Skip to main content

Full text of "Great Epic Of India Its Character And Origin"

See other formats







call No. L 2JL yn j C j Hop 



CD ( 


gale 'Bicentennial publications! 

With the approval of the President and Fellows 
of Tale University , a series of volumes has been 
prepared by a number of the Professors and In- 
structors, to be issued in connection with the 
Bicentennial Anniversary , as a partial indica- 
tion of the character of the studies in which the 
University teachers are engaged. 

This series of volumes is respectfully dedicated to 

Ki)t <Sratmatc3 of ttje Ombcrstti? 



Its Character and Origin 





Cofiyri^' 1901, 


p ’<U‘^d, Jun . ranr 

OEN't'K.* . . •• i :&L 

ijjiia;!.' .\v,-.v delbll 

Ace. ;N ». • .S.11.W 

Dado - 

^all No mum i^i 

■j Hrp 

Uf, I''ERsi 

ry PRESS . 

AND S °* • C^UEK,Tr T LS ° N 

L > U - s. A. 


The sub-title of this book places analysis before specula- 
tion. In recent studies of the great epic this order has been 
reversed, for a method calling itself synthesis has devoted 
itself chiefly to dwelling on epic uniformity, and has either 
discarded analysis altogether or made it subject to the 
results of “ synthetic ” speculation. 

The best way, of course, to take up the historical investiga- 
tion of a literary product the origin of which is well known 
is to begin with the source and afterwards to study the 
character of the completed whole. Tut if the origin be 
unknown, and we wish to discover it, we must invert the 
process, and begin our study with an examination of the 
character of the work. When the results of our analysis 
become plain, we may group together those elements which 
appear to have existed from the first, and thus, on the basis 
of analysis, reconstruct the past. To begin with a synthesis 
(so called) of whatever is preserved in the product, and so 
to postulate for the beginning exactly what we find to be the 
completed whole, is a process that leads us only to the point 
from which we started. As vaguely incorrect as is the des- 
ignation synthesis for the method so called is the method 
itself, which thus does away with all analysis. Analysis is 
an examination of constituents. As a method it is, like any 
other, obnoxious to error, but it is not on that account an 
erroneous method. It is in fact, as turned upon history, 
nothing but inevitable critique; and synthesis without such 
critique becomes merely the exploitation of individual opin- 
ion, which selects what pleases it and rejects, without visible 
cause, what is incompatible with the synthetic scheme. 

viii PREFA CE. 

In the case of the great epic of India, the peremptory 
demand that we should reject the test of analysis is the more 
remarkable as the poem has never been completely analyzed. 
The literature mentioned in it has been ably collected in the 
well-known memoirs of Professor Holtzmann, who has also 
indicated what in his opinion may be supplied from allusions ; 
but the poem has not been thoroughly examined to see what 
literature it reflects from the age of the later Upanishads or 
Vedic schools ; it has not received a careful investigation 
from the metrical side; its philosophy has been reviewed 
only in the most haphazard fashion ; and its inner relation to 
other epic poetry has been almost ignored. Yet critic after 
critic has passed judgment on the question of the date and 
origin of this poem, of which we know as yet scarcely more 
than that, before a definitive answer can be given, the whole 
huge structure must be studied from many points of view. 
And last of all the synthesis! comes also, with his ready-made 
answer to a problem the conditions of which have not yet 
been clearly stated. 

Thus far, indeed, the synthetic theory has not succeeded 
in winning over a single scholar to accept its chief con- 
clusions, either as regards the contention that the epic was 
composed 500 is. c., or in respect of the massed books of 
didactic material and their original coherence with the nar- 
rative. Though the results of the method have not proved 
to be entirely nugatory, yet they are in the main irrecon- 
cilable with a sober estimate of the date and origin of the 
epic; but the hypothesis is, in truth, only a caricature of 
Bidder's idea, that the epic was older than it was thought 
to be. In its insistence upon the didactic element as the 
base of the whole epic tale it bears a curious resemblance 
to a mediaival dogma, the epitaph of which was written 
long ago. For there were once certain ingenious alchemists 
who maintained that the Legend of the Golden Fleece was a 



legend only to the multitude, whereas to the illuminati it 
was a didactic narrative teaching the permutation of other 
metals into gold; on the tomb of which brilliant but fal- 
lacious theory was finally inscribed: Xoyos o? ecm ry /iev 
ToXfiy [I eyas ry S’ enrobe i^ee nevos. 1 

But though this theory has failed as a whole, yet, owing to 
the brilliant manner in which it was first presented by its 
clever inventor, and perhaps also to its sharing in the charm 
which attaches to all works of the imagination, it has had 
a certain success with those who have not clearly distin- 
guished between what was essential and adventitious in the 
hypothesis. The Rev. Mr. Dahlmann, to whom we owe the 
theory, has shown that epic legends and didactic motif are 
closely united in the epic as it is to-day ; but this is a very 
different proposition from that of his main thesis, which is 
that complete books of didactic content were parts of the 
original epic. One of these statements is an indubitable 
fact; the other, an historical absurdity. 

This historical absurdity, upheld by the Rev. Mr. Dahl- 
mann in a rapidly appearing series of somewhat tautological 
volumes, is of much wider application than has perhaps 
occurred to the author. For in the later additions, which 
the Rev. Mr. Dahlmann regards as primitive parts of the 
epic, are found those sections which reflect most clearly the 
influence of Buddhism. If these sections revert to 500 B. C., 
all that Buddha as a personality stands for in the history 
of Hindu religious thought and practice belongs not to him 
but to his antecedents, and therewith vanishes much of the 
glory of Buddha. Though the author has not publicly rec- 
ognized this obvious result of his theory, yet, since it is 
obvious, it may have appeared to some that such a darken- 

1 Almost identical, in fact, is tlie verdict on the synthetic argument 
delivered by the veteran Trench critic, M. Barth : “ conclusion audaeieuse 
. . . theorie absolument manque'e ” (Journal des Savants, 1807, pp. 337, 148). 



ing of the Light of Asia added glory to the Light of the 
World, and this is possibly the reason why the synthetic 
theory has been received with most applause by the reviewers 
of religious journals, who are not blind to its bearings. But 
however important inferentially, this is a side-issue, and the 
historian’s first duty is to present the facts irrespective of 
their implication. 

On certain peculiarities (already adversely criticised by 
disinterested scholars) characteristic less of the method of 
investigation than of the method of dialectics which it has 
suited the Rev. Air. Dalilmann to adopt, it is superfluous to 
animadvert in detail. Evidence suppressed by one seeker, 
in his zeal for truth as he sees it, is pretty sure to be turned 
up by another who has as much zeal and another method; 
nor has invective ever proved to be a satisfactory substitute 
for logic. As regards the claims of synthesis and analysis, 
each method has its place, but analysis will always have the 
first place. After it has done its work there will be time 
for honest synthesis. 

The material here offered is by way of beginning, not by 
way of completing, the long task of analyzing the great 
epic. It is too varied for one volume, and this volume has 
suffered accordingly, especially in the chapters on philosophy 
and the interrelation of the epics. But the latter chapter was 
meant only as a sketch, and its worth, if it has any, lies in 
its appendix ; while the former could be handled adequately 
only by a philosopher. The object of these and other chap- 
ters was partly to see in how far the actual data rendered 
probable the claims of the synthetic method, but more par- 
ticularly to give the data without concealment or misstate- 
ment. For this reason, while a great deal of the book is 
necessarily directed against what appeared to be errors of 
one sort or another, the controversial point of view has 
not seldom been ignored. Pending the preparation of a 



better text than is at present available, though Dr. Winter- 
nitz encourages the hope of its eventual appearance, the 
present studies are intended merely as signboards to aid 
the journey toward historical truth. But even if, as is 
hoped, they serve to direct thither, they will be rendered 
useless as they are passed by. Whether they are deficient 
in their primary object will be for travellers on the same 
road to say. 

January, 1901. 






The Vedas 2 

Divisions of Veda 7 

Upanishads 9 

Upavedas and Upafigas 11 

Sutras 15 

Dharmayiistras 17 

Vedic citations in the Epic 23 

Upanishads in the Epic 27 

The Cvetaevatara Upanishad 28 

The Kathaka or ICatha Upanishad 29 

The Maitri Upanishad in the Epic 33 

The Atharvajiras Upanishad 46 

Afvaliiyana Grhya Sutra 47 

Furiinas and Itihasas 47 

Drama 54 





Epic Systems 85 

Heretics 86 

Authority 90 




V edanta 93 

Nyaya 95 

Vaiyesika 96 

The Four Philosophies 96 

Ivapila and his System 97 

Sariikhya and Yoga 101 

Fate and Free-Will 103 

Sariikhya is atheistic 104 

Yoga as deistic and bralimaistic 106 

Difference between Samkhya and Yoga Ill 

Sects 115 

The different Schemata 116 

The Gunas 119 

Plurality of Spirits 122 

The Twenty-fifth Principle 125 

Samkhya is Saiiikhyana 126 

The Samkhya Scheme 127 

The Twenty-sixth Principle 133 

Maya, Self-Delusion 13S 

Pahca(,'ikha’s System 142 

The Thirty-one Elements (Pancat, ikha) 152 

The Secret of the Vedanta 157 

Details of philosophical speculation 162 

The Sixty Constituents of Intellect 163 

The Seventeen 165 

The Sixteen (A) Particles 168 

The Sixteen (B) or Eleven Modifications 169 

The Eight Sources 170 

The Vital Airs and Senses 171 

The Five Subtile Elements. Gross and Subtile Bodies 173 

The Colors of the Soul 179 

The Five Faults of a Yogin 181 

Discipline of the Yogin 

The Destructible and Indestructible 182 

The Gods and the Religious Life 183 

Heaven and Hell — Death 

The Cosmic Egg and Creations 187 

The Grace of God ] gg 






Epic Versification 191 

Cloka and Tristubh. The Padas 194 

‘'''Rhyme 200 

Alliteration 202 

Similes and Metaphors. Pathetic Repetition 205 

Cadence in Cloka and Tristubh 207 

Tags 211 

Common forms of Cloka and Tristubh 214 

The Epic Cloka. The Prior Pada of the Cloka. The Pathya . . 219 

The Vipulas 220 

The Posterior Pada of the Cloka 239 

The Diiambus 242 

Poetic Licence 244 

The Hypermetric Cloka 252 

Dialectic Sanskrit 261 

Prose-Poetry Tales 266 

The Epic Tristubh. i. The Regular Tristubh in the Mahabharata 273 

Bird’s-eye View of Tristubh Padas 275 

The Ramayana Tristubh 276 

The Scolius 277 

Catalectic and Hvpermetric Tristubhs 281 

ii-iii, The Catalectic Tristubh 282 

iv-ix, The Hypermetric Tristubh. iv-vi, Simple Hypermeters . 286 
vii-ix, Double Hypermeters or Tristubhs of Thirteen Syllables 298 

Defective Tristubhs 299 

v, b, and ix, Mora-Tristubhs 301 

The Tristubh-Stanza. Upajatis. Upendravajras and Indravajras 309 

The Syllaba Anceps 314 

Emergent Stanzas 317 

The Fixed Syllabic Metres 321 

Rathoddhata 322 

Bhujamgaprayata 323 

Drutavilambita 324 

Vaifvadevl 325 

Atijagatis. Rucira 326 




The Fixed Syllabic Metres ( continued ) — 

Praharsini 329 

Mrgendramukha 331 

Asambadha 332 

Vasantatilaka 333 

Malini 334 

Qardulavikridita 336 

Ardhasamavrtta (Matrachandas). A — Pnspitagra and Aparavaktra 336 

B — Aupacchandasika and Vaitaliya 341 

Matrachandas in the Mahabharata 343 

Matrasamakas 353 

Ganacchandas 354 

The Distribution of Fancy Metres in the Epic 356 





APPENDIX A. Parallel Phrases in the two Epics . . 403 

“ B. Illustrations of Epic Cloka Forms . . 446 

“ C. Illustrations of Epic Tristubh Forms . 459 





As most of the references in this volume are to the Mahabharata, all 
numbers without alphabetical prefix refer to this epic (Bombay edition, or 
with prefix C. to Calcutta edition); but when necessary to distinguish a 
reference to the Mahabharata from a reference to the Ramayana, I have 
prefixed M., which therefore does not refer to Manu, but to the great epic. 
To bring the two parallel editions of the epics into line, I have used R. or 
RB. for the Bombay edition of the Ramayana also (rather than for the 
Bengal text), and for clearness I employ G. for the Gorresio (Bengal) text 
thus : — 

M. or MB., Mahabharata, Bombay edition. 

R. or RB., Ramayana, Bombay edition. 

C., Mahabharata, Calcutta edition. 

G., Ramayana, Gorresio’s edition. 

Other abbreviations, such as those usually employed to indicate native texts, 
or, for example, ZDMG. and JAOS. for the Journals of the German and 
American Oriental Societies respectively, require no elucidation for those 
likely to use them. Those using the old edition of RB. must add one to all 
references to sargas after vi, 88, and two to all after vi, 107 . Sanskrit 
words usually anglicized have so been written. 

$aie 'Bicentennial publications 





Paradoxical as it may seem, the great epic mentions post- 
epical as well as prae-epical works. To solve the paradox 
it is necessary to assume that the text has been interpo- 
lated, a fact admitted as a last recourse even by him who 
holds that the epic was originally what it is to-day. But 
interpolations to be referred to when everything else fails 
will not suffice. A large part of the present epic is inter- 
polation, some of it self-interpolated, so to speak. For, not 
content with receiving accretions of all sorts, narrative and 
didactic, the Bharata, in default of other sources of inter- 
polation, copied itself. Thus the same story, hymn, and 
continuation are found in iii, 83, 116 £f. and ix, 38, 39 ff. 
The matter of xii, 223 is simply enlarged in 227, while xii, 
248-9 repeats xii, 194 and then reappears again in xii, 286. 
An example of reproduction with variations is found in ix, 51, 
50, as compared with iii, 133, 12 ff. In one case a youthful 
prodigy encounters venerable sages and teaches them the 
Veda; in the other a priest and king are instructed, but with 
the same setting of proverbial lore. So xii, 185 is a repro- 
duction of iii, 213, 1-19; xii, 277 (8), of xii, 175, etc. 

It is not strange, therefore, that a work thus mechanically 
inflated should have absorbed older literature. But to under- 
stand the relation between the epic and the older literature 
copied by the epic it is essential to know the whole literature 
referred to as well as cited. In this chapter, then, beginning 
with the Vedas, I shall follow the course of revealed and 




profane literature as far as it is noticed in the epic itself, 
reserving, however, for the two following chapters the Ra- 
mayana and the philosophical systems. 

The Vedas. 

Allusions to Vedic literature, veda, chandas, mantra, §ruti, 
are naturally common in every part of the Mahabharata, but 
except in the didactic or later epic these are usually of a gen- 
eral character. It may be assumed that the bulk of Cruti or 
revealed works, if not all of it, was composed before the epic 
began. Nevertheless, it is interesting to see which portions 
of this hereditary literature are especially mentioned, and 
particularly important to observe how the epic cites from 
older works. Even the fact that it does cite verbatim the 
words of the holy texts is of historical moment when it is 
remembered that in other places even women and slaves are 
exhorted to hear the recital of the epic . 1 We find indeed in 
the course of the epic narrative that a woman is taught Vedic 
mantras , 2 but the mantras are from the Atharva Veda, which, 
without being particularly slighted, is less regarded than the 
older Vedas, as is shown by this incident; for no woman 
would have been taught Rig Veda verses, for example. 

The Vedas are all mentioned by name, though the Atharva 
Veda is not always recognized in the formal enumeration. 
The order of precedence is not fixed, though its peculiar 
holiness, vimala, is not the reason why the Santa Veda in the 
Gita and Anuijasana heads the list . 3 Usually the Rig Veda 
stands at the head and the Atharva, if mentioned, at the foot, 
though the order Rk, Yajus, Atharvan, Saman, and even 
Atharvan, Saman, Rk, Yajus is found; but the last order 
occurs only in the didactic or later epic. The four together 
comprise the veda$ caturmurtih, or fourfold Veda, which, in 

1 Compare i, 62, 22 ; 95, 87 ; iii, 85, 103 ; xii, 341, 116, etc. 

2 Tatas tam grahayamasa sa dvijah Mantragramam . . . atharvafirasi 
(jrutam (v. 1. atharvangirasi), iii, 305, 20. 

3 For in v, 44, 28, it has this epithet, yet stands last in the list : “Not in 
R. V., nor in Y. V., nor in Atharvas, nor in the spotless Samans.” 


distinction from the threefold Veda, is often joined with the 
“Veda of the bow.” The epic even has caturveda as an 
epithet of a man, — “one that knows the four Vedas” 
(= caturvaidya), — as earlier triveda, traividya, is used in 
the same way of one learned in the three (caturvidyam is a 
pseudo-epic term for the Vedas). 1 

The tradition of “lost Vedas” 2 and “ divided Vedas ” is 
well known. There was at first but one Veda, but after tire 
Krta age men became men of three, men of two, men of one, 
and men of no V edas, triveda, dviveda, ekaveda, anrk, iii, 149, 
14-29, and v, 43, 42, bhinnesu being Vedas; bhinnas 
tada vedah, xii, 350, 42 (by Apantaratamas). The last pas- 
sage is peculiar in the use (<jl. 41-47) of ved akhyane Qrutih 
karya, and in the name of Kali as krsna (as well as tisya). 3 

The Veda is either recited, declared, or made, srsta, krta. 
The latter word contradicts the dogma declared in the well- 
known words : na hi cchandahsi kriyante nityani ccliandansi, 
“ the Vedas are not made, they are eternal ; ” but the sense is 

1 The word triveda remains the usual form (tritayarh sevitam sarvam, ix, 
64, 21). Besides caturveda as an epitiiet of a god (illustrated in PW.) we find 
in the late passage iii, 313, 110 ff. : pathakah patlmkag cai \-a ye ca ’nye 
gastracintakah sarve vyasasino markka, yah kriyavan sa panditah; catur- 
vedo 'pi durvrttah sa gudrad atiricyate, yo 'gnihotraparo dantah sa brahraana 
iti smrtah. On the order of names referred to above : the lead of the Atharva 
is found also in the Mahabhasya (IS. xiii, p. 432) ; the epic passage is xiii, 17, 
91. The name is here atharvana or atharvana, xiii, 93, 136 ; 94, 44. Exam- 
ples of the usual order are rco yajunsi samani, i, 1, 66 ; ix, 36, 34 ; xii, 252, 2 
(rco yajunsi samani yo veda na sa vai dvijah) ; rgvedah samavedag ca jajur- 
vedag ca atharvavedag ca, ii, 11, 32 ; iii, 189, 14, atharvanah. In v, 18, 6-7, 
it is said that the name Atharvangiras will eventually belong to the Atharva 
Veda. The word samani is not restricted to this Veda. Thus Dhautnya, a 
Purohita and, therefore, as Weber has shown, presumably an Atharvan 
priest, sings incantations of destruction, samani raudrani yam_\ an i (gavan), ii, 
80, 8. On the expression atliarvavede vede ca, see below. For the order of 
names, compare my Ruling Caste, p. 112; and see Holtzmann, lias Mahabha- 
rata, iv, p. 5; for further passages (for the AV. in particular), Bloomfield, 
SBE. xiii, p. liii. 

* On this aeonic occurrence (xii, 210, 16 ff.), compare vedagrutih pranasta, 
xii, 346, 9, the story in 348, and the quotation in the text below. The modi- 
fied vrata, rules, vikriyante vedavadah, are referred to in xii, 233, 38. 

* The former as Kali is still starred in pw. The latter is masculine in R. 
vi, 35, 14 (also starred as such in pw.). The word occurs also in xii, 311, 86. 



not opposed, as tlie maker is God (vedakarta vediingo veda- 
vahanah, iii, 3, 19), who only emits the Y edas as he does all 
else when the new aeon begins. The more decided “make” 
is found of seers, however, in the Harivan§a, mantrabrahmana- 
kartarah, mantrakrtah, 1 seers and descendants of seers, just 
as there is a Mahabharatakrt and Itihasasya karta, or 7roir;Tr?? 
67rwr, though he too is divine. 2 The gods who are credited 
with the making of the Vedas 3 are Fire and Sun, as All-God 
(above), or especially Brahman, and in the later epic Vishnu. 
It was Brahman who “first recited the Vedas,” vedan jagau, 
v, 108, 10. With a natural inversion, “Brahman created 
brahman ” (whereas in reality brahman created Brahman), ac- 
cording to another passage, xii, 188, 1-2. Compare : ya ime 
brahmana prokta mantra vai proksane gavam ete pramanam 
bhavata uta ’ho na, v, 17, 9-10. The Self-existent, according 
to xii, 328, 50, created the V edas to praise the gods, stutyar- 
tham iha devanam vedali srstah svayambhuva. Krsna, who is 
krtagama, in xiii 149, 97, takes the place of the more general 
term. Compare xii, 340, 105 : 

yada vedaqrutir nasta maya pratyakrta punah 
savedah saqrutlkag ca krtdh purvarn krte yuge 
(atikrantah puranesu crutas te yadi va, kvacit), 

and nirmita veda yajnac cau ’sadhibhih saha, ib. 341, 66, with 
xiii, 145, 61, agama lokadharmanain maryadah purvanir- 
mitah. 4 

1 jayanti ’ha punah punah Mantrabrahmanakartarah dharme pragithile 
tatha, H. 1, 7, 56. 

2 Krsna Dvaipayana, also called Kuruvahgakara, xii, 347, 13 ; xiii, 18, 
43-44. The recitation of the Vedas is a matter of scientific study. When 
they are “loudly recited in the proper way,” sagaiksya, they fill (other) winds 
with fear, and therefore should not be recited when a high wind is blowing, 
xii, 329, 23-56. 

8 For the gods and especially for the part of Brahman in creating the Vedas 
and the transfer of his office to Vishnu in the epic, see Holtzmann, ZDMG. 
xxxviii, p. 188, and Das Mahabharata, iv, p. 6. 

4 The v. 1. sarya is wrong. The word agama usually refers to Veda, but not 
always. Compare xiii, 104, 156, agamanam hi sarvesam acarah grestha ucyate ; 
i, 2, 36, itihasah gresthah sarvagamesy ay am ; xii, 59, 139, agamah purana- 
nam. It means any received work, particularly the Vedas. 


In late passages the two earliest forms of the text (the 
latest forms are unknown) together with the accents of the 
texts are especially mentioned. 1 

In the important numerical analysis of xii, 343, 97-98, the 
Rig Y eda is said to “ have twenty-one thousand ” ; while the 
Suma Yeda has “one thousand branches”; and the adhva- 
ryava or Yajus has “fifty-six and eight and thirty-seven 
(one hundred and one) branches.” Probably “twenty-one 
branches” is the real meaning in the case of the Rig Veda. 
Here too are mentioned the gitis, songs or verses (a rather 
unusual word) found in the branches in their numerous divi- 
sions, (jakhabhedali, qakhasu gltayali. 2 

It is evident from this statement that, as Weber says of the 
passage in the Mahabhasya, we are dealing with a period 
when the number of Yajur Yeda schools is greater than that 
recognized in the Caranavyuha, which gives only eighty-six. 
Another verse of this book recognizes ten thousand rcas : 
“ This ambrosia churned from the wealth of all the dharma- 
khyanas, the satyakhyana, and the ten thousand rcas,” xii, 

1 rgvedah padakramavibhusitah, xiii, 85, 90; atharvavedapravarah puga- 
yajniyasamagah sarhhitam Iravanti sma padakramayutam tu te, i, 70, 40. 
Galava, Babhravyagotra, Pancala, the grammarian, through the especial grace 
of the deity and being instructed in the method of Vamadeva, became a 
shining light as a kraraa specialist, xii, 343, 100 ff. ; laksanani svarastobha 
niruktam surapaiiktayah, xiii, 85, 91 (together with nigraha and pragraha) ; 
svaraksaravyanjanahetuyuktaya (gira), iii, 297, 26. 

2 The verse translated above is ekavihpatisahasram (rgvedam mam pra- 
caksate). Twenty-one thousand what? Not stanzas, for the Rig Veda has 
only half so many (Muller, ASL. p. 220). On the other hand, the passage 
agrees closely with one in the Mahabhasya (IS. xiii, p. 430), where the cor- 
responding words are “twenty-one fold,” after varlma (school): ekapatam 
adhvaryupakhah, sahasravartma samavedah, ekavinpatidha bahvrcyam (a 
word implied in Mbh. xv, 10, 11, “Samba the bahvrcah”), navadha atharvano 
vedah. The epic text, closely corresponding, is : ekavinyatisahasram rgvedam 
. . . sahasragakham yat sama . . . satpanca^atam astau ca sapta trinpatam 
ity uta yasmin pakha yajurvede, so *ham adhvaryave smrtah, pancakalpam 
atharvanam krtyabhih paribrmhitam kalpayanti lii marii vipra atharvana- 
vidas tatha. There can scarcely be a doubt that for the text above we 
should read ekavinpatipakham yam, as the parallel suggests, for the text as 
it stands is unintelligible. 1 regret that M eber has not noticed the epic pas- 
sage, so that I cannot cite his opinion. 



247, 14, where the commentator says that this is a general 
number, implying a fraction over lO^SO. 1 

In the account of the later epic we have a parallel to that 
of the Vayu Purana, where the latter, lxi, 120 if., is account- 
ing for the successive editions of the Yedas : 

avartamana rsayo yugakhyasu punah punah 
kurvanti samhita hy ete jayamanah parasparam 
astagitisahasrdni qrutarslnam smrtani vai 
ta eva samhita hy ete avartante punah punah 
grita daksinam panthanam ye gmaganani bhejire 3 
yuge yuge tu tah ijakha vyasyante taih punah punah 
dvaparesv iha sarvesu samhitaq ca srutarsibhih 
tesam gotresv imah qakha bhavanti ’ha punah punah 
tah tjakhas tatra kartaro bhavanti ’ha yugaksayat 

The eighty thousand Yedic seers here mentioned are those 
of the Harivahga (loc. cit.) : ye (jruyante divarn prapta rsayo 
hy urdhvaretasah mantrabrahmanakartaro jayante ha yuga- 
ksaye. They are mentioned elsewhere in the Vayu Purana, 
viii, 184, and in the epic itself, ii, 11, 54, in the same words : 

astaqltisahasrani rslnam urdhvaretasam, 
a verse found also in the Mahabhasya (IS. xiii, p. 483). 

1 Compare further the da?a paaca (ca) yajunsi, learned from Arka by the 
author of the Qatapatha Brahmana, in xii, 319, 21. The word carana, in the 
sense of school, occurs in xii, 171, 2, prstaf ca gotracaranam svadhyayam 
brahmacarikam ; xiii, 63, 18, na prcched gotracaranam. The mantras of the 
special septs are referred to in the late hymn to the Sun (Mihira), iii, 3, 39 : 
(tvam brahmanah) svagakhavihitair mantrair arcanti. The commentator 
cited above gives as his authority for the number of stanzas in the Rig Veda 
a lame couplet of the C^kalaka : ream da 9 asahasrani ream paficagataui ca 
ream agitih pada? cai-’tat parayanam ucyata, iti. 

1 They are referred to, but not as Veda-makers, in Yaj. iii, 186, and in Ap. 
Dh. S., ii, 9, 23, 3-5 (as being mentioned “in a Purana”). Yajnavalkya calls 
them the astagitisahasra munayah punaravartinah . . . dharmapravartakah. 
The Purana referred to by Apastamba may be the one cited above, though 
in another form, since the words have a different application. There is here a 
pragamsa of the urdhvaretasas : astagitisahasrani ye prajam isira rsayah dahsi- 
nena’ryamnat) panthanam te gmaganani bhejire, etc. Compare Prafna Up. i, 9, ta 
eva punaravartante tasmad ete rsaya prajakama daksinam pratipadyante. 


Divisions of Veda. 

Reference is seldom made to Samhita, Brahmana, or Ara- 
nyaka. The “ peruser of Samhita,” samhitadhyayin, is 
alluded to in i, 167, 8, and xiii, 148, 56. The word is used 
also of the epic, Vyasa’s Samhita, the fifth Veda. In xii, 
201, 8, saugha may be used in the same sense of collection, 
but it probably means a quantity. I will give the passage, 
however, as it enumerates the usual (i, 170, 75, etc.) six 
Vedangas, though in an order constrained by the metre (they 
and the Upaiigas will be discussed below, under Upa vedas): 

rksamasangaruj ea yajunsi ea ’pi 
cehandansi naksatragatiiii niruktam 
adhltya ca vyakaranam sakalpaiii 
Qiksam ea, bhutaprakrtim na vedmi, 

“ Although I have studied collections of hymns and chants and 
the sacrificial formulas, and also prosody, astrology, etymology, 
grammar, ritual, and phonetics, I do not know the First Cause of 

Brahmanas are mentioned in xii, 269, 83-84, as the source 
of sacrifice, and in iii, 217, 21 , “ the different Agnis named 
in the Brahinanas,” brahmanesu. In xiii, 104, 137, “rites 
declared in the Veda by Brahmanas,” the word means priests. 
Possibly Gita, 17, 23, brahmanali (and vedali) may be works, 
as the epic is not particular in regard to the gender of these 
words (purana, itihasa, and mahabhuta are both masculine and 
neuter). Yiijnavalkya’s C'atapatha Brahmana alone is named, 
with all its latest additions (krtsnarii sarahasyam sasamgra- 
ham sapariijesam ca), xii, 319, 11, and 16. So ib. 24, 25, and 
34 : “I resolve in mind the Upanishad (BA.) and the Pari- 
qesa (the last part), observing also logic, the best science, 
anvlksikl para, and declare the fourth transcendental science 
or science of salvation, samparayika, based on the twenty-fifth 
(Yoga) principle.” 1 Other Brahmanas may be implied in the 

1 In the expression, loc. eit., ft 10, vedah sakhilah so ’ttarah, uttara refers 
to the Upanishads (not to the philosophy). The Khila Supplement is men- 
tioned again in the Harivahfa (Boltzmann). 



list at xii, 387, 7 ff., Tandya, Katha, Kanva, Taittiri. 1 As 
“prose works,” gadya, this class of works is perhaps recog- 
nized in iii, 26, 3, in the words : “ The thrilling sound of 
yajuhsi, rcah, samani, and gadyani ” (as they were recited). 

Whether pravacana, exposition, means Angas or Brahma- 
nas or perhaps Sutras, I do not know. The (Upanishad) 
word occurs in a verse found also in Manu, where the com- 
mentator explains it as Aiiga, to which the objection may be 
made that the Angas have already been mentioned. But the 
passage is not without importance as showing how the didac- 
tic or later epic adds elements to the simpler statement of 
the earlier law-books. In xiii, 90, 36, the pahkteyas, or men 
who may be invited to sit in the row at a funeral feast, are 
not only the agryah sarvesu vedesu sarvapravacanesu ca of 
Manu iii, 184, and the list of iii, 185, trinaciketah pancagnis 
trisuparnah sadangavid (v. 1. brah n i adey an usan t an ag chandogo 
jyestasamagah) in 90, 26, but, among others, the atharvagi- 
raso 'dhyeta, 29 (a rare word) ; “ those who cause the Itihasa 
to be read to the regenerate,” 33 ; those who are “acquainted 
with commentaries,” bhasyavidas (or know the Maliabhasya?), 2 
and are “ delighted with grammar,” vyakarane ratah, 34 ; 
those who “study the Purana and the Dharmagastras ” ; those 
who “ bathe in holy pools,” ye ca punyesu tTrthesu abhise- 
kakrtagramah, 30 (a practice not extolled by Manu, whose 
view seems to be tiiat of Agastya, asti me kagcit tlrthebliyo 
dliarmasamgayah ! xiii, 25, 5). The bharate vidvan, xiii, 76, 
18, is naturally extolled in the epic, and yet even with this 
latitude we must see in the list above a distinct advance on 
the position held by the early law-makers, to whom it was 
not enough for a man to recite the epic (not to speak of 
grammar and bhasya-knowers as being ipso facto pankteyas) 
to be deemed worthy of invitation. Even Vishnu’s Sinrti is 
here exceeded, and Manu and the Sutras have nothing in any 
degree parallel. Even if we say that the list is on a par with 

1 The Taittiri dispute is referred to in xii, 319, 17 ff. 

2 But bhasya may mean any reasoned exposition, bliasyani tarkayuktani, 
ii, 11, 35. 


Vishnu alone, although it really exceeds it in liberality, we 
thereby put this epic passage on a par with a law-book later 
than any that can be referred to the Sutra period, later than 
Manu also and probably Yajnavalkya. 1 

Almost as rare as the mention of Brahmanas is that of Ara- 
nyakas. In the passage cited above, xii, 343, stanza 98 has 
as elsewhere the singular, gayanty aranyake vipra madbha- 
ktah. So ib. 340, 8: “ Hari sings the four Vedas and the 
Aranyaka ” (as forest, e. g., ib. 337, 11, aranyakapadodbhuta 
bhagah) ; and in xii, 349, 29-31, the Krishna religion has 
“mysteries, abstracts, and Aranyaka.” Compare also v, 175, 
38, gust re ca ’ranyake guruh, “ a man of weight in code and 
esoteric wisdom ” ; xii, 344, 13, aranyakam ca vedebhyah 
(yatha), where the kathamrtam or essence of story of the 
expanded Bharata, Bharatiikhyanavistara of 100,000 glokas, 2 
is compared to the Aranyaka as the essence of the Vedas (a 
simile repeated at i, 1, 265). The word is in fact general- 
ized, like Upanishad. But as a literary class it is found in 
the plural in xii, 19, 17, vedavadan atikramya gastrany 
aranyakani ca . . . saram dadrgire na te, “ they ran over the 
words of the Vedas, the ^astras, and the Aranyakas, without 
discovering their inner truth.” Here Veda does not connote 


The Upanishads are alluded to in the singular, collec- 
tively, or distributively in the plural. They are generally 
grouped with the Angas and are called Upanishads, rahasyas, 
mysteries, Brahma Veda, and Vedanta; while like the Ara- 
nyakas they are logically excluded from the V eda of which 
they are supposed in ordinary parlance to form part. 3 The 

1 Vishnu, ch. 83 ; Manu, loc. cit. ; Yaj. i, 219 ; Ap. ii, 17 ; Gaut. xv ; Vas. xi. 
I doubt whether the “ Atharva 5 iras-reader ” can imply the Qiras-vow, but even 
this is a comparatively late touch, Baudh. ii, 14, 2, in this regard. 

2 Note that the number of verses show that the Harivanfa already existed 
when this passage was written. Compare ib. 340, 28. 

8 I mean that in the current phrase vedah safigah or sopanisadah the sa 
should differentiate as much as it does iu the parallel phrase rg\ edali saya- 



word upanisad has two distinct but current meanings in the 
epic. It means on the one hand mystery, secret wisdom, 
essential truth, essence, as in xiii, 78, 4, gavam upanisadvid- 
van, “ wise in cow-mysteries,” and in iii, 207, 67 = xii, 252, 
11, vedasyo ’panisat satyam, satyasyo ’panisad darnah, “ truth 
is the secret wisdom (essence) of the Veda, patience the 
essence of truth.” So in the common phrase, vedag ca sopa- 
nisadah, xiii, 85, 92, etc., the word may mean mysteries. This 
I think is the explanation of the employment of the word 
mahopanisad in vii, 143, 34-35, where Bhurigravas devotes 
himself to prdya before death in battle. He is a muni here 
and desires to ascend to the world of Brahman, so he sits 
down in Yoga contemplation and meditates the “ great Upa- 
nishad,'' dliyayan mahopanisadam yogayukto 'bhavan niunih. 
On comparing the scene where Drona is in the same situa- 
tion, vii, 192, 52, we find that he says om, and this mystery 
of om is probably the meaning of mahopanisad, which cannot 
be a work here, as is mahopanisadam in xii, 340, 111. But in 
other cases Upanishad is clearly a literary work, even stand- 
ing in antithesis to the mysteries with which it is sometimes 
identical, as it is in the form upanisa in the Pali scriptures. 1 

jurvedah, or in yad etad ucyate gastre se ’tihase ca chandasi, xiii, 111, 42. 
But it is very likely that the term was used to mean “including” (as part of 
the Veda). On the use of singular and plural referred to above, compare sa 
raja rajadharmang ca brahmopanisadam tatha avaptavan, xv, 35, 2 ; saiigo- 
panisadan vedan viprag ca’ dhiyate, i, 64, 19, etc. For Vedanta and Vedantah, 
meaning Upanishads, compare iv. 51, 10, vedantag ca puranani itiliasam (!) 
puratanam; xiii, 16, 43, ( fiv a ) yam ca vedavido vedyam vedante ca pratisthi- 
tam . . . yam viganti japanti ca ; H. 3, 10, 67, puranesu vedante ca. I may 
mention here also the works called Nisads, which are referred to (or invented) 
only, if I mistake not, in xii, 47, 20, yam vakesv anuvakesu nisatsupanisatsu 
ca grnanti satyakarmanam satyam satyesu samasu. 

1 Kern, SBE. xxi, p. 317. Compare for the use of the word, xii, 245, 15, 
where it is said that the Upanishads inculcate the four modes of life, caturthag 
cau ’panisado dharmah sadharanah smrtah ; and xiii, 84, 5, where it is said 
that Vedopanisadas inculcate that earth, con-s, or gold must be the sacrificial 
fee. As we find vedah sarahasyah sasamgrahah and vedavedangabhasyavit, 
xii, 325, 22-23, so in viii, 87, 42, reference is made to “ all the Vedas, with 
Tales as the fifth Veda, together with Upavedas. Upanishads, mysteries, and 
abstracts” (samgraha). Narada is said to be vedopanisadam vetta itihasa- 
puranajaah . . . sadangavit and smrtiman, ii, 5, 2 ff. The use in iii, 251, 23, 


Upavedas and Upangas. 

The Upavedas or subsidiary Vedas are three in number, 
Ayur Veda, Dhanur Veda, and Gandharva Veda. To these 
is added in other works Sthapatya Veda, but this term is not 
recognized in the epic, and the commentator on vii, 202, 75, 
recognizes only three, those just given, or Medicine, Archery, 
and Music ; but the fourth, Architecture, is known (only 
in the epic introduction), as Vastuvidya. 1 Authors are as- 
signed to these and other works in xii, 210, 20, Brhaspati 
being the originator of all the Vedangas ; Bhrgu’s son, of 
Nltigastra, law; Niirada, of music; Bharadvaja, of the sci- 
ence of arms (particularly archery) ; Gargya, of tales of 
the doings of seers (devarsicarita) ; and Krsniitreya, of med- 
icine (cikitsita). They are all contrasted with other Nyiiya- 
tantrani, which like these were created at the beginning of 
the aeon as an aid in understanding Brahman (expounded by 
hetu, agama, and sadacara, or reason, faith, and common con- 
sent of good men, ib. 22). It is noteworthy that Niirada, not 
Bharata, is found in this connection, and that Krsniitreya 
takes the place elsewhere given to Bharadvaja. 

Of the first of these subsidiary Vedas, tire epic naturally 
gives little information, though burdened with much medici- 
nal knowledge which may be referred to some uncited work 
on medicine. Native scholars imagine that the correspond- 
ing Upanishad passages imply the circulation of the blood, 
also thought to be recorded in xii, 185, 15, prasthitii hrdayat 
. . . vahanti annarasan nadyah : “ The veins convey (all over 

would suggest that Upanishad is a sort of Sutra, for here a spirit is summoned 
by means of “ mantras declared by Brhaspati and Upanas ; by those declared 
in the Atharva Veda ; and by rites in the Upanishad,” yap co ’panisadi krivah. 
I am not certain how to interpret pathyase stutibhip cai ’va vedopanisadaih 
gaydih xii, 285, 126. 

1 Thus the architect, sutradhara, sthapati, is vastuvidyaviparada, i, 51, 15 
(the sutrakarmaviparada of G. ii, 87, 1). Architectural Qastras are mentioned 
in i, 134, 10-11. As a fourth to the three is elsewhere set the Arthapastra. 
These as a group are added to the other vidyas (see note below on the sixty- 
four arts and fourteen sciences). But in the epic, Arthapastra is not grouped 
with the Upavedas. 



the body) the food-essences, starting from the hrdaya ” (heart 
or chest). But a direct citation is the allusion, under the 
cover of an “it is said,” to the constituents pitta, glesman, 
vayu (also vata, pitta, kapha), which make the threefold 
body, tridhatu, according to the Aryurvedins. 1 In the epic 
Khila and in the Kaccit and eleventh chapters of Sabha, both 
late additions to the epic, 2 the science of medicine is said to 
have eight branches (ii, 5, 90 ; 11, 25). Possibly in iii, 71, 
27, Qalihotra may' represent the veterinary science of iv, 12, 7. 

The Dhanur Veda, literally Veda of the bow, is often 
joined with the regular Vedas, as is to he expected in epic 
poetry, ix, 44, 21-22, etc. It is called also isvastra, weapons, 
and is said to be fourfold and to have ten divisions. In the 
Kaccit chapter just referred to it is said to have a Sutra like 
other Vedas, and at the time this was written it is very prob- 
able that such was the case, though, as I have shown else- 
where, the knight’s study of Dhanur Veda consists in prac- 
tice not in study of books. This Bow-Veda, archery, is 
opposed sometimes to the four Vedas alone, sometimes to 
the Upanishads and Brahma Veda, while on the other hand 
it is associated with various Sutras, arts, and Nitigastras. 
The priority of Dhanur Veda in the phrase dhanurvede ca 
vede ca, found in both epics, is due partly to metrical con- 
venience and partly to the greater importance of this Veda 
in the warrior’s education : 3 na tasya vedadhyayane tatha 
buddhir ajayata yatha ’sya buddhir abhavad dhanurvede, 
“His intelligence was more developed in learning how to 
use a bow than in perusing holy texts,” i, 130, 3 ; dhanur- 

1 xii, 343, 86-87 : pittam <;lesma ca vayu? ca esa samgliata ucyate, etai? 
ca dharyate jantur etaih ksinai? ca ksiyate, ayurvedavidas tasmat tridha- 
tum mam pracaksate. Compare vi, 84, 41, cited in PW, and also xiv, 12, 
3, fitosne cai ’ya vayu? ca gunah . . . ?arirajah, whose equality is health 
(N. kaphapitte). Some notes on epic anatomy will be given later. 

2 The lateness of the Kaccit chapter I have discussed elsewhere, Am. 
Jonrn. Phil., vol. xix, p. 147 S. A noteworthy statement on disease is that 
of xii, 16, 9, which attributes all mental disease to the body and all bodily- 
disease to the mind,manasaj jayate ?arirah (vyadhih), “bodily ailment arises 
from mental (ailment).” 

3 The same is partially true of atharvavede vede ca, xiii, 10, 37, etc. 


veclaparatvat, ib. 4. 1 It is the Ksatra Veda or knightly science 
par excellence, R. i, 65, 23 (with Brahma Veda). 

The science of music, Gandharva Veda, consists according 
to iii, 91, 14, in the knowledge of singing, dancing, chanting, 
and playing on musical instruments, git am nrtyam ca sama 
ca vaditram ca, not including apparently the Natasutra or 
manual for actors mentioned by Panini. The seven musical 
scales, van! saptavidha, ii, 11, 34, are a branch of study. 
The three notes of the drum are spoken of 2 and the names 
of the notes of the regular scale, gamut, are given. Further 
citations in this regard will be made hereafter. 

These Upavedas are associated with the chief Vedas (vedah 
and upavedah, vii, 202, 75, etc.), much as are the Vedangas, 
Upanishads, and Tales, and are distinguished as well from the 
(Jastras and Sutras mentioned in the passage already noticed, 
ii, 11, 32-33, though (lustra is a general term including Upa- 
veda. The Ahgas are the customary six mentioned above, 
and are generally referred to as in i, 104, 12, vedam sadafigarii 
pratyadhlyata ; or without number, as in i, 156, 5, brahmaih 
vedam adhiyana vedangani ca s a r vat; ah, nlt^iistrarii ca sarva- 
jnah. 3 These again have their subsidiary branches, Upangas, 
vedah siihgopahgah savistarah, iii, 64, 17 ; Unarms’ and Brha- 
spati’s §astra with Angas and Upangas, i, 100, 36-38. The 
similarity of phrase in iii, 99, 26 and elsewhere, vedah siiiigo- 
panisadah, might suggest that U pangas were Upanishads, but 
they are more probably a species of Upavedas. The term is 

1 This Veda is constantly mentioned, e. g. i, 130, 21 ; 221, 72 ; iii, 37, 4; ix, 
6, 14, da^angam ya? catuspadam isvastrarii veda tattvatah, sangahs tu caturo 
vedan samyag akhyanapancaman. The phrase dhanurvede ca rede ca occurs, 
for example, in i, 109, 19. In R. y, 35, 14, Rama is described as “trained in 
the Yajur Veda . . . and skilled in dhanurvede ca vede ca vedaiigesu ca (the 
Yajur Veda only, to which Valmiki belonged, is here mentioned). Elsewhere 
the science takes its proper place, as in M. iii, 277, 4, yedesu sarahasresu dha- 
nurvedesu paragah, where the plural is noteworthy. 

2 iii, 20, 10, trihsama hanyatam esa dundubhih. The Tina madhuralapa, 
sweet-voiced lyre, is spoken of as gandharvam sadhu murchati (= murcha- 
yanti), iv, 17, 14. The gandharvam is the third note of the seven, xii, 184, 39 
=: xiv, 50, 53. 

8 Compare brahme vede ca paragah contrasted with astranam ca dha- 
nurvede, vii, 23, 39. So Brahma Veda, R. i, 65, 23 (above), not as AV. 



one associated with Jain rather than early Brahmanic litera- 
ture, and is not explained by the commentator. 1 Yedas, 
Puranas, Angas, and Upangas are sometimes grouped to- 
gether, as in xii, 335, 25 (veclesu sapuranesu sangopangesu 
giyase, the prior pada found again, e. g. in 342, G). The 
Angas commonly mentioned in particular are the calendar- 
knowledge, Jyotisa, and etymology, Niruktam. The latter 
word, indeed, generally means only an explanation of the 
meaning of a word, but it occurs also as the title of a specific 
literary work in xii, 343, 73, where we find mentioned not 
only “Yaska’s Nirukta,” together with Naighantuka, hut 
vocabularies and lexicographies. 2 A curious contemplation 
of Krishna as the divine sound in xii, 47, 46 analyzes him 
grammatically, “with joints of euphony and adorned with 
vowels and consonants.” 3 

Astronomical similes are not infrequent. Thus Arjuna 
storms about “ like Mars in his orbit.” 4 An indication that 
one science as a specialty is not much regarded is seen in the 

1 The later Upangas are the Puranas (and upa-) : Logic, nyaya and vai- 
yesika; Philosophy (including Vedanta), mimahsa; and Law-books (including 
Samkhya-yoga and epics), dharmayastra. The epic use, as will be seen from 
the citation above, differentiates Puranas from Angas and Upangas. For the 
later meaning, see Weber IS. i, p. 13. 

2 ib. 83, 88 : naighantukapadakhyane, niruktarii vodaviduso vedayabdartha- 
cintakah. The common meaning, “ explanation,” may be surmised in xii, 340, 
50, caturvaktro niruktagah (in both editions), where the avagraha is certainly 
required, “ inexplicable,” despite Taitt. Up. ii, 6. 

3 In xiii, 17, 111 (where siddhartha, according to Nilakantha, is siddlianta), 
fiva is siddharthakari siddharthay chandovyakaranottarah. Kalpa and 
Jyotisa are united, kalpaprayoga and jyotisa, in xiii, 10, 37. In ii, 4, 18, 
Kalapa and Katha are mentioned; in R. (not G.) ii, 32, 18, the Kathakalapas 
(after the acaryas taittiriyanam in 15). SI. and G. (only) have (Jandiiya and 
ICauyika (with Gargya in G.) in the same list, and SI. has Tittiri (with Yajfia- 
valkya). In M. they are vedavedaiigaparagah ; in R., vedaparagah. R. calls 
Trijata (Piiigala) a Gargya in 20 (Aiigirasa in G. ; cf. R. 33). 

4 viii, 19, 1, vakrativakragamanad angaraka lva grahah. Compare budh- 
angarakayor iva (a battle-phrase). The Yedangas and Upavedas are often 
grouped together, as in i, 1, 67, where yiksa, phonetics, is grouped with nyaya, 
rules, and cikitsa, medicine. In i, 70, 40-44, the same passage where pada 
and krama are mentioned (above), yabda (samskara), yiksa, chandas, nirukta 
and kalajfiana are found with philosophy. A priest who is yiksjk^aramantra- 
vit gets gold niskas, etc., iii, 23, 2 ; 30, 42. 


fact that the cultivator of the Upaveda medicine and of the 
Afiga astrology are both excluded from society, although it 
should be added that the man intended is one who “lives by 
the stars,” naksatrair yag ea jivati. Such a fortune-teller is 
classed with rhapsodes and physicians, xiii, 90, 11. The diffi- 
culty of reconciling the data of astrology (fortune-telling) and 
the theory of Karma is alluded to in iii, 209, 21 : “ Many are 
seen to be bom under the same lucky star, but there is a 
great difference in their fate.” The most surprising astro- 
nomical statement in the epic is to the effect that stars are 
really very large and only appear small on account of their 
distance. 1 The kiilajnana or “knowledge of time,” already 
mentioned, is attributed especially to Garga, who, as Weber, 
Lectures, p. 237, has noticed, is associated with Kalayavana : 
“ Kalayavana who is endued with Garga’s (brilliancy or) 
power,” xii, 340, 95. This same Garga is credited not only 
with having kalajflanagati and jyotisfuh vyatikrama, “thor- 
ough knowledge of times and mastery of science of stars,” 
ix, 37, 14-16, but also with kalajnana, or the fine arts. That 
the epic has a different order of planets from that of the 
third century A. d. has already been observed by Jacobi. 2 

The Upavedas, however, pass the Vedic stage. There re- 
mains a word to say on the older Sutras, to which may be 
added an account of those more frequently mentioned Sutras 
and other treatises which are quite beyond the Vedic pale. 


A Vedasutra, apparently a Crautasutra, but perhaps only 
Veda in general, 3 is mentioned once, in xii, 341, 63. Grhya- 
sutras are not mentioned by name, but may be implied in the 
word Veda, as will be seen in the quotation given below. 
The Dharmasutras are apparently implied in one passage of 

1 dipavad yiprakrstatvat tanuni sumahanty api (tararupani), iii, 42, 34. 

2 ZDMG. rol. xxx, p. 307; Holtzmann, Das Midi. vol. iv, p. 114. 

8 The Supreme Lord says that the god who gives him a share gets by the 
Lord’s grace a corresponding (Veda-arranged) sacrificial share in (i. e. accord- 
ing to) the Vedasutra. 



tlie thirteenth hook, where a Sutrakara in one verse corre- 
sponds to Vedas in the next, in a passage cited from the Saihhita and Law-books (see below) ; and in another, 
where agaknuvantag caritum kimeid dharmesu sutritam, “ un- 
able to do what is sutrified in the laws,” xii, 270, 30, must 
refer to the general class of legal Sutras. The Gita, 13, 4, 
mentions the Brahmasutra, which is probably nothing but 
an equivalent of Vedasutra, that is, equivalent to Veda in 
general ; but it may be one of the late marks of this poem 
(the Brahmasutra being otherwise unknown before the Hari- 
vahtja) and mean the philosophical Sutra. 1 Sutrakaras and 
Sutrakartars, “who will arise,” are mentioned prophetically 
a few times in the didactic epic. 2 

Profane Sutras are jumbled together in one of the latest 
stanzas of the Kaccit chapter, ii, 5, 120, to which I have 
alluded before j « Dost thou understand the Sutras on 
elephants, horses, chariots, catapults, and the Dhanurveda 
Sutra? ” 

As early as Panini there were Sutras of all sorts and the 
mention of such works has only the special value of indicat- 
ing that the epic belongs to a time when Sutra meant works 
which were probably popular and not written in aphoristic 
style. They were doubtless the same as the various (yastra 
and other treatises to which reference is often made. Some 
of these works are called ^astras and are grouped with the 
fine arts mentioned above as known to Garga. Arthacastra 
and Kamagastra, by-names of the epic itself, are mentioned 
in the late introduction to the whole work. The fine arts, 
kalas, are mentioned or implied in three places. First the 
slave-girls of Yudhisthira are said, at ii, 61, 9-10, to be 
“ versed in dancing and songs,” saroasu, and “ skilled in the 

1 In xii, 327, 31, there is mentioned a Moksagastra, inspired by gathah pura 
gltah, a treatise which is based on verses recited (by Yayati) in regard to 
proper behavior, and it is partly philosophical. 

2 xiii, 14, 101-104, granthakara, sutrakarta (bhavisyati), granthakrt ; 16 70 
eutrakartar. In xii, 245, 30, svagastrasutrahutimantravikramah, sutra may 
be the thread (a brahma-sutra as elsewhere), but in the connection seems 
more likely to mean Sutra. 


sixty-four,” which must imply the sixty-four kalas. Then 
Garga, who knows kalajnana and omens, utpatas, is also 
acquainted with kalajnana catuhsastyanga, xiii, 18, 38, which 
shows that the fine arts were not exclusively for women 
and slaves ; as is also indicated by the passage xiii, 104, 
149 ff., where, as befitting a king to know, are mentioned 
treatises on logic (or behavior?), on grammar, on music, 
and the fine arts ; and to hear, Legends, Tales, and adven- 
tures of the saints. 1 It is interesting to see that these 
“sixty-four arts,” still typical of culture, are proverbial in 
India to-day. A Marathi proverb says eauda vidya va cau- 
sasta kala, “ fourteen sciences and sixty-four arts.” 2 


But if Sutra literature, except in the few instances cited 
above, is practically ignored, all the more fully is (Jastra 3 
and particularly Dharma§astra literature recognized; which 
I may say at the outset shows that the later epic was 
composed under the influence of Dharma§astras rather than 
of Dharmasutras. 

The general term Nltig.astra, code of polity, has already 
been noticed. A number of such codes is recognized, xii, 
138, 196, and Dharma(gastras) are cited not infrequently; 

1 yuktiyastram ca te jfieyarii yabdayastram ca, Bharata, gandharvayastram 
ca kalah parijiieya, naradhipa ; puranam itihasay ca tatha ’khyanani yani ca, 
mahatmanam ca caritam yrotavyam nityam era te. The yuktiy astram is not 
explained. According to PW, it is a manual of etiquette, but perhaps logic ; 
possibly the unique system of logic and rhetoric developed by Sulabha in 
xii, 321, 78 ff. 

2 Man waring, Marathi Proverbs, No. 1175. This is late. Cf. Yaju. i, 3, 
and Vayu Purana, lxi, 78—79. In the latter passage, the four A edas, six Angas, 
Mlmansa, Nyaya, Dharmayastra and Purana make the “fourteen vidyas ” or 
“eighteen” including the three Uparedas and the Arthayastra. 

:i Or Smrti, but this word seems of wide bearing. Just as agama (above) 
includes more than Veda, so Smrti includes all tradition. In xii, 200, 30, 
mahasmrti and anusmrti seem to be interpreted by the commentator as Samhi- 
tas and Vedaiigas (with Munu and others) respectively, but his first words 
may refer to the inferred Veda of the preceding japaka (the reciters of both 
go ipso facto to heaven). Besides Manu (above), lama, Angiras, Brhaspati, 
Uyanas, and Parayara are specially cited as law-givers. 




while a general rule is given as a Dharma-gasana, e. g., i, 
72, 15: 

Three fathers have we, for e’en thus 
Law's statute says, ’t is meet 
To call our sire, and who saves life. 

And him whose food we eat. 

Manu’s Dharmagastra is referred to under that name only 
in one of the latest books of the pseudo-epic. In the early 
books his Rajadharmas are once mentioned, iii, 35, 21, which 
might imply a chapter of our present code, but otherwise 
only his Dharmas are referred to, though generally merely an 
ipse dixit of Manu is cited, which, however, is often a dic- 
tum opposed to the actual words of the extant Manu text. 
The epic poets do not always recognize Manu as in any wise 
supreme, often not even as prominent. A typical example is 
furnished by iii, 150, 29: “Gods are upheld by Vedic sacri- 
fices; men are upheld by the laws (not of Manu but) of 
Uganas and Brhaspati.” 1 But in xii, 336, 39-45, a primeval 
code, anugasana, of 100,000 glokas, gives rise to the “laws 
which Manu the self-existent will declare and Uganas and 
Brhaspati,” where there is a clear reference to the code of 
Manu ; as in the next stanza, where are mentioned the “ laws 
of the Self-existent, the Castra made by Uganas and the opin- 
ions of Brhaspati” (a gastram sangopanisadam, 54 ) ? 

The mere order of names, however, is no more indicative 
of priority than in the case of the Vedas mentioned above. 
Another list of Rajagastra-pranetaras at xii, 58, 1-3, 13, 
begins with Brhaspati and Uganas (Kavya, cited with two 
gathas at xii, 139, 70), and then follows Praeetasa Manu, 
Bharadvaja, and Gauragiras, with the gods between. So in 
the next section, 59, 81 ff., (liva reduces Brahman’s work, 

1 So in iv, 58, 6, Bharadvaja was “equal to Uganas in intelligence, to Brha- 
spati in polity,” naya; ix, 61, 48: “Have you not heard the instructions, 
upadega, of Brhaspati and Uganas? xii, 122, 11: “You have perused the 
opinions, matam, of Brhaspati, and the £astra of Uganas,” as the authorities 
generally recognized. Bharadvaja has three roles in the epic, as archetypical 
jurist, physician, and teacher of arms, according to the passage. 

2 Compare xii, 59, 80, ff. 


which in turn is reduced by Indra, as the bahudantaka, and 
then by Kavya Yogacarya, a work which embraces Itihasas, 
Vedas, and Nyaya (141) or laws. 

More important is the fact that references to Manu’s laws 
in the early books are seldom verifiable in our present code, 
while references in the didactic epic more often than not cor- 
respond to passages of the extant text. 1 Hence it may be 
inferred that that part of the epic w'hich agrees most closely 
in its citations with our code is later than that portion which 
does not coincide, or, conversely, that the text of Manu was 
shaped into its present form between the time of the early 
epic and that of the didactic epic. In the first period, when 
Manu’s Dharmagastra was unknown, Manu was merely a 
name to conjure with. The verses thus ascribed to Manu 
were not all put into the code when it was formed and for 
this reason the earlier citations are not generally found in 
our text. Some of them were adopted, however, and the 
later epic writers therefore agree more closely with the £ns- 
tra as it is to-day; though no one who understands how 
works are enlarged in India will expect to find all the quota- 
tions verified, even in the later epic, for there is no reason 
to suppose that the code was exactly the same two thousand 
years ago as it is to-day. But in fact, out of eleven quota- 
tions from Manu in the thirteenth book, there is only one 
which does not correspond with our Manu text, and this is 
of a general character, to the effect that a graddha with tila 
is undecaying, “said Manu.” 

1 So in the Ramayana there are two evidently interpolated chapters at ir, 
17 and 18. Kama in the subsequent chapters is incidentally charged (with 
great truth) with having violated every knightly rule in slaying Vali. To 
offset this clear case of sin on the part of the divine hero, a formal charge 
and defence is inserted (just the procedure in the Mahabharata !) in chapters 
which metrically belong to the classical period, so close is the adherence to 
vipula rule. Just here it is that Mamina gitdu f lokau are cited, viz., Manu, 
viii, 318 and 316 (inverted order), almost verbatim. Elsewhere Manu is a 
sage merely, not a cited law-giver, as here, iv, 18, 30-31 (without reference to 
Manu in G.). These chapters need no further proof than the reading to show 
their true character. They are simply banal, especially Rama's speech, as 
well as contradictory in substance to the preceding and following chapters. 



In a previous discussion of tbis subject in the Journal Am. 
Or. Soc. xi, p. 239 ff. (where will be found more data on the 
subject of legal literature in the epic), 1 in order not to force 
my argument I included as unverified a quotation at xiii, 65, 
3, “ Manu said that the highest gift is something to drink,” 
pamyam paramam danam, because it was in connection with 
Tirthas. In this I was certainly over-scrupulous, for the 
words could easily refer to the passage I there cited from 
Manu, iii, 202, vary api (jraddhaya dattam aksayayo ’pakalpate, 
“even water given with faith fits for immortality.” I can 
now add to this another quotation, xiii, 67, 19, toyado . . . 
aksayan samavapnoti lokan ity abravln Manuh, “ a giver of 
water obtains imperishable worlds.” Further, I rejected as 
uuverifiahle the statement that Manu said the king gets a 
fourth part of the sin of the people (instead of the usual 
sixth), although, as I pointed out, this proportion actually 
occurs in Manu, only it is for a specific occasion. Neverthe- 
less as Manu, viii, 18, says pado rajanam arhati (or rcchati), 
it is clear that the quotation caturtham etc. in xiii, 61, 34 
cannot be said to be “unverifiable.” It is simply a free ren- 
dering verbally of a statement actually found in Manu. 3 

We have here the incontrovertible fact that, while the 
other books of the epic before the thirteenth in giving quo- 

1 For example, the fabulous books of divine origin of xii, 59, 80 ff. (like the 
origin of Narada’s law-book), called Barhaspatya, etc., according to the dia- 
dochos ; the “ law and commentary,” savaiyakho dharmah, of xii, 37, 10, etc. 
(pp. 254 and 248), and other points to which I may refer the reader without 
further remark than the references already given. 

2 Besides the quotation given above from the thirteenth book and verifiable 
in our present code, I may add iii, 92, 10 : “ By Manu and others (it is said 
that?) going to Tirthas removes fear,” manvadibhir maharaja tirthayatra 
bhayapaha, if this be the meaning, which is rather doubtful. In any case 
it only adds one more to the unverified citations from the early books, but 
it may mean only that Manu and others have journeyed to Tirthas. Compare 
also xii, 266, 5, sarvakarmasv ahihsa hi dharmatma Manur abravlt, “ Manu 
the righteous proclaimed that one should not injure (animals) at any cere- 
mony.” From the context, killing cattle at a sacrifice is here reprobated. 
This is a perversion for sectarian purposes of Manu’s rule v, 43, na ’vedavi- 
hitam hinsatn apady api samacaret, to which perversion some color might be 
given by the following verses, which speak harshly of all injury to living 
creatures. I think no other quotations from Manu will be found in the epic. 


tations from Manu agree with our present text of Manu only 
in one third to one half the instances, the thirteenth book has 
eleven citations, of which ten agree with the statements of 
our code. To this must be added the fact that only the thir- 
teenth book recognizes “ the (^astra declared by Manu.” I do 
not know any other literature where such facts would not be 
accepted as of historical importance, and they have been so 
regarded here by competent scholars. In the opinion which 
I first set forth in 1885, the late Professor Biihler in general 
concurred, though inclined to believe that the authors of the 
twelfth and thirteenth books did not know the identical 
^astra which we have to-day. As Professor Biihler’s position 
has not always been cited with the reservations made by him, 
I will cite his own words : “ It remains indisputable that 
the author or authors of the first, twelfth, and thirteenth 
Parvans of the Mahabharata knew a Manava Dharmnga&tra 
which was closely connected but not identical with the ex- 
isting text,” Manu p. lxxix, and again : “ The answer which 
we are thus obliged to give to the question whence the author 
of our Manu-Smrti took his additional materials agrees very 
closely with Professor Hopkins’ hypothesis,” p. xci. Never- 
theless, despite this admission, Professor Biihler, by a line of 
argument which is based chiefly on the lack of absolute 
identity, assumes finally that the authors of the epic “ knew 
only the Dharmasutra,” ib. p. xcviii. The arguments other 
than the lack of total identity are, first, that Manu shows an 
acquaintance with the epic because he says that in a former 
kalpa the vice of gambling has been seen to cause great en- 
mity; in regard to which Professor Biihler says: “This asser- 
tion can only point in the first instance to the match played 
between Yudhisthira and Duryodhana,” p. lxxx. But why 
not to the story of Nala, as Professor Biihler himself suggests, 
or any other story of dicing resulting in “ enmity ” which may 
have preceded our epic ? Another argument is, that legends 
referred to in the (JJastra are found in the epic, ib. But it is 
of the very character of the epic that it contains many ancient 
legends, gathered from all sources. It does not follow in the 



least that Manu took them from the epic. On the other hand 
it is important to observe that in no such passage does Manu 
refer a single one of them to an epic source. Thirdly, it is 
claimed that the passages parallel in epic and yastra often 
have verses in a different order, with omissions, etc., that, in 
short, they are not actual copies one of the other. But Pro- 
fessor Biihler himself has shown that “the existing text of 
Manu has suffered many recasts,” p. xcii, so that we do not 
know the form of the Castra to which the epic explicitly refers 
and from which it cites as the (yastra set forth by Manu. For 
my part, it still is impossible for me to believe that when the 
pseudo-epic, in particular the Anugasana, refers to (yastras , 1 
and cites correctly from “Manu’s Castra,” it really knows 
only Sutras. 

A Manava Dharmagastra, specifically, must from the evi- 
dence be regarded as older than the later epic but later than 
the early epic, which knew only a mass of royal and general 
rules, dharmas, generally ascribed to Father Manu but differ- 
ent from those in our extant (yastra. With this result too 
agrees the fact that the metrical form of the extant code is 
distinctly earlier than that of the later epic. Not unimpor- 
tant, finally, is the circumstance that the extant code only 
vaguely refers to epic Tales, but recognizes neither of the 
epics, only legends that are found in the epics. In all prob- 
ability the code known to the later epic was not quite our 

1 In xii, 341, 74, are mentioned “ teachers in Dharmagastras,” acarya dharma- 
gastresu ; in xiii, 61, 34, Manu’s anugasana ; in xiii, 47, 35, “ the £astra com- 
posed by Manu,” manuna ’bhihitarii gastram ; in xiii, 45, 17, “ those that know 
law in the law-books,” dharmagastresu dharmajhah, in reference to the sub- 
ject discussed in Manu iii, 52-53 ; iv, 88. Similarly, xiii, 19, 89. In most cases 
here Qastras are the authority, which in iii, 313, 105, are set beside the Vedas 
as two standard authorities. In the face of these citations it is difficult to 
understand Biihler’s words, “ the authors . . . knew only the Dharmasutras,” 
especially as the words contradict what he says in the same essay on a 
different page, “the authors . . . knew a Manava Dharmagastra” (loc. cit. 
above). It has seemed to me that the great scholar was unduly influenced 
in his final word by his general desire to put back the epic as far as possi- 
ble. Professor Holtzmann, who has collected the material, loc. cit., p. 115 ff., 
is of the opinion that “our Manavaadharmagastra is certainly much later 
than the older parts of the Mahabharata." 


present code, but it was a code much like ours and ascribed 
to Manu, a (yiistra which, with some additions and o mi ssions, 
such as all popular texts in India suffer, was essentially our 
present text. 

Vedic Citations in the Epic. 

We have now reached and indeed already passed, in the 
notice of some of the works mentioned, the point where the 
epic impinges on the earlier literature. Before going further 
I will illustrate the statement made at the outset that the 
epic cites freely or parodies Yedic documents. The free 
rendition in Veda-like verse of the older hymnology is not 
uncommon. Thus in v, 16, the opening hymn is not strictly 
Vedic, but it is very like a collection of Vedic utterances put 
into popular form and these verses are called brahma man- 
trah, gl. 8. Apart, however, from such instances of more or 
less exact imitation of general Vedic verses, 1 we find a num- 
ber of verses plainly imitative of extant Vedic passages or 
almost exactly reproducing them. This applies to reproduc- 
tions or imitations 2 of the chief Vedic literature from the 
Rig Veda to the Sutras, as will be seen from the following 
examples : 

Rig Veda x, 117, 6, 

mogham annam vindate apracetah 

1 There are, of course, also a vast number of verses such as gaur me mata 
vrsabbah pita me, introduced, as here, with the fiat imarii frutim udaharet, 
xiii, 76, 6-7 ; or with the more usual tag, iti crutih, as for example, agnayo 
mansakamai; (starred in pw.) ca ity api oruyate frutih, iii, 208, 11 ; or with 
smrta, as in a^vinau tu smrtau ?udrau, xii, 208, 24 ; as well as such phrases 
as that of xiv, 51, 26, yas tam veda sa vedavit, all of which reflect the litera- 
ture of the earlier periods. 

2 The Yedic work most frequently referred to is the Yajur Yeda Hymn, 
trisauparnam brahma yajusam cjatarudriyam, xii, 285, 138 ; samaveda 9 ca ve- 
danam yajusam <;atarudriyam, xiii, 14, 323 ; tad brahma fatarudrivam, vii, 81, 
13; vede ca ’sya samamnatam jatarudriyam uttamam, vii, 202, 120; grnan 
brahma param £akrah ^atarudriyam uttamam, xiii, 14, 284. It is imitated 
over and over again, and some of the epic hymns call themselves by the 
same name, a fact alluded to in the words : vede ca ’sya vidur viprah fata- 
rudriyam uttamam, Vyaseno ’ktam ca yac ca 'pi upasthanam, xiii, 102, 23. 



Mbh. v, 12, 20, 

mogham annam vindati ca ’py acetah 

Bohtlingk, Spruch 4980. 

Rig Veda vii, 89, 2, 

drtir na dhmato, adrivah 

Mbh. iii, 207, 47; xii, 95, 21, 

mahadrtir iva ’dhmatah 

(papo bhavati nityada, iii, 207, 47) 

Rig Veda i, 10, 1, 

gayanti tva gayatrino arcanti arkam arkinah 
brahmanas tva (jatakrato ud vanqain iva yernire 

Mbh. xii, 285, 78, 

gayanti tva gayatrino arcanti arkam arkinah 
brahmanam tva 9 atakratum urdkvam kham iva 

Holtzmann, Das Mahabharata, iv, p. 12 ; also for the following 
parallel, p. 13: 

Rig Veda x, 129, 1-3, 

na ’sad asln no sad asid tadanlm . . . 

no ratria ahna aslt praketah . . . 
tama aslt tamasa gulham agre 

Mbh. xii, 348, 8, 

(nidarqanam api hy atra) nasld aho na ratrir asln na sad asln 
na ’sad aslt, tama eva purastad abhavad vi^varupam 

Compare also with Rig Veda, i, 13, 4, asi hota manurhitah, Mbh. 
ib. 10-11, 

tvam agne yajnanam hota viqvesam hito devanam manusanam 
ca jagata iti, nidarcanaiii ca ’tra bhavati, viqvesam agne yajnanam 
tvarii hote ’ti, tvam hito devair manusyair jagata iti 

Rig Veda x, 14, 1, 

vaivasvatam sariigamanam jananam 

Mbh. xiii, 102, 16, 

vaivasvatl samyamani jananam 


Further, with Rig Veda i, 164, 46, ekam sad vipra bahu- 
dha vadanti, and x, 114, 5, viprah . . . ekam santam bahu- 
dha kalpayanti, may be compared Mbh. (v, 16, 2, and) i, 232, 
13, manlsinas tvarii jananti bahudha cai ’kadha ’pi ca. In xv, 
34, 11, devayana hi panthanah grutas te yajnasamstare 1 is an 
allusion to Rig Veda x, 18, 1; while in xii, 312, 5, dyava- 
prthivyor iti esa . . . vedesu pathyate, the citation of a Vedic 
phrase is acknowledged ; whereas in the epic phrases ma 
rlrisah and bhuvanani vi§va, vii, 201, 77, no indication of 
Vedic origin is given. 

Taitt. Samhita i, 16, 11, 1 ; 4?at. Br. i, 5, 2, 16, 
ye yajamahe 

Mbh. iii, 180, 33, 

idam arsam pramanam ca ye yajamaha ity api 

Compare iii, 31, 22, yasya na ’rsam pramanam syat, etc. 
Aufrecht, apud Muir, OST. i, 137. Also Taitt. S. ii, 5, 1, 1 
is repeated verbatim Mbh. xii, 343, 28, as shown by Weber, 
Ind. Stud, i, p. 410. 

Mait. Samhita i, 10, 11, 

stry anrtam 

Mbh. xiii, 40, 12 and 19, 6-7, 

striyo 'nrtam iti qrutih; anrtah striya ity evam vedesv api 
hi pathyate; anrtah striya ity evam sutrakaro vyavasyati. 

Compare Bauclh. Dh. S. ii, 3, 46, with Biihler’s note, and 
Manu ix, 18, striyo 'nrtam iti sthitih (v. 1. §rutih). The 
double reference in the epic, Sutrakara and V edah, may point 
to the same place, or the writer may have had in mind a 
Sutra passage parallel to Baudhayana, if not Baudhayana 
himself, whose text here is corrupt. 

1 In the preceding verse is cited an arvaim-dliarruti, apropos of the a^va- 
samjnapana : lokantaragata nityam prana nityam raririnam. With the text 
cited above, compare dvav etau pretya panthiinau, etc., xii, 329, 30. The 
Upanishads would suffice to explain some of these phrases. 



Atharva Veda? Hbh. xiii, 98, 30, 

osadhyo raktapuspag ea katukah kantakanvitah catrunam 
abhicarartham atharvesu nidargitah; viii, 69, 83-86, tvam ity 
atra bhavantam hi bruhi . . . tvam ity ukto hi nihato gurus 
bhavati . . . atharvangirasl hy esa Qrutlnam uttama qrutih . . . 
avadhena vadhah prokto yad gurus tvam iti prabhuh. 1 

Ait. Brah. i, 1, 

agnir vai sarva devatah 

Mbb. xiv, 24, 10 (read vedasya ?), 

agnir vai devatah sarvah, iti devasya Qasanam 

Mbb. xiii, 84, 56, 

agnir hi devatah sarvah, suvarnam ca tadatmakam 

Holtzmann, loc. cit. p. 14. 

(kit. Brahmana in Mbh. xii, 343, 13-15, 

yajnas te devahs tarpayanti devah prthivTm bhavayanti, Qata- 
pathe 'pi brahmanarnukhe bhavati, agnau samiddhe juhoti yo 
vidvan brahmanamukhena ’hutim juhoti, evam apy agnibhuta 
brahmana vidvahso 'gnim bhavayanti. 

On this and other citations from Samhitas and Brahmanas, 
compare Holtzmann, loc. cit., p. 14 ff., with especial reference 
to verses cited by Weber, Lectures, p. 137-138; IS. i, p. 277. 
To these I may add a passage reflecting the Brhad Aran. Up. 
of this Brahmana, Up. 1, 5, 14 (where the chief verbal iden- 
tity is in soda 9 aya kalaya), expressly said to be from the 
Rsi’s “ more extended ” exposition of the subject : viddhi 
candramasarhdanje suJcsmaya kalaya sthitam, tad etad rsina 
proktam vistarena ’numiyate, Mbh. xii, 242, 15-16 (compare 
sodagakalo dehah ; and 305, 4). The commentator refers the 
passage to this Upanishad, as cited. 

1 According to xiii, 163, 53, tvamkara (to superiors) is vadha, and is em- 
ployed only in speaking to equals, inferiors, pupils, etc. Compare Chand. 
Up. vii, 15, 2. Echo arose in the mountains (compare Callimachus, Ep. xxviii) 
from the care with which Ouka addressed his superior Vyasa with bho, bho , xii, 
334, 25-26. 


The citations in the Ramayana I have not examined, but 
have noted by chance two; Rig Yeda i, 22, 20; Katha Up. 
iii, 9 ; Maitri, vi, 26 : tad visnoh paramam padam (sada 
pagyanti surayah) ; G. vi, 41, 25, tad visnoh paramam padam 
(nihato gantum icchami) ; and satye sarvam pratisthitam in 
Mahanar. Up. 22, 1 ; satye lokah pratisthitah, R. ii, 109, 10. 

Upanishads in the Epic. 

Sporadic parallels between the epic, generally the Gita, 
Anuglta, and <y'anti, and various Upanishads have often been 
noticed. As illustrative material all these passages are val- 
uable, but they give no evidence that the epic has copied, if 
the mutual resemblance is only of general content or is given 
by similar or even identical verses, when these are not con- 
nected as in the supposed model. As this material has been 
put together by Holtzmann, loc. cit., p. 21 ff., I may refer the 
reader to his parallels, 1 while pointing out that it is histor- 
ically of little importance whether the oldest Upanishads are 
cited if we can satisfy ourselves that the epic draws on Upa- 
nishads of the second and third period, not only sporadically 
but connectedly. In regard to the earliest works, it is enough 
to refer to the passage condensed from the Brhadaranyaka and 
cited above. This is the only one of the oldest Upanishads 
certainly cited, though the Chandogya, Aitareya, and Kau- 
sltaki have many parallels with the epic, as have among the 
later works of this class the Ivena, Mundaka, Pragna, and a 
few others. Oddly enough, the Maitrayana has been scarcely 
compared, 2 but I purpose to show that this and the earlier 
Kathaka were certainly copied by the later epic poets. 

1 Not all the “ Vedic ” verses are here verified, e. g., Taitt. iii, 7, has prano 
va annam. This is cited in the epic as Vedic: annam prana iti yatha vedesu 
paripathyate, xiii, 95, 22. The Gita distributes older material, e. g., (Jvet. iii, 
17 = Gita, 13, 14, but the following pada, navadvare pure dehi, is found in 
Gita, 5, 13, etc. 

2 The verse dve brahmani (as duly recorded by Holtzmann) was located by 
Hall, and Biihler has compared two more verses with xii, 330, 42-43 (Manu, 
p. 212), while Telang has illustrated the Gita with general parallels. 



The Cvetacvatara Upanishad. 

This may be loosely copied, but, except for one parallel, 
the mutual passages are common to this and other sources. 
I cite as exemplifying a possible copy (though the Upanishad 
itself is a copy of the older Katbaka) : 

Upanishad. Epic. 

iii, 8 = V. S. 31, 18, 

tamasah parastat ; na ’ayah pan- 
tha yidyate ayanaya. 

iii, 10, 

tato yad uttaratararii tad arupam 
anamayam, ya etad vidur amrtas 
te bhavanti. 

iii, 13, 

angusthamatrah purusah, see be- 

iii, 18, 

navadvare pure dehl hansah, see 

iii, 19-20, 

sa vetti vedyam . . . anor ani- 
yan, etc. 

iv, 2 and 19, 

tad era gukram tad brahma ; 
yasya nama mahad yagah, see 

iv, 5, 

ajam ekam lohitaguklakrsnam. 

iv, 6, 

Birds and pippal, see the passage 
from Drona, cited hereafter. 

iv, 17 and 20, 

na samdrge ; hrda manlsa, see be- 

v, 44, 29 and 24, 

tamasah parastat ; na ’nyah pan- 
tha ayanaya yidyate- 

v, 44, 81, 

anamayam tan mahad udyatam 
yago ( Katha, vi, 2, mahad bhay am 
vajram udyatam) vaco vikararii 
kayayo vadanti yasmin jagat 
sarvam idam pratisthitam ye tad 
vidur amrtas te bhavanti (com- 
pare BAU. i, 5, 1 ; Chand. iii, 12, 
2; Katha, vi, 9). 

v, 43, 53; 46, 31 (Gita, 10, 

yo veda vedyam na sa veda sa- 
tyam; anor aniyan (Katha i, 2, 
20). In 44, 29, aniyo ruparii ksu- 
radharaya samam (Katha, iii, 14). 

v, 44, 25 and 26, 

abhati guklam iva lohitam iva 
krsnam (followed by ayasarn 
arkavarnam with v. 1., atha’iija- 
nam kadravam va) ; Malianar., 
ix, 2 ; also Chand. viii, 6, 1. On 
account of the varied reading in 
the same verse the three first 
colors may be the only original, 
but even here the reference is to 
Prakrti in the Upanishad and to 
Brahman in the epic. 


These are the best examples of sporadic parallels to be 
found in the Upanishads. I turn now to the Kathaka. 

The Kathaka or Katha Upanishad. 

From the Katha, iii, 10, indriyebhyah para hy artha, arthe- 
bhyag ca param manah, manasas tu para buddhir, buddher 
atma mahan parah, and ii, 19, na ’yam hanti ua hanyate, the 
Gita, 3, 42, has inclriyani parany ahur indriyebhyah param 
manah, manasas tu para buddhir, yo buddheh paratas tu sah 
(the Sa is higher than intellect) ; and in 2, 19-20, it inverts 
and modifies the na jayate and hanta cen manyate hantum 
stanzas. Less precise in rendering, but important on account 
of the Gita modifications, are two other stanzas. Katha i, 22, 
has vakta ca ’sya tvadrg anyo na labhyah, etc., a tristubh, 
whereas Gita, 6, 39, has tvad anyah saihgayasya ’sya chetta 
na hy upapadyate, a gloka (compare M. ii, 15, 1, samgayanam 
hi niimoktil tvan na ’nyo vidyate bhuvi, addressed to Krishna). 
The Katha is older also in the stanza ii, 15, 

sarve veda yat padam amananti, tapahsi sarvani 
ca yad vadauti 

yad icchanto brahraacaryam caranti, tat te padam 
sangrahena bravlmi, 

as compared with Gita, 8, 11, 

yad aksararii vedavido vadanti, viganti yad yatayo 

yad icchanto brahmacaryam caranti, tat te padam 
sangrahena pravaksye. 

Other parallels will be found between Katha ii, 7, 

agcaryo vakta kugalo 'sya labdha, agcaryo jnata 

and Gita, 2, 29, 

agcaryavat pacyati kagcid enam, agcaryavad vadati 
tathai ’va ca ’nyah, etc. ; 

between Katha vi, 1 and Gita, 15, 1 (the idea developed in 
xii, 255, 1 ft.) ; and in a few more instances, such as tasya 



bhasa sarvam idam vibhati, Katha v, 15, and ekah sihyah 
sarvam idaiii vibhati, Mbh. iii, 134, 8. 1 

But it is not necessary to dwell upon these, as the third 
chapter of the Upanishad is epitomized in a section of Canti. 
The later feature begins at the start, xii, 247, 1 ff. The 
vikaras, modifications of Prakrti, do not know the ksetrajna, 
or spirit, but he knows them. Then follows the image of the 
Upanishad iii, 2 ff. The senses are subservient steeds, and 
the spirit is the driver who controls them, samyanta. After 
this general imitation follow the three stanzas of Katha iii, 
10, 11, 12, one of which appears in the Gita (above), 2 but 
with the substitution of amrta for purusa in the second 
stanza, and evam for esa in the third. Then a general like- 
ness follows between the Upanishad’s next stanza (“ restrain 
mind in knowledge, in self ”) and the epic, which says “ sink- 
ing the senses with mind as the sixth in the inner self,” 
“endowing the mind with wisdom,” “one that is not mas- 
tered (by the senses) gets the immortal place.” The instruc- 
tion is a mystery, to be repeated to Snatakas (compare Ivatha, 
iii, 17), and besides containing the gist of former wisdom, “is 
recited in the Upanishads” vedantesu ca glyate, 247, 16, 19, 
21. I think there can be no doubt that the epic section is an 
abbreviation of Ivatha iii, perhaps under the influence of the 
Maitrayana, as shown below. A preceding section may be 
compared with Katha v, 1-2, where the city of eleven doors 
is followed by a reference to the hansa, lord, R. V. iv, 40, 5. 
The epic (see under the “ group of seventeen ”), like the later 
Upanishad, admits only “ nine doors,” and says in xii, 240, 32, 
the hansa lord, 19a, and controller, vagi, enters the city of 
nine doors, because it is controlled, niyatah, by the senses. 

Other stanzas reflecting the last chapters of this Upanishad 

1 Compare in the Up., ib. 9 and 12, agnir yathai ’kah and ekam rupam 
bahudha yah karoti, with eka evagnir bahudha samidhyate, just preceding 
in the epic. Gita, 13, 30, may be a modification of Katha vi, 6. The Gita 
stanza, by the way, is repeated verbatim in xii, 17, 23. 

2 The last of the three verses is cited again in Vana in a copy of the Mai- 
trayana Upanishad, which substitutes bhutatma for gurlho 'tma, and jnanar 
vedibliih for suksmadaryibhih. See the next paragraph. 


are found mingled with copies from other Upanishads in the 
last chapter of the Sanatsujata Parvan. In every case where 
evidence exists it points to the epic being a copy of the Upa- 
nishad. Thus in BAU. v, 1, we read purnam adah purnam 
idam purnat purnam udaeyate, purnasya purnam adaya 
purnam eva, ’vagisyate, which in the epic, v, 46, 10, appears 
as purnat purnany uddharanti purnat purnani cakrire haranti 
purnat purnani purnam eva ’vagisyate. Again the stanza of 
Katha vi, 9, 

na samdrge tisthati rupam asya, na caksusa pagyati 
kagcanai ’nam 

hrda mamsa manasa ’bhiklpto, ya etad vidur arnrtas 
te bhavanti 

is modernized already in £vet., iv 17 (idem) and 20, hrda 
hrdistham manasa ya enam evaiii vidur arnrtas te bhavanti, 
and this in the epic, v, 46, 6, appears as 

na sadrgye tisthati rupam asya, na caksusa pagyati 
kagcid enam 

manisuya ’tho manasa hrda ca, ya enam vidur amr- 
tas te bhavanti, 

or, as ib. 20, 

na dargane tisthati rupam asya . . , 

ye pravrajeyur arnrtas te bhavanti. 

The section begins with an explanation of the gukram brahma 
which is rnahad yagah and tad vai deva upasate, a phrase, 
prior pada, metrically borrowed from the licence of the Upa- 
nishads, where the epic usually writes upasante to avoid di- 
iambus. 1 Here gukram brahma and mahad yagah are from 
Katha v, 8 ; vi, 1 ; £vet. iv, 19 (yasya nama mahad yagah). 
Below, gl. 9, the Agvattha and its birds may be drawn from 
Katha vi, 1, and, after the purnam stanza cited above, gl. 11, 

1 The later Upanishads resort to a similar device. Thus in the Yoga- 
tattvop. i, 6 (alle gute Dinge sind drei) : trayo lokas trayo vedas trayah 
samdhyas trayah surah, trayo 'gnayo gunas trini (sthitah sarve trayaksare). 



tasmad vai vayur ayatah . . . tasming ca prana atatah, is a 
parallel to Katha vi, 2. 1 Then follows, in the epic, gl. 15 : 

angusthamatrah puruso 'ntaratma, liiigasya yogena 
sa yati nityain 

tam Igam Idyam anukalpam adyam, pagyanti mudha 
na virajamanam, 

which appears ib. 27 as : 

angusthamatrah puruso mahatma, na drgyate *sau 
hrdi samnivistah 

ajag caro divaratram atandritag ca, sa tam matva 
kavir aste prasannah, 

with which Katha iv, 4 (matva dhlro na gocati) may be com- 
pared, and especially iv, 12 : 

angusthamatrah puruso madhya atmani tisthati 
igano bhutabhavyasya na tato vijugupsate, 

and Katha vi, 17 : 

angusthamatrah puruso 'ntaratma, sada jananam 
hrdaye samnivistah 

tam svac charlrat pravrhen munjad iva-islkaih dhair- 
yena (tam vidyac chukram) 

The last words are found in the epic, 44, 7, as : 

ta atmanam nirharantl ’ha dehan, munjad isikam iva 

while just before 46, 27, is found in gl. 25 : 

evam yah sarvabhutesu atmanam anupagyati 
anyatra ’nyatra yuktesu kirn sa gocet tatah param, 

which is like Iga 6-7 in contracted form. 

1 There is here a general resemblance, noticeable chiefly because of the 
correlation of one idea with the next following, interrupted in the epic by 
the purna stanza. With 44, 27, “His form is not in stars, lightning, clouds, 
wind, moon, sun,” compare Katha v, 15, “Not there the sun shines, moon, 
stars, nor lightnings.” 


The Maitri Upanishad in the Epic. 

Especially instructive is the form in which the Maitri or 
Maitrayana Upanishad appears in the epic. In the case of 
many of the Upanishads there is lacking any characteristic 
mark sufficiently peculiar to identify the Upanishad when it 
appears in epic form. But the Maitri, as is well known, con- 
tains some special stanzas and above all some special terms 
not found elsewhere except in still later Upanishads. It is, 
therefore, more easily identified, and the possibility that we 
are dealing with material common to the age of the older 
Upanishads is not so great. In all probability it is a later 
Upanishad. Deussen, Sechzig Upanishads, p. 312, success- 
fully maintains this view, and in his Geschichte der Philo- 
sophic i 2 , p. 24, groups it with the Pra§na and Manclukya 
as belonging to the group of “ later Prose Upanishads,” 
regarding it not only as later than the old prose, but even 
as later than the metrical Upanishads, from both of which 
earlier groups I have given epic parallels in the list above. 

This Maitri Upanishad is found reflected in the epic at 
iii, 218, and in a later imitation in the twelfth book. The 
former epic section is based entirely on the Upanishad, and 
the preceding sections appear to be due to an expansion of 
the same material. The order followed is in general that 
of the Upanishad. 

The teaching is called brahml vidya, iii, 210, 15. There 
is an introductory systematization, the assumption of the 
universe (as Brahman) consisting of five elements, 1 earth, 
water, light, wind, air, w'hich have as their characteristics (in 
inverted order), sound, touch, color, taste, smell, so related 
that earth has all five ; water, four ; light, three ; wind, two ; 
air, one (sound), making altogether fifteen in combination 
in all created things (210, 17; 211, 8). With these five 
“ gunas ” begins a group of seventeen : cetana or inanas, mind, 

1 In 210, 17, these are given in reverse order, hut in 211, 3, in their usual 
epic order, bhumir apas tatha jyotir vayur akasam eva ca (reversed, kbam 
vayur agnir apas tatha ca bhuh). 



as sixth; intellect as seventh; egoism as eighth; the five 
senses ; atman, soul, the fourteenth ; and the three gunas, 
rajas, sattvam, tamas. This is “the group of seventeen,” 
which has as its designation the Unmanifest (avyakta) ; to 
which are added objects of the senses and the manifest and 
unmanifest, making the category of twenty-four. 1 

This is the introductory chapter of the discourse, and its 
likeness to the Maitri Upanishad consists in the initial dis- 
cussion of the elements (which, however, are not called 
fine elements, tanmatra, as they are in the Upanishad, iii, 
2, mahabhutiini and gunas), 2 and the statement that this is 
a brahrnl vidya, like MU. ii, 3, brahmavidya. As an indica- 
tion of the age of the discourse, it may be observed in pass- 
ing that, in 211, 9, the fifteen gunas are said to be properly 
correlated in the remarkable verse : 

anyonyam na ’tivartante samyak ca bhavati, dvija 

where the use of bhavati for bhavanti (subject, paHcaclaga 
gunah), though declared by the commentator to be an archa- 

1 Otherwise the commentator. Objects of sense and action-organs are not 
included in the seventeen : ity esa saptada^ako rayir avyaktasamjnakah, 
sarvair ihe ’ndriyarthais tu vyaktavyaktaih susamvrtaih caturvingaka ity esa 
vyaktavyaktamayo gunah (210, 20-21). Guna is obscure. The entirely differ- 
ent group of seventeen in xii, 270, 28, casts no light on the subject, but in xii, 
330, 46, a similar verse has (in B) sarvair ihe ’ndriyarthais ca vyaktavyaktair 
hi samhitah (v. 1. samjuitah) caturvinjaka ity esa vyaktavyaktamayo ganah, 
which gives the needed ganah for gunah and makes the construction some- 
what clearer, though the latter passage is such a careless imitation of the 
one above that in making up the previous list of seventeen, atman, ahaihkara, 
and manas are all omitted from the list (buddlii being represented by mahad 
yat param afravat) and 5 + 1 + 5+ 3 =17! The first group is similar to 
the group of seventeen in the Vedanta-sara, though there the organs of action 
and the breaths are included with the organs of sense, buddhi and manas. 
The formal definition of vyakta and avyakta in iii, 211, 12, repeated in xii, 
330, 40, with grhyate for srjyate and with slight v. 1. in xii, 189, 15, is that 
vyakta, the manifest, is what is comprehended by the senses, while avyaka 
is what is supersensuous, comprehended only by the “fine organs” (liiiga- 
grahyam atindriyam). If the reading guna be retained above, it will imply 
the interpretation of all the constituents as gunas. 

2 That is, here, as synonym of dhatu or the elements, which after the dis- 
solution of the universe appear in every newly formed body, dhatavah panca- 
bhautikah, iii, 211, 11 ; xii, 184, 1. 


ism, is really a late carelessness. It is further to be observed 
that though in this introduction, and incidentally in a pre- 
ceding section, iii, 207, 72, the organs of sense are given as 
five, yet in iii, 211, 24, they are spoken of as six, 1 in a figure 
which not only reproduces the exact language of the Gita, 
2, 60 and 67, but contains the imagery of the Maitri Upa- 
nishad (ii, 6, rathah §arlram, mano niyanta, prakrtimayo 'sya 
pratodah) : 

sannam atmani yuktanam indriyanam pramathinam 
yo dhiro dharayed racmln sa syat paramasarathih 
indriyanam prasrstanaiii hayanam iva vartmasu 
. . . indriyanam vicaratam, etc. 

This image of the senses to be kept under control like horses 
held in check by a charioteer is indeed too general to have 
any bearing on the relation of the epic to the Upanishad (it 
occurs, as said above, in the Katha Upanishad, for instance, 
and again in the epic in purely Buddhistic form at i, 79, 2—3 
= Dhammapada 222-223) and might pass unnoticed, were it 
not that the corresponding section of the twelfth book brings 
the two into somewhat closer relationship. As already ob- 
served, the teaching of the Yana in 210 and 211 is more or 
less closely reproduced in xii, 330, which, however, omitting 
the stanzas in regard to the six senses, condenses them in the 
statement that one is “ tossed about ” by the effects of evil 
actions, but then closes with a stanza, 58, which has direct 
reference to transmigration and is in turn omitted from the 
end of iii, 211, paribliramati samsaram cakravad balmvedanah, 

1 So both groups of organs, those of sense and of action, are sometimes 
counted as making not ten but eleven, including the thinking faculty, as in 
xiv, 42, 12. Compare the same image and number in xii, 247, 2 (above), ma- 
nahsastair ihe ’ndriyaih sudantair iva sarhyanta, etc. In the passage above, 
iii, 211, 13, the sense-organs, indriyani, are defined as apprehenders of objects 
of sense, grahakany esarii gabdadinam. The word is derived from Indra, 
xii, 214, 23, tribijam (apapatha nrbljam), indradaivatyam tasmad indriyam 
ucyate, with a preceding description of the seeds, the ten chief dhamanyah, 
the three humors, vata, pitta, kapha, and other medicinal intelligence, with 
especial weight on the heart-artery, manovaha, and its action as known to 


that is, “ like a wheel he revolves through transmigrations.” 
Just so the Maitri Upanishad, ii, 6, says first that the senses 
are horses and then, after developing the figure, concludes 
with anena (pratodena) khalv Iritah paribhramatl ’dam £arl- 
ram cakram iva mrtpacena, “ thus goaded he revolves in bod- 
ily form like a potter’s wheel.” 

The next chapter of the teaching, iii, 212, discusses the 
three gunas as (in general) in Maitri, iii, 5. The section 
before this in the Upanishad, iii, 4, is a close prose prototype 
of the Canti verses (omitted in iii) just preceding the group 
of seventeen (the rest of the section, xii, 830 being parallel 
to iii, 211). This (xii, 330, 42) verse begins asthisthunam 
snayuyutam . . . carmavanaddham (just as in the Upanishad, 
carmana ’vanaddham), and in 28-9, kosakara iva suggests 
(against the commentator and Deussen) that in the Upani- 
shad, the ending kosa iva vasuna should be interpreted 
accordingly, “filled like a cocoon with (deadly) wealth.” 
The next chapter of Yana, the special chapter under consid- 
eration, begins with the question how the vital flame can 
combine with earth-stuff to make the incorporate creature, 
and how air causes activity. To which the answer is that 
the flame enters the head and directs the body, while air acts 
by being in the head and in the vital flame. This is like the 
opening of the Upanishad where it says, ii, 6, that the spirit 
is fire. The answer continues : “ All is established upon 
breath;” which is identified with spirit, Purusha, intellect, 
buddhi, and egoism. Then follows a disquisition upon the 
different kinds of bodily airs or breathings. These are 
named as the usual five, but are incidentally referred to as 
ten, which makes it necessary to understand with the com- 
mentator that the other five are those called naga, kurma, 
krkala (sic), devadatta, and dhanaihjaya, besides the usual 
(in-) breathing, with-breathing, off-breathing, up-breathing, 
and through-breathing, which are specifically mentioned. 1 

1 iii, 2X3, 16, dafapranapracoditah. The ten are named as above in the 
Yedantasara of Sadananda, 99, Bdhtiingk’s Chrest. p. 204. The (usual) five 
are prana, samana, apana, udana, vyana. The same thing occurs in xii, 185, 


This also corresponds to Maitri ii, 6, where the five breaths 
are associated with the vital flame (Agni Vai§vanara as 

After the breaths are discussed, there is a passing refer- 
ence to the eleven (not sixteen) vikaras, or transformations 
by which the spirit is conditioned like fire in a pot ; 1 just 
as Maitri iii, 3, has first yatha ’gnina ’yaspindo 'nyo va ’bhi- 
bliutah, etc., and then the transformations, gunani (= vika- 
ras). The corresponding passage in (Janti, here 242, 17, has 
karmagunatmakam for nityaih yogajitatmakam, but then both 
passages continue with the stanza : 

devo 2 yah samsthitas (v. 1. sarhqritas) tasminn, ab- 
bindur iva puskare 

ksetrajnam tam vijanlhi (v. 1. °iyat) nityaih yoga- 

“ Know that the divine being who stands in the body like a 
drop of water on a lotus, is the spirit eternal but overcome 
by its association.” The epic texts vary in the next stanza, 
but the sense is the same, to the effect that the individual 
life-spirit, jrva, though conditioned by the three gunas, has the 
characteristics (gunas) of the atman, while atman again is one 
with the Supreme Atman (paratmakam, 213, 21). The third 
version of the passage, found in xii, 187, 23-25, explains the 
individual spirit, ksetrajna, as atman conditioned by the gu- 
nas of Prakrti, and as Supreme Atman when freed from 

15, where the phrase above reappears in a copy of this section. In xii, 329, 
31 ff. (and elsewhere) the pranas are seven personified creatures, Udana born 
of Samana, etc., as winds, pra, a, ud, sam, vi, pari, and para (vahas). Com- 
pare also xii, 184, 24, below. 

1 ekadagavikaratma kalasambharasambhrtah murtimantam hi tam viddhi 
nityam yogajitatmakam, tasmin yah samsthito hy agnir nityam stbalyam 
iva’hitah atmanam tam vijanlhi nityam yogajitatmakam, 213, 18-19. 

2 In xii, 246, 29, deva may be jiva, devam tridhatuiii trivrtam suparnarii 
ye vidyur agryam paramatmatarii ca, but on the other hand there may be a 
textual error here of devo for dehe. Compare xii, 187, 24, tasmin yah samgrito 
dehe hy abbindur iva puskare. The Supreme Spirit is devo (nirgunah), xii, 
341, 101, as in <?'’et. Up. i, 8 (here called, 99, vajnesv agraharah). 



them; 1 with a varied reading of nityaih lokahitatmakam and 
viddhi jivagunan in the following verses ; 26, however, being 
almost the same as iii, 213, 22 : — 

sacetanam jlvagunaiii vadanti 
sa ce state cestayate ca sarvam 
(t)atah paraiii ksetravido vadanti 

prakalpayad (v. 1. pravartayad) yo bhuvanani 

“ They say that the individual spirit is characterized by intel- 
ligence; it moves and causes all to move. 2 The wise say, 
that he who caused the many creations to form is still 
higher (or the Highest).” 

The reading in xii, 187, 23 brings the passage into still 
closer connection with the Upanishad. The latter, at iii, 2, 
has atma bindur iva puskare followed by sa va eso 'bhibhutah 
prdkrtair gunaih, while the epic has abbindur iva puskare 
preceded by atma ksetrajna ity uktah samguktah prdkrtair 
gunaih, where the Yana version keeps (what is here lost) 
the image of the fire in the pot. Then the stanza above, 
sacetanam, etc., 3 closely reproduces the words as well as the 
thought of the Upanishad, ii, 5: cetanene ’dam ganram ceta- 
navat pratisthapitam pracodayita vai ’so 'py asya (compare 
acetanam 9ariram, ii, 3). The fact that the epic Vana is 
not based on the lotus-phrase of earlier Upanishads but is 
following the Maitri is shown even more clearly in the phra- 
seology of the following stanza, 213, 23, which at this point 
does not correspond to (Snti above, but to a later chapter, 

1 For the text, see the end of the last note. A passage in xii, 316, 15-17 
combines freely the two traits mentioned above: "The fire is different from 

the pot, uklia; the lotus is different from the water, nor is it soiled by touch 
of water,” etc. — a fact which is said to be “not understood by common 
people,” as in the example below. 

3 The commentator says that as individual soul the atman is active, and 
as the Lord-soul causes activity (compare xii, 47, 65, yag cestayati bhutani 
tasmai vayvatmane namah); but the Highest is above both these. In xii, 
242, 20, jivayate takes the place of cestayate. 

8 C. has acetanam in the Vana passage, but both texts in both the fanti 
passages have sacetanam, xii, 187, 26 ; 242, 20. 


xii, 247, 5. The Yana passage says: “Thus in all beings 
appeals the bhutiitman (conditioned spirit), but it is seen 
only by the subtile intellect ; ” whereas the (Snti passage 
has not bhutatma samprakayate, but gudho 'tma na praka§ate, 
“ concealed it is not apparent,” that is, it has the text of the 
Kathaka. 1 But in Vana there is the characteristic bhutatman 
of the Upanishad, which says at iii, 3: “(Pure) spirit is no 
more overcome (by environment) than fire is overcome -when 
the mass of iron (enclosing it) is hammered ; what is over- 
come is the bhutdtman, which is abhibhuta, overcome, because 
it is bound up with (the transformations) ; ” and further, 
iii, 5 : “ Filled with the effect of the gunas (which condition 
it) the bhutiitman is abhibhuta (the same etymological tie), 
overcome, by them, and so enters different forms.” 2 A few 
more passages contain this word bhutiitman. Of these, two 

1 See the analysis above, p. 30, note 2. 

2 The etymological connection between abhibhuta and bhutatman may 
have suggested to the commentator his explanation of bhutiitman as an 
epithet of mahatman in xiii, 31, 15, where he says that mahatmans are called 
bhutatmans because they have overcome or controlled their thoughts (bliuta 
— vajikrta). In the epic, bhutatman appears as incorporate spirit in xii, 201, 
1, where “ how can I understand bhutatman 1 ” is to be thus interpreted ; and 
as intellect, buddhi, in the reabsorption process described at xii, 313, 12, mano 
grasati bhutatma. Differently employed, the combination appears in Gita, 5, 
7, where one is said not to be contaminated by action if one is sarvabhuta- 
tmabhutatma, which, as is shown by parallel passages, is not to be divided 
into sarvabhutatma and bhutatma, but into sarvabliuta, atmabhuta, atma, 
where sarvabhutiitmabhuta means one with all, or the All-soul. Compare xii, 
240, 23, sarvabhutatmabhutasya vibhor bhutahitasya ca deva ’pi marge mu- 
hyanti; xii, 47, 82, sarvabhutatmabbutaya . . . namah. Bhutatman means 
also elemental spirit, as in xii, 298, 17-19, where it is said that before the 
disembodied jiva, or spirit, secures a new resting place (ayatana, body), it 
wanders about as a bhutatman, “like a great cloud.” So in xii, 254, 7, the 
bhutatman of Yogins wanders through space and has seven subtile gunas 
(according to the commentator, the fine elements, intellect and egoism), like 
sattvatman, ib. 6 ; but here, too, it is the bhutatman, “ standing in the heart,” 
ib. 12. I observe, by the way, that the citation above, “ the gods are con- 
founded at the track of him who is identical with all created things ” (com- 
pare the anirdefva gatih, “indescribable course, which the moksmah foresee,” 
xii, 19, 15), shows, as does xiii, 113, 7, apadasya padaisinah, that in xiii, 
141, 88, padarii tasya ca vidyate should be changed to na vidyate, as in C. 
6477 (sattvarh sarvabhutatmabhutastham is found in xii, 210, ot>). Compare 
Dhammapada 420, yassa gatim na jananti deva. 



or three deserve particular attention. In xii, 240, 21, it 
appears in a stanza like one to be cited presently, where 
another Maitri word is found, but here the text says merely 
that the bhutatman (ceases to be conditioned and) enters 
Brahman, where it “ sees self in all beings and all beings in 
self.” In §1. 11 of the same chapter the bhutatman appears 
as the controller of mind in the same simile of the wild horses 
noticed above, but with a different turn : “ Mind, as a char- 
ioteer his horses, directs the senses ; and the bhutatman 
which is seated in the breast directs mind; as the mind, 
restraining and letting out the senses, is their lord, so the 
bhutatman in respect to the mind.” In xiv, 51, 1, on the 
other hand, the mind itself is called bhutatman, because it 
rules the mahabhutas. Finally the same term is used of 
Vishnu in xiii, 149, 140, where it is said : 1 

eko Visnur mahad bkutam prthag bhutany anekagah 

trln lokan vyapya bhutatma bhunkte viqvabhug 

“ Vishnu as one is a great spirit (bhuta), and separately is all 
beings; he, permeating, enjoys the three worlds as bhutatman, 
he the all-enjoyer, indestructible.” 

It is clear from these passages that bhutatman is not used 
in one strict sense in the epic, but its signification varies 
according to different passages. In one case it is a free spirit 
of elements, 2 but in another the conditioned spirit in the 

1 The quotation here given may be the one cited in PW. from t'KTir. g. 
bhutatman I, 1. But compare also xii, 207, 8, where the Lord Govinda is 
bhutatma mahatma. In the “Secret of the Vedantas" (ITpanishads) the 
Intelligence as Lord bhutakrt, maker of elements, is called Bhutatman, xii, 
194, 7 = 248, 4, and 14 as Buddhi. 

2 Hence called suksma, fine. This seems to be the sense in xii, 203, 0-7 : 
“As no one has seen the back of Himalaya or of the moon, but cannot say 
it is non-existent, so the fine bhutatman which in creatures has a knowl- 
edge-soul, jnanatmavan, cannot be said not to exist because it has not been 
seen.” With this jnanatman compare, by the way, what is said of the soul, 
ib. 240, 22, yavan atmani vedatma tavan atma paratmani (just after the vc-rse 
cited in the text 240, 21, above, on bhutatman) : “ The soul is as much in the 
All-soul as there is knowledge-soul in itself.” 


body. 1 It is the latter meaning which applies both in the 
Upanishad and in the epic imitation of it. In these cases 
bhutatman is the atman, spirit, not as being pure Purusa, 
but as being in connection with and conditioned by bhuta, 
that is, imprisoned in matter. It is apparently a popular (not 
philosophical) term for spirit in general, and when used in 
philosophy answers to the ordinary philosophical jlva, incor- 
porate spirit. It is not found in other (old) Upanishads. 

But there is still a closer parallel between the epic and the 
Upanishad. After the verse cited above, it is said, iii, 213, 
24-27, that salvation is attained by peace of mind and by per- 
ceiving self in self, and that this purified spirit by the aid 
of the lighted lamp (of knowledge), seeing self as free of self, 
becomes released. 2 Here again we have a peculiarly Maitri 
word in nirdtman , “free of self,” that is, free from the de- 
lusion of subjectivity. But the two works are here evidently 
identical. First, just as the epic says that one must have 
peace of mind, prasiida, and be pure, and then becomes nirat- 
man, so in ii, 2-4, the Upanishad, after an allusion to sam- 
prasada, the same peace of mind, says that one becomes pure 
and nirdtman (guddhah putah gunyah ganto 'prano niratma). 
The sign of this peace is explained as when one sleeps sweetly, 
iii, 213, 25 = xii, 247, ll. 3 In the epic the word niratman 
occurs again in much the same way, xii, 199, 123, gantibhuto 
niratmavan, like the collocation above in the Upanishad. 

1 Compare what is said, Mait. Up. iii, 2. “The bhutatman is affected by- 
ignorance, and so gives itself up to objects of sense,” it is said in xii, ‘104, 5. 

2 “ For self is the friend of self, and even so self is the foe of self,” V, 34, 
64; Gita, 0,5. 

3 Samprasada is susupti, unconscious slumber. Unconscious existence is 
the goal of the soul, for the conditioned spirit, jiva, “glorious, immortal, an- 
cient ” is a part of this unconsciousness, and on becoming pure enters it. In 
a preceding section this samprasada, or unconscious existence, is declared to 
be the body of the universe : Yah samprasado (am, C.) jagatah cariram, sarvan 
sa lokan adhigacchati ’ha, tasmin hitam (hi sam, C.) tarpayati ’ha devans, 
te vai trptas tarpayanty asyam asya, xii, 246, 33, where the sense seems to 
be that the reabsorption of the universe pleases the mouth of unconscious- 
ness; that is, the mouth of Time as Lord of all, a metaphor from the pre- 
ceding verses. So samprasada is a spirit at peace, in Cliand. Up., cited 
on the next page. 



Another passage reads : “ The spirit (atman, but conditioned) 
knows not whither it goes or whence, but the inner-spirit, 
antaratman, is different ; it sees all things ; with the lighted 
lamp of knowledge 1 it sees self in self. Do thou, too, seeing 
self in (or with) self, become freed from self, become all-wise ” 
(niratma bhava sarvavit, xii, 251, 9-10). This verse, is in fact, 
only a different version of the “ lighted lamp ” verse above. 
This latter, in turn with its environment, must be compared 
in the original with the Upanishad to see how close are the 
two. But for this purpose I take, not the samprasada passage 
referred to above, which is parallel to Chand. Up. viii, 3, 4, 
but one from the sixth book, where the Upanishad, vi, 20, 

tada ’tmanci ’tmanam drstvd niratma, bkavati, 

whereupon follows a stanza cited, ity evarii hy aha, as : 
cittasva hi prasadena hanti karma Qubhacubham 
prasannatma ’tmani sthitva sukham avyayam acnute 

In the epic, iii, 213, 24, this whole stanza (gloka) appears, 
cittasya hi prasadena, etc., in exactly the same words , 2 and 
then, after the definition of prasada and the injunction that 
one must be viguddhatmii, of purified soul, as explained above, 
come the words, gl. 27, drstvd ’ tmanam nirdtmdnam sa tada 

When this stanza is repeated in the Upanishad at vi, 34, it 
is preceded by the verse yaccittas tanmayo bhavati, so that 
together we have : 

yaccittas tanmayo bhavati guhyam etat sanatanam 

(i. e., the guhyam of Dhammapada 1, mano settha manomaya; 
compare Pracna Up. iii, 10, yaccittas tenai ’sa pranam ayati) 

1 Here jnanadipena (compare Gita, 10, 11) diptena ; above, pradlptene ’va 
dipena manodlpena. Compare dlpavad yah sthito hrdi, Maitri, vi. 30 (and 

2 In the corresponding filnti chapter, in which I pointed out above the 
simile of the six senses as horses, and gudho ‘tma for bhiitatma, this verse 
is found in a different form, cittaprasadena yatir jahati ’ha (,'ubhai;ubbam, 
vii, 247, 10. 


eittasya hi prasadena hanti karma cubhagubham 
which the Anuglta takes up xiv, 51, 27, and 36, in inverse 
order : 

27, yaccittam tanmayo 'vagyam, guhyam etat sana- 

36, prasade cai Va sattvasya prasadam samava- 

If all these points be compared, first the general order of 
discussion, then the peculiar words which are used in the 
same way in both texts, and finally the identical passage just 
given, it is clear that one of these texts must have followed 
the other. The dispersion of the epic chapter over different 
books certainly makes it seem more likely that it is a copy 
than an original. This opinion is strengthened by the late 
features added in the epic, the freedom in metre, almost 
exclusively characteristic of the later epic, and the late Ve- 
danta grouping of seventeen at the beginning. For this 
group is not the old Samkhya group, which occurs often 
enough elsewhere in the epic, but a modification of it as in 
the Vedantasara. 

The citation in the Maitrayana of the stanza eittasya hi 
prasadena from some source might be referred to the epic, 
but it seems more likely that this, like a dozen other “ some 
one says” verses in the same Upanishad, is a general refer- 
ence, and it is quite counterbalanced by the fact that the 
Vana version in the epic adds a hidden reference to its 
source in the words maitrai/ana-gatag caret, a strange expres- 
sion, which is found only in this verse and in its repetition in 
the twelfth book ; 1 while the speaker in the last verse of the 
Vana chapter confesses that what he has been teaching “is 
all a condensed account of what he has heard.” 2 

1 iii, 213, 34; xii, 279, 5; with a slight varied reading in xii, 189, 13. 

2 yatha cjrutam idam sarvam samasena . . . etat te sarvam akhyatam, iii, 
213, 40. I suppose no one will lay any weight on the statement of xii, 247, 
which copies Vana here (see above), that (12-14) this is a “secret not handed 
down by tradition,” anaitihyam anagamam (atmapratyayikam pastram), but 
an ambrosia “ churned from dharmakhyanas, satyakhyana, and the ten 



It is perhaps worth noting further that in the Upanishad 
vi, 20-21, one sees the real soul and becomes isolated (where 
the goal is kevalatva), whereas in iii, 211, 15 of the epic, the 
result of this same seeing of self truly is brahmanah sariiyo- 
gah, union with Brahman; which carries on the antithesis 
already noticed between the Samkhya tanmatras of the 
Upanishad and the omission of the same in the epic. This 
special designation of tanmatra in iii, 2 is complemented by 
the viyesas mentioned in vi, 10, and is important as showing 
that the Upanishad, as a Upanishad, is late, for none of the 
older Upanishads has either of these terms. Its priority to 
the epic, however, may be urged on still another ground 
than those mentioned above. The Upanishad quotes stanzas 
freely, and it is scarcely possible that if the epic and Manavic 
verse cited above on p. 27 had existed in verse the prose form 
of the Upanishad would have been used. As Miiller says in 
his note on the Upanishad passage : “ Part of this passage 
has been before the mind of the author ” (of Manu together 
with the epic poet). So perhaps, too, with the recognition 
of the eleven (vikaras) in v, 2. The epic has both groups, 
eleven vikaras and also the system’s sixteen, as I shall show 
in a later chapter. As compared with the epic, moreover, 
the Upanishad is distinctly earlier in knowing Yoga as “six- 
fold,” vi, 18, whereas the epic makes it “eightfold,” xii, 817, 
7 ff. as does Patanjali, ii, 29. 

I think another circumstance may point to the fact that 
the epic refers directly to the sixth chapter of the Upanishad. 
The word tatstha is not, indeed, used in a pregnant sense in 
the Upanishad. It is simply an ordinary grammatical com- 
plex in the sentence vi, 10, purusay ceta pradhanantahsthah, 
sa eva bhokta . . . bhojya prakrtis, tatstho bhunkte, “ Prakrti 
is food; when standing in it (Prakrti), the Purusa enjoys.” 
But in the epic, xii, 315, 11, we read sa esa (purusah) pra- 
krtistho hi tatstha ity abhidhiyate, “ Purusha is designated as 
tatstha when he is in Prakrti.” As the expression tatstha 

thousand Rks,” for this applies only to paipyaty atmanam atmani, seeing 
self in self, not to the exposition. 


occurs only in this Upanishad, according to Col. Jacob’s Con- 
cordance, it seems very likely that the epic verse alludes to 
the tatstha = prakrtistha of the Upanishad, where Purusa is 
expressly purusa g cetii, and the epic also follows, 14, with 
cetanavans tatha cai ’kah ksetrajna iti bhasitah. 1 

In Up. vi. 15 and Mbh. xi, 2, 24 occurs Kalah pacati bhu- 
tani ; and in the companion-piece to the image of the body as 
a house, cited above from Up. iii, 4, as the same with xii, 330, 
42, namely, Up. i, 3, occurs anistasamprayoga = Mbh. xi, 2, 
28, but I do not think that these universal expressions taken 
by themselves are of any significance. 

On the other hand I cannot regard as unimportant the fol- 
lowing stanzas, beginning with the extraordinary, unsyntac- 
tical, verse found in the epic, xii, 241, 32, — 

sanmasan nityayuktasya gabdabrahma ’tivartate 
compared with 237, 8 (Gita 6, 44, jijnasur api yogasya, etc.), 
api ‘pi gabdabrahma ’tivartate 
and with xiv, 19, 66, 

sanmasan nityayuktasya yogah, Partha, pravartate 
and with Miiitr. Up. vi, 28, 

sadbhir masais tu yuktasya nityayuktasya dehinah 
anantah paramo guhyah samyag yogah pravartate 

and with Miiitr. Up. vi, 22 = IMbh. xii, 233, 30, 

dve brahmanl veditavye gabdabrahma paraih ca yat 
gabdabrahmani nisnatah param brahma ’dbigacchati. 

The last stanza occurs only here and in this Upanishad (ex- 
cepting later copies). 2 The first is a meaningless compound of 

1 It may be noticed here also that in caitanya the vocabulary of the pseudo- 
epic is that of the Upanishad in its later part, vi, 10 and 38 (the word is found 
else only in late Upanishads). Compare : acaitanyam na vidyate (the tree has 
a jiva), xii, 184, 17; cetanavatsu eaitanyarii samam bhutesu pafyati, “the sage 
sees one and the same soul in all conscious creatures,” xiv, 18, o3. The term 
is unknown to the Gita and early epic. 

3 With the var. lec., dve vidye veditavye, 3Iund. Up. i, 4 ; Brahmabindu 
Up. i, 17 Compare a sort of parody in xii, 100, 5, ubhe prajne veditavye 
rjvi vakra ca, Bharata. The dve vava bfahmano rupe of BAU. ii, 3, 1, are 
perhaps the first pair, though there it is higher and lower Brahman in a meta- 
physical sense. 



the “six months” stanza and the “two brahman” stanza. The 
second is a theoretical advance on the latter, which says that 
when one is thoroughly conversant with the word-brahman he 
gets to the highest Brahman. The later Yogin does not think 
this necessary, and emends to “ even one desirous of knowl- 
edge (of Yoga, in Gita) surpasses the word-brahman,” while 
the “ six months ” stanza in the epic is adjusted to the occa- 
sion (nityayuktasya of the MSS. is to be read in the Upani- 
shad as in the epic). Here again, the Maitrayana alone has 
this stanza, nor does nityayukta occur elsewhere except in 
the same way in the Gita, 8, 14, nityayuktasya yoginah. 

In my opinion these parallels together with the cittasya hi 
prasadena stanza above indicate that the epic has copied from 
the sixth chapter of the Upanishad as well as from the earlier 
portions. 1 

The Vedic period, then, is represented in the epic down to a 
pretty late stage of Upanishads. The tanmatra era of philos- 
ophy, the trinitarian era of philosophy, these are represented 
by the Upanishad and by the epic ; but only the latest philo- 
sophical and religious chapters of the epic recognize tanma- 
tras (the name) and the trinity, as only the later Upanishads 
recognize them. 

Of still later Upanishads, it is possible that the pseudo-epic 
may know 

The Atharvaciras Upanishad. 

The title is applied to Narayana, xii, 339, 113, and the 
commentator explains it as referring to the Upanishad. 2 But 
we must, I think, rest content with the certainty that the 
epic cites (a) the Brhadaranyaka Up., (b) the Kathaka, (c) the 

1 The general lateness of the Upanishad is shown by its recognition, v, 2, 
of the trinity (Muir ap. Holtzmann), which is also recognized in the later 

1 On this and on i, 70, 39-40 in the £ a kuntala episode, bharundasamagita- 
bhir atharvaciraso ’dgataih . . . atharvavedapravarah, compare Weber, IS., 
vol. i, pp. 383-4. See also above, pp. 8 and 9 (note 1). 


Maitrayana, or, in other words, copies at least one of each of 
the three kinds of Upanishads, old prose, metrical, and later 

Acvalayana Grhya Sutra. 

In this Sutra i, 15, 9, occurs a stanza which is found also 
with varied readings in the Kausltaki and BA. Upanishads 
(ii, 11 ; vi, 4, 9, respectively) as a single stanza. This is cited 
in the epic as Vedic, the reading following that of the Sutra 
and adding one stanza, which clearly belongs to the citation, 
i, 74, 68-64: 

vedesv api vadantl ’mam mantragramaih dvijatayak 
jatakarmani putranam tava ’pi viditarii tatha 
afigad angat sambhavasi hrdayad adhi jayase 
atma vai putranama ’si sa jlva qaradah qatam 
jlvitaiii tvadadhlnarn me santanam api ca ’ksayam 
tasmat tvam jiva me putra susukhl qaradaiii qatam 1 

The general conclusion to be drawn from these citations is 
twofold. First, the epic, synthetically considered, post-dates 
the latest Yedic works. Second, the final redactors were 
priests, well acquainted with Yedic literature. Of these 
points there can be no doubt; nor is a third open to serious 
objection, namely, that the restriction of philosophical citation 
to philosophical chapters does not prove anything in regard 
to the date of the epic that preceded the insertion of these 

Puranas and Itihasas. 

Whether the Puranas, ascribed to Romaliarsa (sic) in xii, 
319, 21, precede or follow epic literature, is not a question 
that can be answered categorically. Nothing is commoner 
than the statement made by some epic character that a story 
was heard by him long ago in a Purana. 2 But most of the 

1 Acvalayana is mentioned only in the pseudo-epic, xiii, 4, 54. On this 
and his mention of the epic, see below, and Holtzmann, loc. cit., p. 27, with 
other supposed references to Sutras. 

2 For example, xiii, 84, 59, maya frutam idam pur v am purane. For the 
relation between the extant Puranas and the epic, compare Holtzmann, loc. cit., 
p. 29 ff. There is no earlier allusion to an extant Purana (SBE. ii. p. xxyiii) 



extant Puranas are in their present shape certainly later than 
the epic. Nevertheless, before the great epic was completed 
the eighteen Puranas were known, since they are mentioned 
as a group xviii, 5, 46 (not in C.) and 6, 97. Further, a Vayu 
Purana is referred to in iii, 191, 16 : 

etat te sarvam akhyatam atltanagatam tatha 
Vayuproktam anusmrtya Puranam rsisamstutam. 

This statement, however, implying that the Purana treats 
of future events, though illustrated in this instance by the 
epic’s account of later ages, scarcely tallies with the early 
epic use of the word, which regularly connotes atlta, the 
past, but not anagata, (account of) things to be ; yet it corre- 
sponds exactly to the ordinary contents of the later Puranas. 
On the other hand, the pseudo-epic contains this later sort of 
Purana, known as Purana as well as akhyana and mahopa- 
nisada, where future events are described. 1 It is to be re- 
marked, moreover, that this reminiscence of Vayu’s Purana, 
a work which is referred to again in the Harivanija, is con- 
tained in the Markandeya episode, which long interpolation 
is itself virtually a Purana. That some of the verses in the 
extant Vayu are like some in the epic proves nothing in 
regard to the relative age of either. 2 There is no real iden- 

tlian that in Ap. Dh. S., ii, 9, 24, 6, where a Bhavisyat Purana is cited, the words 
haring an epic strain, perhaps to be filled out with vijarthah srarge (jivanti 
yavad) abhutasamplavat. See also above, p. 6. On the Puranas as deposi- 
tories of Yedic <?ruti, see the quotation above, p. 4, and compare H. 3, 33, 5, 
etat te kathayisyami puranam brahmasammitam naniigrutisamayuktam. 

1 xii, 340, 95-125, future avatars, conquest of Kalayavana, etc., called 
mahopanisadam (sic, neuter), in fl. Ill, puranam in 118 and 124, akhyanam in 
125. Closely united are “praise and Puranas” (known to Sutas) in xii, 53, 3 
(not like the stuti?astra, praise-treatises, of the late passage, ii, 452, where, 
however, B. 11, 35, has stuticastrdni). 

2 Even the Garuda and Varaha Puranas may precede the final revision of 
the whole epic, though the evidence for references is far from conclusive ; 
but on the other hand our present Puranas may have been so changed as 
not to agree in any detail with Puranas that once bore these names. The 
arguments are given by Holtzmann, loc. cit. The epic passages supposed to 
refer to the Puranas are H., 3, 33, 5 (above) and i, 31, 3. The epic declaration 
i, 2, 386, that it is the base of all Puranas, presupposes a goodly number 
already in existence ; but this statement is as late an addition to the poem 


tity in the account cited from the Vayu Purana and the 
extant Vayu Purana. In the description of the Kali age, 
for instance, where the epic (in the part said to be from the 
Vayu Purana) has, 190, 64, ^udra dharmam pravaksyanti, 
brahmanah paryupasakfih, the Vayu, lviii, 41, says Cudriicar- 
yac ca brahmanah, and where the epic, ib. 97, has utsadayi- 
syati inlecchaganan, the Vayu, ib. 78, has mlecchan hanti, 
but here there is nothing characteristic. On the other hand, 
the most striking features in the epic account, the edukas, 
and Kalki, with the heavy taxes laid upon priests, §1. 62, 
65-67, 93 ff., are not found in the Vayu at all. Noticeable 
also is the fact that the epic account not only has more 
than the Vayu, but has contradictory statements. Thus in 
§1. 58, the Vayu declares one of the signs of the evil age to 
be that girls less than sixteen will bear children ; while in the 
epic the sign is that girls of five or six will bear and boys of 
seven or eight will beget children : pancame va ’tha saste va 
varse kanya prasuyate, saptavarsa ’stavarsag ca prajasyanti 
naras tada, 190, 49. Taken altogether, the epic account 
seems to be an extended and exaggerated reproduction of 
that in the Vayu Purana, but it is impossible to say whether 
it is really based on the extant text or not. The Puranic 
version, however, does not seem to be taken from the epic 
account, and as the latter is expressly said to be from the 
Purana it is reasonable to suppose that the Miirkandeya 
episode was inserted into the epic after the Vayu Purana 
was written, though this must remain only a supposition. 

Another long intrusion in the same third book of the epic, 
this time in the Tlrtha stories, iii, 110 ff., leads to a result 
somewhat more definite in respect of the relation between 
the particular story intruded into the epic and the Padma 

as is the mention of the eighteen. I suppose most scholars will accept the 
u eighteen Puranas ” as actually referring to eighteen, and I am inclined to 
do so myself. At the same time the number is more or less conventional in 
the epic (see the groups of eighteen spoken of below), and even in the period 
of the Upanishads literary works may have been grouped in eighteens : yajha- 
rupa astada£o£tam avaram yesu karma, with Deussen’s remark on ukta and 
attempt to explain the number, Mund. Up. i, 2, 7. 




Purana. Here, according to the acute investigation of Dr. 
Liiders, Die Sage von Rsyagrnga, the epic account in its 
present form is based upon that of the Purana. Dr. Liiders 
thinks indeed, p. 103, that there was an earlier epic form of 
the story which antedated the Puranic account. But it is at 
least certain that the present epic form is subsequent to the 
present Puranic form, and that the tale is drawn from popu- 
lar sources that antedate in all probability all the literary 
versions in Sanskrit. 

Leaving the modern Purana, as it is described, e. g., in 
Vayu Purana, iv, 10, 

sargaq ca pratisargaq ca vanqo manvantarani ca 

vaiicanucaritam ce ’ti puranam pancalaksanam, 

and turning to the meaning of the word in the epic, there 
is no essential difference between atita, akhyana, 1 purana 
and itihasa. Together with the more general katlm, all these 
words mean ordinarily an old tale, story, legend or incident. 
Rarely is Purana itself used of cosmogony, but a case occurs 
in xii, 201, 6, where the phrase tad ucyatam puranam refers 
to the origin of earth, heaven, creatures, wind, sky, water, 
etc. The birth of Asuras and Suras is a Puranic topic in i, 
65, 38. When not an adjective to akhyana, which is a com- 
mon function of the word, it is an equivalent substantive. 
Thus the Nandinl tale is an akhyanam puranam, i, 175, 2, 
while in xii, 343, 2, hanta te vartayisyami puranam, the word 
in the phrase takes the place of Itihasa; as it does in i, 196, 
14, gruyate hi purane 'pi Jatila nama Gautaml. 

From remote antiquity these Puranas or tales of old were 
associated with Itihasas, legends, whether cosmological or 
not (the distinction is quite artificial). They were narrations, 
kathiis, composed partly in prose and partly in verse, gathas. 
Katha itself is entirely non-specific, and may be a causerie 
rather than a tale, as in ix, 38, 16, where are mentioned reli- 

1 Synonymous with this is the word upakhyana. Thus the fakuntala 
episode and Namuci myth, ix, 43, 33, bear the name upakhyana, and in v, 
18, 16, and 19 it is synonymous with akhyana. The Fowler’s tale is a dharma- 
khyana, iii, 216, 36 (compare a reference to many such, p. 5, above). 


gious conversations , 1 citrah katha vedam prati. A legend, 
such as that of Agastya, is a katha divya, iii, 100, 2. The 
inahopanisadam alluded to above is a kathamrtam, the essence, 
sara, of hundreds of upakhyanas, xii, 340, 127. So the (jvo- 
tadvlpa story is a kathasara, xii, 336, 16. 

But the especial characteristic of the old legend is that it 
relates the story of great kings or gods 2 and their acts in the 
past. In iii, 298, 7, Dyumatsena is solaced “ by the help of 
tales of former kings,” citrarthaih purvarajHaih kathmjrayaih, 
according to the recommendation in the epic itself : “ Comfort 
those afflicted in mind with tales of the past,” yasya buddliih 
parihhavet tarn atitena santvayet, i, 140, 74; an instance 
being the story of Nala, klrtana, itihasa, itihasah puranah, 
as it is indifferently called, iii, 79, 10, 11, 13, 16. 

The word itihasa may also have the meaning “saying,” 
rather than “ legend.” Thus in iii, 30, 21 : 

atra ’py udaharantl ’mam itihasam puratanam 
I^varasya vaqe lokas tisthante na ’tmano yatha, 

where itihasa is equivalent to pravada, a proverbial saying 
(in this instance repeated in 9 I. 25 and in other parts of the 
epic). But ordinarily the word means a tale, of which the 
hemistich just cited is the stereotyped introduction, as in iii, 
28, 1 and passim . 3 It is important to notice that, as itihasa is 
used for proverb and glta gatha is also used in the same way, 

1 So a philosophical discourse of religious content, moksadharma, is an 
Itihasa, xii, 334, 42; and the tale of a good Brahman is a katha on duty, 
xii, 354 ff. 

2 The tale of Atharvan finding Agni when the latter disappeared is an Iti- 
hasa puratana, iii, 217 and 222. In iii, 183, 40, puravrttah katliah punyah, 
are “tales of kings, women, and seers.” With puravrtta as adj. compare 
kathayanti puravrttam itihasam, xii, 18, 2 ; as a noun it is not uncommon, 
rajnam puravrttam, “ a tale of kings,” etc., as is illustrated sufficiently in PW. 
(compare rrttanta). Khandava’s burning is a paurani katha rsisamstuta, i, 
223, 16. “ Men, snakes, and demons ” is the subject of a “ divine tale,” katha 
divya, in iii, 201, 4. 

3 A word of analogous formation is aitihya, equivalent to traditional re- 
port, Veda. It is found, e. g., in xii, 218, 27 and 247, 13, and G. v, 87, 23, as 
one of a group of sources of knowledge besides anumana and pratyaksa. 
Compare itivrtta, as legend, in i, 1, 10. 



for example, the na jatu kamah proverb, i, 75, 49-50, so the 
phrase to introduce a tale, Itihasa, may substitute gathas, as 
in iii, 29, 35, atra ’py udaharanti ’ma gathah . . . gltah. 
Such gathas refer to action or to ethical teaching (compare 
the same formula for both, loe. cit. and ii, 68, 65). A differ- 
ence may be imagined in the element of song of the gatha, 
but this is illusory. The gathas are indeed said to be sung, 
as in the case just cited (§1. 34-44 are the gita gathah), but 
singing is too precise a translation. As shown above, even 
the Aranyakas are “ sung,” and in point of fact the gathas 
are synonymous with glokas and are recited. Stanzas of 
Puranas are thus said to be sung. 1 Conversely, gathas are 
not always sung, iii, 135, 45, atra ’py udaharanti ’ma gatha 
devair udahrtah ; while ib. 54 is another illustration of the 
word gatha meaning only a current proverbial §loka. But in 
this case it is woven together with the legend of Dhanusaksa, 
whose direct curse not succeeding in slaying his enemy, he 
destroyed the mountain, in the life of which was bound up 
the life of the invulnerable foe. Hence they say “ man can 
never escape his fate:” 

ucur vedavidah sarve gatham yam tarn nibodha me 
na distam artham atyetam iqo martyah 2 kathamcana 
mahisair bliedayamasa Dhanusak^o mahldharan 

Such gathas 3 are even incorporated into the law-books : 
“Verses recited by Yama” are cited (by those that know 
antiquity and the law) “ in the law-books ” on the sin of 
selling a son or daughter, xiii, 45, 17. 4 

1 Compare Tirtha gatha and Tirtha gloka, iii, 88, 22 ; 89, 17 ; 90, 6 ; “ the 
gloka sung in a Purana,” purane gruyate gltah glokah, v, 178, 47 ; puranah 
yloko gltah, iii, 300, 33 (a proverb on fame) ; Holtzmann, loc. eit., p. 29 ff. 

2 The reading amartyah in B. would require api. C. has martyah. The 
proverb appears in a different form, v, 40, 32, na distam abhyatikrantum 
jakyam bhutena kenacit. 

8 In the Ramayana also, eti jivantam anando naram varsagatad api is given 
as a kalyani or pauranl gatha laukikl, v, 34, 6 ; vi, 126, 2 (G. 110, 2). 

4 atra gatha Yamodgitah kirtayanti puravidah dharmajna dharmajastresu 
nibaddha dharmasetusu, yo manusyah svakam putram vikrlya dhanam icchati 
kanyam va jlvitarthaya yah yulkena prayaccliati, saptavare, etc. 


The best known example of the last case, gathas recited 
by a divinity, is found in the Harigltas (plural), xii, 347, 
11, that is the Bhagavad Gita (Upanishad). 1 Here the “sing- 
ing” is that of the Aranyakas. As Vedantas are Upanishads 
(above, p. 9), so we find in xii, 247, 21, yat tan maharsi- 
bhir drstam (= Veda), vedantesu ca glyate, “what is re- 
vealed in the Veda and sung in the Upanishads.” 

Such tales and legends are said to be the epic itself, which 
is called indifferently an Itihasa, a Purana, or Krsna’s Veda. 2 
As the Chandogya Upanishad applies the title “fifth Veda” 
to the Itihasapurana, so the epic claims the same title : 

itihasapuranah pancamo vedanam, Chand. Up., vii, 1, 2, 4 
(So each is a Veda in Qat. Br. xiii, 4, 3, 12-13.) 
adhltya caturo vedan sangan akhyanapancaman, vii, 9, 29 
sangopanisadan 3 vedaiig catur akhyanapancaman, iii, 45, 8 
vedan adhyapayamasa Mahabharatapaneaman, i, 63, 89 and 
xii, 341, 21. 4 

In the opening stanzas 5 of the great epic it is described as 
a Samhita, collection, a grantha, book, a Purana, an akhyana, 
an Itihasa, a Kavya, a poem containing various Castras, full 
of Vyakhyas (vaiyakhya) or narrations, and Upanishads. It 
is true that it is also called a Dharmagastra, yet this repre- 
sents but one side of its encyclopaedic nature, as it is besides 
Arthagastra, Dharmagastra, and Kamagastra, i, 2, 383. When 
the character of the work as a whole is described, it is in 

1 bhagavadakhyanam, ib. 2 ; here a recitation about the Lord, not by the 
Lord. But the Gita is a recitation by the Lord, gita bhagavata svayam, ib. 
349, 8. 

2 i, 62, 16-18, idam puranam . . . itihasam . . karsnam vedam vidvan. 

So the imitation of the Gita in the twelfth book is called “ Krsna’s Religion," 
Satvato dharmah (see below). 

3 The other form occurs, e. g., iii, 206, 2, sangopanisado vedan adhite. 

4 Compare also v, 43, 41 ; ix, 6, 14 (as above), and vedaiig ca ’dhijage sangan 
setihasan, i, 60, 3; itahasapuranesu nanayiksasu bodhitah vedavedangatat- 
tvajnah, i, 109, 20; vedesu sapuranesu rgvede saya j urvedc . . . purane so- 
panisade tathai ’va jyotise ayurvede tathai ’va ca, xii, 342, 6-9; ye 'dhlyate 
setihasam puranam, xiii, 102, 21 ; yad etad ucyate yastre setihase ci chandasi, 
xiii, 111, 42. 

5 i, 1, 16, 49, 55, 61, 72. 



terms of epic story, not of didactic code. Even the Hari- 
vahqa poet does not fail to distinguish the two elements. 
He boasts that the epic is an akhyanam bahvartham gruti- 
vistaram, but still says that it is the Bharatl kathii, Bharata 
story, the root of which is the dramatic episode of the Raja- 
suya, which led to the development of the story (II. 3, 2, 
13 ff.). So another poet proclaims: “I will relate the great 
good fortune of that great-hearted king the Bharata, whose 
brilliant Itihasa, story, is called the Mahabharata,” i, 99, 49. 
The reason that Krsna Dvilipayana spent three years in mak- 
ing the epic was not only that he wished to do a good thing 
but that he wished to “ extend the glory of the Pandus and 
other warriors.” 3 

Constituting a small but important part of the various 
tales told in the epic are found genealogical verses, anu- 
vanija-glokas (or gathas), which commemorate the history of 
the race of valiant kings and great seers of the past. I 
shall speak of them again hereafter. Here it suffices to say 
that such verses are either sung by professional rhapsodes, 
or recited by narrators. The rhapsodes, however, were quite 
distinct from the Brahmans, who recited the epic stories. 
For a priest to be a professional story-teller or a rhapsode was 
as bad for him as to be a juggler or a physician. 2 


There remains only one class of literature which may 
doubtfully be included under the head of literature known 
to the epic poets, the drama. Whether there was already a 
literary drama is, however, chiefly a matter of definition. 
It is conceivable that the story-tellers and rhapsodes may 
have developed dramatic works before any such works were 
written, that is, became literature in a strict sense, and that 

1 i, 62, 27-28. 

2 xiii, 23, 15, gayana nartaka 9 cai ’va plavaka vadakas tatha kathaka 
yodhakag cai ’va rajan na ’rhanti ketanam; ib. 90, 11, among apankteyas 
are kufilavas, rhapsodes, and idol-makers (above, p. 15). A priest is insulted 
on being called a professional eulogist, bandin, i, 78, 9-10. 


the akhyana may have been dramatically recited. But it is 
also true that the early epic does not mention the play or 
drama. Nevertheless a kind of drama existed before the 
epic was ended. Compare iv, 16, 43: 

akalajna ’si, sairandhri, qailusl ’va virodisi 

From the expression “ thou weepest like an actress ” one 
might hastily conclude that we have here a reference to real 
drama. But pantomime expresses weeping, and no mention 
of real drama occurs in the epic except in the passage ii, 11, 
36, where Drama is personified : 

nataka vividhah kavyah kathakhyayikakarikah, 

which is anything but an early verse. 1 In the Harivaiiga, on 
the other hand, which probably dates from a time posterior to 
our era, we find not only pantomime, abhinaya, but even the 
dramatic representation of the “great Ramayana poem,” in 
which the vidusaka, or stage-jester of the regular drama, 
takes part, H. 2, 89, 72 ; 92, 59. 

But even abhinaya, or pantomime, is not mentioned in the 
epic proper under that name and no technical dramatic term 
is found anywhere in it. This is the more surprising as the 
manner in which the epic is told gives abundant opportunity 
to introduce both the terms and allusions to dramatic repre- 
sentation. Shows of dances are frequently mentioned, but 
the spectators never hear the players even when mentioned 
as natas, a doubtful word which might be actor and may be 
pantomimist. Not to speak of the absence of Qaubhikas and 

1 Dramatic recitations are of course another matter, and pantomime must 
be separated from drama. According to Fick, Soeiale Gliederung, p. 188, the 
same relation exists in the Jatakas, where also nata and nataka do not 
yet mean actors but pantomimes, as “ dramatic performances are nowhere 
described.” This is, in my opinion, the state of affairs in the epic prior to 
the writing of the late additions (see the allusion below), ii, 11, 36, belongs 
clearly to an interpolated scene, and the fact that real drama, nataka, is 
mentioned only here in the whole epic till the Harivanya, should show its 
age. He who refers the passage to 500 b. c., must ignore its uniqueness and 
the fact that the rest of the epic knows no such word. See my Ruling Caste, 
p. 320, and also Professor Rhys Davids’ interesting note on the Brahma-jdla 
Sutta, Dialogues of the Buddha, p. 7 (with my note below, p. 67, on prekkha). 



others elsewhere mentioned as actors, and of the dramatic vitas, 
gakaras, and vidusakas, when groups of people of this grade 
are given, 1 even the granthika appears only as a rhapsode 
processional singer, and the characters are described merely 
as “ seing,” pagyanto natanartakan, ii, 38, 49 ; i, 218, 10, etc. 
The expression “ stage ” and the various vague terms for 
actors can be referred to mimes with perfect propriety and 
in the absence of everything that would indicate real drama 
ought perhaps to be so referred. In the expression “ God 
treats men as men do a doll on a string,” iii, 30, 23, the refer- 
ence must be to the sort of Punch and Judy show which is 
still performed in town and village. Even in xii, 36, 25, 
rangastii, “ stage- woman,” may perhaps most reasonably be 
explained as the equivalent of the actress mentioned above. 
Like the Harivahga, the Ramayana speaks of theatrical exhi- 
bitions, natakany aliuh (or eakruh), R. ii, 69, 4 ; G. 71, 4. 
Rhapsodic drama is alluded to also in the Mahabhasya, where, 
as Weber has shown, the actors are seen and heard and tra- 
gedies are presented in costume. But the Mahabhiirata 
neither alludes to such dramatic plays nor does it notice the 
Natasutra. 2 All that is heard seems to be songs and instru- 

1 Such groups are frequently found in lists of persons who are not eligible, 
and are generally regarded as vulgar or dangerous, but in all these groups 
among dancers, singers, rhapsodes, etc., no technical word of the regular 
drama is found. 

2 Compare Weber, IS. xiii, p. 487 ; Holtzmann, loc. cit., p. 78 ff. The latter 
scholar says “die ganze dramatische Literatur ist spater als das Mahabha- 
rata.” He means therewith, I presume, the received drama of Kalidasa and 
others. There is certainly in the epic nothing like the nataklkrta Ramayana 
of the Harivanfa. The chronological value of the Mahabhasya data would 
be greater if one knew to which century they reverted, but Weber himself 
warns against taking them as of certain worth for any time earlier than the 
end of the eighth century a. d., loc. cit., p. 320. A Punch and Judy show 
is implied in v, 39, 1, sutraprota darumayi ’va yosa. The Sutradhara appears 
only in i, 51, 15, where he is a sthapati, or architect, and a Sutah pauranikah. 
The application of the name here is apparently to the sutra, lines or plans, 
drawn up by the architect (xii, 10,983, but B. has mudra for sutra, 299, 40). 
Lists of natanartakagayanas are found in iii, 15, 14; xii, 09, 00; rangavata- 
rana, ib. 295, 5. In i, 184, 16, though natas and Sutas come with dancers and 
praisers and boxers, niyodhakas, only praisers are heard (Sutas, 188, 24). So 


ments : “ The musicians sounded their instruments together ; 
the dancers danced also; the singers sang songs,” nanrtur 
nartaka§ cai ’va jagur geyani gayanah, i, 219, 4. 

The conclusion seems inevitable that the technical nataka 
with its vidusaka, etc., that is, the drama in its full form, was 
unknown to the epic proper. What was known was clearly 
pantomime. Dramatic recitation like that of the Bhasya may 
be inferred only if one ignores the facts mentioned above, 
which is possible if the (non-hearing but) seeing of shows 
be taken as a general expression. On the other hand, the 
akhyana-reciters may have been dramatic without the set- 
ting noticed in the Bhasya. They are heard rather than 
seen. I have already noticed the fact that Narada is the 
representative of Bharata as the genius of music, and that 
the latter is not known to the epic in his later capacity. 1 

in ii,4, 7, (with vaitalikas) ; and in the danamahakratu at it, 14, 17, which is 
natanartakalasyadhyah. A dance-hall, nartanayala, nartanagara, is mentioned 
in iv, 22, 3, 16, and a preksagara, “hall for seeing,” is made according to 
(Jastra rule in i, 134, 10-11, a temporary affair for a joust, helped out with 
mancas; a samajavata (more elaborate) in 185, 16; while “spectators at an 
arena,” preksakah . . . raiigavata iva, iii, 20, 27, are alluded to. Other stage- 
words, rangabhumi, etc., occur occasionally without specific application to 
acting. The use to which preksa and samaja are put, when they are explained 
in the epic, should make one hesitate to translate the same words in Manu 
more specifically than “ shows and meetings,” and the same is true of prekkha 
in Pali. 

1 The pseudo-epic, xiii, 33, 12, says that some priests are thieves, some are 
liars, and some are natanartakas, which the commentary illustrates by saying 
that Valmlki and Viyvamitra are examples of the thief, while Bharata and 
others are examples of natanartakas (Narad a is an example of the liar, as 
he is kalahapriyah). Here, and in the quotation above, natanartaka is one, 
“ actor-dancer.” For the part played by dolls in the early Hindu drama, see 
Professor Pischel’s illuminating essay. Die Heimat des Puppenspiels (1900). 
He also gives references to previous literature on the drama. 



Of the two early epics of India, the Maliabharata, the great 
epic, is traditionally attributed to a distributor, vyiisa, who is 
also credited with the distribution or editing of the Vedas 
and of several other works. Different editions and former 
declarers are also noticed. In other words, there was no one 
author of the great epic, though with a not uncommon confu- 
sion of editor with author, an author was recognized, called 
Vyasa. Modern scholarship calls him The Unknown, or 
Vyasa for convenience. 

But if the great epic lacks an author with a real name, the 
little epic, the Ramayana, is the work of a definite personality. 
Here there is no question of disputed authorship, only of 
more or less plainly marked interpolation and addition. The 
great, maha, Bharata-epic is really, as it is designated, a col- 
lection, Samhita, the reputed author of which, corresponding 
generally to the parallel figure in Greece, yet out-Homers Ho- 
mer; while beside the huge and motley pile that goes by 
Vyasa’ s name stands clear and defined the little Ramayana of 
Valmiki, as (in this respect) besides Homer’s vague Homeric-a 
stands the distinct Argonautika of Apollonius. 

As the relation between the two Hindu epics, especially in 
point of age, has often been discussed, I do not purpose to 
repeat all the details here, but to take up the study of the 
great epic from a new point of view. For the reason why so 
much theorizing in regard to relative age has been spent on 
the epics without satisfactory result — adhuc sub judice — is 
that hitherto there has been no recognition of the underlying 
unity of epic speech. Hence discussions in regard to the pos- 
sibility of totally different origins of the two epics and the 



different ages they represent, while their common base has 
been ignored. 

In regard to the final growth of each, it may be said at once 
that neither epic was developed quite independently of the 
other. The later Ramayana implies the Mahabharata, as the 
later Mahabharata recognizes the Ramayana of Valmiki. It 
is not, then, a question of absolute separation, but only of the 
length we may go in separating. 

Neither epic has a definitive text. The question therefore 
naturally arises whether there is any use in arguing about the 
original form of either poem. In regard to the Mahabharata, 
this question has been answered negatively by Dr. Winternitz, 
who holds that all work on the epic is useless till we have the 
text of the Southern recension, of which he has lately pub- 
lished, in the Indian Antiquary, some interesting specimens. 
But it is doubtful whether the publication of the whole 
Southern version would result in a text any more definitive 
than that of the Ramayana. At most we should have two 
versions, more or less independent of each other, each showing 
omissions and interpolations as viewed in the light of the 
other. This would be of considerable value indeed, as proving 
that the text has been freely altered, a conclusion inevitable 
even without this support, but based with its aid on objective 
reality. Nevertheless, though the Southern recension would 
be thus valuable, its absence does not preclude the possibility 
of obtaining provisional data of importance from the Northern 
recension alone, either in regard to its relation to the Rama- 
yana or in respect of its own development. Such data must 
finally be checked in detail by a comparison with those of the 
alternate text; but as a whole they suffice to cast much light 
on several moot points, and in themselves are useful in de- 
monstrating that the great epic is the result of the labors of 
different writers belonging to different schools of style and 
thought; a result diametrically opposed to the view of the 
method calling itself synthetic, and likely to be rather twice- 



proven than disproven by the eventual publication of the 
Southern text. 

In regard to the texts of the Ramayana, I need only refer 
to the invaluable essays of Professor Jacobi, seconded by the 
recent analyses of Dr. Wirtz and Dr. Liiders , 1 especially as 
this epic is not the chief object of consideration in this vol- 
ume. It is, however, obvious that exactly the same conditions 
obtain here as in the case of the great epic, and it may be 
added that if there were a third epic the same conditions 
would obtain there. There is no fixed epic text because Hindu 
epic poetry was never fixed. All epic poems were transmitted 
at first orally, and the various rewriters treated them exactly 
as the rhapsodes had previously done, altered and added as they 
pleased. Reconstruction of the original text is therefore out 
of the question. All that can be done is to excise the most 
palpable interpolations in each traditional rendering. 

Neither of the epics, as such, is recognized before the late 
period of the Grhyasutras, and the first epic recognized here 
and in other Sutras is the Bharata. The question has often 
been raised which epic is the older. In our present state of 
knowledge it may be said that this question cannot now and 
probably never can be answered in one word. In the first 
place, it will always be idle to speak of either epic as the older 
without specifying whether one means the present text or the 
original text; for that these, in the case of either epic, are 
convertible terms is an idea refuted by even a superficial 
acquaintance with the poems. Assuming, however, that the 
question implies priority of epic qua epic as a new genus of 
literature, and whether this form first arose as Ramayana or 
(Maha) Bharata, tins too cannot be answered categorically, 
because parts of the latter are older than the former, and the 
former is older than the mass of the latter, as will be shown. 
Personally I have no doubt that the Pandu (pandava) form of 
the great epic is later than the Rama epic ; but, since one was 

1 Das Ramayana (together with special studies mentioned hereafter), hy 
Professor Jacobi; Die Westliche Rezension des R., by Dr. Hans Wirtz; Die 
Sage yon Rsyayrnga, by Dr. Heinrich Liiders, Gott. Nadir. 1897, p. 87. 



a slow outgrowth from a Punjab Kuru epic, and the other, of 
unknown antecedents, was developed far to the East, in much 
more polished form, while only the Bharata is recognized in 
Vedic literature, I have as little doubt that there was a Bha- 
rata epic before there was a Ramayana ; whereof also I shall 
speak again in a subsequent chapter. Here I wish merely to 
notice, in passing, the ridiculous claim that the Ramayana dates 
from the “ twelfth or thirteenth century ” b. c. This claim 
has been made not only by Hindus but by Occidental scholars. 
Whether there was a Rama story at that period or (just as 
well) twelve or thirteen centuries earlier no man can know. 
But that Valnriki’s Ramayana can lay claim to no such age 
the slightest historical consideration will show, not to speak 
of an examination of the almost classical metre of the poem. 

The Maliabharata, besides giving the Rama story as an epi- 
sode, Rama-upakhyana, has four direct references to the Rama- 
yana (apart from an allusion to Great Itiliasas). The first is i 
the citation of a verse actually found, as Professor Jacobi has 
shown, in the extant poem of Valnriki, api ca ’yam pura gltah ' 
qloko Valmlkina bliuvi, vii, 143, 67 (R. vi, 81, 28). 1 The 
second is the citation of a verse from Bhargava’s Ramacarita 
(Bhargava being, as Professor Weber has shown, a title of 
Valmlki), which agrees in sense and Avords closely enough 
Avith R. ii, 67, 11, to indicate that the Maliabharata poet of this 
passage, xii, 57, 40, had in mind this or the original form (for 
it is to be noticed that the name is not fixed) of this verse 
in the Ramayana, 2 and to make improbable the synchronous 
collection of the former epic at xii, 67, and 68 (cf. §1. 15) : 

M. cdokae ca ’yam puragito Bhargavena mahatmana 
akhyate Ramacarite nrpatim prati, Bharata, 
rajanam pratliamam vindet tato bharyaiii tato 

rajany asati lokasya kuto bharya kuto dhanam 

1 na hantavyah striya iti, “AVomen may not be slain.” The general rule 
is found also in R. ii, 78, 21, avadhyah sarvabhutanam pramadah ksamya- 
tam iti. 

2 Rather than a common source, as I thought previously, AJP. xx, p. 34. 



R. arajake dhanam na ’sti na ’sti bharya ’py arajake 
idain atyahitam ca ’nyat kuto satyam arajake 

The third and fourth cases refer to the Ramayana without 
mention of the poet: iii, 147, 11, “ Hanmnat is very renowned 
in the Ramayana ; ” xviii, 6, 93 (repeated in the Harivan§a) : 
“In the Veda (which is) the beginning (of literature), in the 
holy Ramayana (which is) the end, and in the Bharata (which 
is) the middle, in all (literatures), Vishnu is besung.” 1 The 
Harivaiiga adds three more references, two to V almiki , and 
one to a dramatic representation of the Ramayana. Valmiki 
in these passages and perhaps in i, 55, 14, as Professor Holtz- 
mann surmises, is credited with being a poet. This is also 
implied in xiii, 18, 8-10. Everywhere else, and he is men- 
tioned several times, ii, 7, 16 ; iii, 85, 119 ; v, 83, 27 ; xii, 207, 
4, he is recognized only as a saint. 

In this material, which I recapitulate here only for a view 
of the chief data, 2 the most striking fact is the antithesis be- 
tween the notices of the Ramayana as found in the early and 
later Mahabharata. The Rama story is referred to over and 
over, and the whole tale is told independently at iii, 273, if., 
but until we come to the much expanded Drona and the 
didactic epic, references to the poem are merely to the Rama 
tale, references to the reputed author are merely to a saint 
recognized as an ascetic but not as a poet. Even as a saint 
the evidence is conflicting, for, though usually a Vishnu adhe- 
rent, in the passage cited above from the Anu§asana, Valmiki 
is a (,’ivaite. The individual allusions prove, therefore, noth- 
ing in regard to the general priority of Valmiki as the first 
epic poet. They prove only that the Mahabharata was not 
completed before Valmiki wrote, just as the mention of the 

1 vede Ramayane punye (may go with the next word) Bharate, Bharata- 
rsabha, adau ca ’nte ca madhye ca, Harih sarvatra glyate. The last clause 
may be taken more indefinitely, “ in V., R., and M. ; in the beginning, end, and 
middle, everywhere.” But such correlation is common (e. g., vede loke frutah 
smrtah, R. ii, 24, 28) and seems to me to be implied here. 

2 Weber, TTeber das Ramayana, first collected it; Jacobi, Das Ramavana, 
added to it ; Holtzmann, Das Mahabharata, iv, p. 60 ff ., has briefly summed 
it, with other references (omitted here) and independent additions. 



Vayu Purana in the Mahabharata shows only that there was a 
Purana of that name not before the Bharata’s beginning but 
before its end. They show also that no antipathy or wish to 
suppress Valmlki's name influenced the Bharata poets, who, 
therefore, had they simply retold or epitomized a poem recog- 
nized as Valmlki’s would probably (as it seems to me) have 
mentioned his name in connection with the Rama-upiikhyana. 

Professor Jacobi is of the opinion that a verse of inferior 
form in the episode points to borrowing because it is inferior. 
But a great poet is more apt to take a weak verse and make 
it strong than is a copyist to ruin a verse already excellent. 
Further, the subject-matter of the Kavya and episode is 
treated differently in several particulars (details, loc. cit.), 
which points to different workings-over of older matter rather 
than to. copying or condensing. Professor Jacobi also em- 
phasizes the fact that the great epic cites Viilnffki but Valmiki 
does not cite or refer to the Bharata. This holds good for 
the great epic only from a “ synthetic ” point of view, which 
Professor Jacobi of course rejects. The normal attitude of a 
Hindu toward his sources is silence. He is rather careful not 
to state than to proclaim that he is treating old material, so 
that there is nothing surprising in Valmlki’s not speaking of 
a predecessor. Moreover, in the later Ramayana, which un- 
questionably betrays acquaintance with the Mahabharata, there 
is no more recognition of the latter than there is in the earlier 
part of the poem; a fact -which weakens considerably the 
argument of silence as applied to that earlier part. 

Apart from vii, 143, 67, the Mahabharata knows the poet 
Valmiki only in the twelfth and thirteenth books ; whereas it 
knows everywhere the Rama tale, a poem called the Rama- 
yana, and a saint known not as a poet but as an ascetic called 
Valmiki. It gives the Rama-episode as it gives other ancient 
tales handed down from antiquity without having been as- 
signed to a specific author. The Rama-upakhyana stands to 
the Ramayana somewhat 1 as the N ala-upak hyana stands to 

1 Emphatic, of course, as tlie example is a great exaggeration in difference 
of age and style. 



the Naisadha, in that it is an early tale of unknown author- 
ship which a poet made his own. Long before there is any 
allusion to Valmlki’s Ramayana, the base of the great epic, 
the substance of the Bharat! Katha, is recognized in Hindu 
literature ; while the latest addition to the great epic refers to 
Valmlki himself as a man who is to be, that is, who is already, 
famous, yagas te 'gryam bhavisyati, xiii, 18, 8-10. Between 
these extremes lies the Ramayana. 

The Ramayana recognizes Janamejaya as an ancient hero, 
and known Kurus and Pancalas and the town of Hastinapur 
(ii, 68, 18). The story of the Pandus, the gist of the present 
epic, is presumably later than the story of Rama ; the former 
everywhere recognizing the latter as an ancient tale. 1 We 
must therefore on these data make the following distinctions : 

(1) The story of Rama is older than the story of the 

(2) The Pandu story has absorbed the Bharat! Ivatha. 

(3) The Bharati Katha is older than Valmlki’s poem. 

Although we have but two ancient Sanskrit epics, there is 
no reason to suppose that epic poetry began with the extant 
poems in our possession. As w'as remarked above, the Maha- 
bharata alludes to the “ Great Itihasas,” which may perhaps 
imply other poems of epic character and considerable extent. 2 
Nor can it be supposed that epic poetry was suddenly 

1 ii, 76, 5, asambhave hemamayasya jantos tatha ’pi Ramo lulubhe 
mrgaya; iii, 11, 48, Vali-Sugrlvayor bhratror yatha strikanksinoh purii ; ix, 
31, 11, Ravano nama raksasah, Earaena nihato rajan sanubandhah sahanu- 
gah ; so ix, 55, 31 ; sometimes interpolated, as when Ravana and Indrajit 
are mentioned in i, 155, 44, but not in C., which omits all 41-44 (after 0081). 
Other references will be found in iii, 25, 8 ; 85, 65, etc. Compare Holtzmann, 
loc. cit., p. 62 ff. According to xii, 340, 85 ff., Rama comes at the beginning 
of the last era; Krishna, at the beginning of the present era (Rama’s two 
adjutant monkeys are here Ekata and Dvita). Rama is recognized here as 
an incarnation of Vishnu, and also in iii, 99, 40. 

2 I say perhaps only, for “great” is a word often used without reference 
to extent. Thus the mahad akhyanam of xiii, 2, 1, is only a philosophical 
fable (about a snake and Karma), 83 jlokas long. 



invented by one poet. The numerous “ancient tales” of 
epic character must have furnished a large body of epic phrase 
as well as fable, out of which and on the basis of which arose 
our present epics. This is rendered probable also by the fact 
that such brief epic verses as are preserved in other works, 
although not always from the extant epics, yet have the same 
character as the verses of the Bharata and Ramayana. Fur- 
thermore, as said above, the epic itself admits that the present 
text is not an original work . 1 

We cannot suppose then, even if one epic could be shown 
to be prior to the other, that this prior epic was the first work 
in epic versification. We must let pass the statement of the 
Ramayana itself that Valmlki invented the gloka verse, for, 
though Viilmlki may have been the first to set out to write an 
epic in §lokas, it is scarcely worth while to discuss such a 
palpable bit of self-glorification as that in which the later 
Ramayana here indulges . 2 * * 5 As the two Greek epics were both 
based to a certain extent on the general rhapsodic phraseology 
of the day, so the two Hindu epics, though there was without 
doubt borrowing in special instances, were yet in this regard 
independent of each other, being both dependent on previous 
rhapsodic and narrative phraseology. 

I cannot, in short, think that such a very large number of 
identical phrases as I shall enlist below can owe their identity 
simply to one poet’s copying of another. For the similarity 
goes too deep, into the very grain of the verse. The exposi- 
tion, I fear, will be tiresome in its study of minute detail, but 
it is necessary to a full understanding of the conditions of the 

1 i, 1, 26 : acakhyuh kavayah kccit sampratyacaksate pare akhyasyanti 
tathai ’va ’nye itihasam imam bhuvi (cited by Holtzmann). 

2 So with the tale of the two rhapsodes who “sang” the poem with musi- 

cal accompaniment, after it had been composed and taught to them (so that 
in the first instance it was recited as a narrative). But all this is the product 

of a later age making up its own fictions and myths, such as the singing sons 
Ku?a and Lava made out of kucilava, an ordinary word for rhapsode. That 
Valmiki could not have “invented the 5 loka”is shown by the presence of 
an earlier form of ylokas in the Brahmanic literature retained in Mbit. 




A characteristic of the common basis of epic verse may be 
traced back to the Rig Veda. This consists in a rhetorical 
duplication of a dissyllabic iambic noun, which favors the 
diiambic close of the octosyllabic pada or verse, as in these 
first three examples, or of the twelve-syllable pada, as in the 
last example : 

rtavana jane-jane, RY. v, 65, 2 

yac cid dhi tvam grhe-grhe, ib. i, 28, 5 

haskartaraih dame-dame, ib. iv, 7, 3 ; vii, 15, 2 

sa dareataqrTr atithir grhe-grhe 

vane-vane qieriye takvavJr iva 

janam-janam janio na ’ti manyate 

viqa a kseti vitjio vieam-viQam, ib. x, 91, 2 

With the last, compare also RY. i, 123, 4, where grhaiii- 
grham, clive-dive, agram-agram stand at the start, not at the 
end. Sometimes a whole pada consists of only such com- 
posita, as in x, 97, 12, aiigam-angam parus-parus (cf. v, 53, 
11 ; x, 163, 6). In the Rig Veda, again, pure adverbs thus 
duplicated are never found at the end of the pada ; only such 
nominal adverbs as those above, the nearest approach to pure 
adverbs so used being idam-idam, a pronominal adverb closing 
a pada at vii, 59, l. 1 In the epic, however, the forms are usu- 
ally adverbs, usually at the end, 2 usually in ^lokas ; in the Rig 
Veda, never pure adverbs, usually at the beginning or in the 
middle, seldom at the end of the pada, and usually not in 
ijlokas, but in gayatrx and especially in jagat! or tristubh 
verses. The first examples given above are, therefore, rather 
the exception than the rule as far as their position goes. But 
I think we may see in them the precursors of the epic for- 
mula used in closing the hemistich. The V eda puts the form 
where it best shows the iterative intensity; the epic puts it 
where it best helps the metre. Thus : 

1 Compare the list of such composita in Professor Collitz’s paper, Abhandl. 
d. V. Orient. Congress, 1881, p. 287. 

3 Exceptions of course occur, as in M. vii, 7, 53, punah punar abhajyanta 
sinhene 've ’tare mrgah ; R. iv, 43, 53, ahany ahani vardhante. So upary upari 
sarvesam and sanunam, Nala 1, 2; and R. v, 13, 10, respectively. 



punah-punar matara navyasl kah, RV. iii, 5, 7 
punah-punar jayamana panic I, II V. i, 92, 10 
nihtjvasya ca punah punah R. i, 54, 5 
(nihovasya) pratyaveksya punah punah, M. ix, 29, 49 

The epic uses this metrical convenience constantly, some- 
times too often, as in ix, 82, 6, 8, 9, where punah punah is 
repeated three times. Other adverbs of the same sort in both 
epics are prthak prthak, muhur rnuliuh, ganaih ganaih. In a 
word, both epics close the hemistich in this antique Yedie 
manner, though the epic style lias somewhat changed the 
relation of the phrase to the pada. 1 

Like these stereotyped terminals in their epic application is 
the countless number of verses ending with the same diiambic 
form, vocative, nominative, or oblique case, of one compound, 
and the less frequent (because less needed) common form of 
the prior pada’s pathya ending, such as mahabala, paramtapa, 
ariihdama (prior, mahabaho, °prajna, °virya, maharaja, ra- 
jendra) ; pratapavan, paravlraha, mahiinirdhe, ranajire, rana- 
murdliani, ranakarkagah, the oblique cases of mahatman 
(constantly used), and such diiambic phrases as balad ball, 
suto ball. All of these are used in the same way in both epics, 
most of them repeatedly. In some, the word passes back of 
the diiambus and leads us toward the whole pada-phrase 
though not quite reaching it. Of such sort are ranakarka- 
§ah (above), yuddhadurmada, saiiigramamurdhani, (Varunah) 
satyasamgarah, nama namatah, Qatrunisudana, akutobhayah, 
krodhamurcchitah. In others, the word falls short, but the 
position of the adjective is fixed and it is generally preceded 
by the same combination as in (eapam, gadam. or clhanur) 
udyamya vlryavan, and the common final manada. 2 

1 And also extended it in the form gate gate (instead of the noun) in 
dagahe vSi gate gate, xiii, 107, 43. Of epic phrases, I have noted also grhe 
grhe, M. ii, 15, 2; R. v. 20, 20; and (passim) pade pade, yoge voge, rane rape, 
and in M., jane jane and, in the more unusual initial position, miisi masi 
(Vedic and M. ix, 37, 4), kale kale, ix, 37. 23. Of the phrases quoted above, 
muhur muhuh occurs often; panaih panaih, e. g., M. ix, 20, 104; R. ii, 40, 22 
and G. vi, 111, 13; prthak prthak, e. g., AI. ix, 37, 23; G. vi. 54, 50 ; 77. 1. 

2 Among those mentioned, paraviraha is converted into hanta in tristubh, 



From these compounds, not only in form but in fixed posi- 
tion common to both epics, we may pass to cases like (svate- 
jasa, often) svena tejasa, where the piida ends with two words 
wliich take in more than the diiambus, for example, biblira- 
tim svena tejasa, jvalantim svena tejasa, the former in M. xii, 
825, 2 ; the latter in R. vi, 107, 11 and G. 80, 33. 

The fixed form is shown most conspicuously in similes that 
are common to both epics, and are of the mechanical form 
instanced in the last two sorts of examples, namely in diiam- 
bic or more than diiambic terminals. Thus there are fixed 
phrases w'hieh are different except for the terminal, which 
again is common (as a fixed terminal) to both epics, for 
example : 

dandahata ivo ’ragah, in M. and in R. 
pancaclrsa ivo ’ragah, “ “ 

dandahasta iva ’ntakah, “ “ 

pacahasta iva ’ntakah, “ “ 

vyattananam iva ’ntakain, “ “ 

jvalantam iva pavakam, “ “ 

didhaksur iva pavakah, “ “ 

vidhuma iva pavakah, “ “ 

patariiga iva pavakam, “ “ 

tjalabha iva pavakam, “ “ 

Such phrases are common not only to the two epics but to 
outside literature. Thus the iva pavakah formula appears in 
the Dhammapada, 71, as bhasmacehanno va ptivako (epic, 
bhasmapanno iva ’nalah), and the same is true of a limited 
number of whole pada-phrases, not only in pure proverbs, but 

R. iv, 31, 5 (°ghna is a common side-form) ; pratiipavan is perhaps least com- 
mon in R., but it serves with vlryavan ; for example, in R. vi, 69, 109 ; 76, 21, 
27, ff., where follow a quantity of mahabalas. Like vlryavan is vegavan with 
vegitah (vegena in the prior pada). M. has ativiryavan, as in iii, 283, 7. 
The simple form is rare in any other position, e. g., G. v, 2, 23; 3, 71. As 
a terminal it occurs in R. about forty times in the sixth book, uncounted 
often in M. The common Mahabliarata terminal marisn, I have not noticed 
in the Ramayana. It appears to belong to later diction and indicates an 
epic recasting, as does, e. g., the late tatrabhavant of R. ii, 106, 30. 



in current similes and metaphors, like kalarh na ’rhanti soda- 
glm, xii, 277, 6; Manu, ii, 86; and Buddhistic, Dh. P., 70, 
kalaih na ’gghati solasim ; or maikagonitalepanaui, Dh. P., 
150; Manu, vi, 76; Mbh. xii, 330, 42 (Mait. Up. iii, 4). 1 

In some cases the variety of paclas constructed on a com- 
mon terminal is very large, such as the various forms of what 
appears most simply as ganta ’si Yamasadanam, yilto 'si Yama- 
sadanam. Thus both epics have yiyasur Yamasadanam and 
anayad Yamasadanam, along with other forms more peculiar, 
Yamasya sSdanam prati, R. vii, 21, 1; praliinod Yamasadanam, 
prahinon mrtyulokaya, 2 garair ninye Yamaksayam, M. ix, 26, 
29, ninye vaivasvataksayam, M. vii, 26, 53, gato viiivasvata- 
ksayam, G. vi, 82, 183, yami vaigravanalayam, G. vi, 82, 167 ; 
nayami lokam (with Yamasya omitted, tristubh), M. viii, 85, 
31; nayami Yamasya gehabhimukham, R. vii, 68, 20; gami- 
syarni Yamasya mulam, R. v, 28, 17 ; mrtyupatham nayami, 
G. vi, 36, 118 ; mrtyumukhaih nayisye, M. viii, 42, 11 ; 
mrtyumukhagatam (anesyamah), G. iv, 45, 9. Evidently in 
these cases the ancient plmases Yamasadanam, Yamaksayam, 
are built upon in several ways, and then the desire for variety 
leads to the pulling away of the base of the old-fashioned 
phrase, and the superstructure is sliiftecl to a new base, gen- 
erally in the later epic, the double meaning of ksaya helping 
in anayat ksayam, ix, 27, 48. Like changes occur in the 

1 There are also clear traces of dialectic influence in the adaptation of 
some of these standing phrases. On this subject I shall speak more fully 
below. Here I will illustrate what I mean by one example from the Rama- 
yana. There is a common phrase which begins tam apatantam sahasa, or 
some similar final word, the first two referring to a masculine noun (weapon). 
When we find, in R. vi, 67, 47, this same phrase used of a neuter noun, tad 
apatantam, we are justified neither in assuming that the poet was wholly 
indifferent to grammar nor in agreeing with the commentator that the mas- 
culine form is an archaism countenanced by Vedic usage, punstvam arsam. 
It is simply a case of borrowing a convenient grammatical form (not San- 
skrit, but Prakrit), for apatantam is a regular patois neuter participle. Forms 
of this sort are adopted into the epic merely for metrical reasons, showing 
that they were borrowed from the common speech of the day when con- 
venient; which shows again that the epics (both are alike in this particular) 
were written in Sanskrit and not made over from Prakrit originals. 

2 See for references, Appendix A, s. v. 



sutumulam yuddham phrases, generally ending with lomaliar- 
sanani, but occasionally in a new setting, Yamariistravivardh- 
anam, as in, 79, 60; ix, 10, 61; 11, 5, etc. ; in tristubh, 
°vardhanah, vii, 145, 97. 

Especially is the monotony varied in the conventional 
phrases of conversation. Both epics have etac chrutva tu 
vacanam, tasyai ’tad vacanarh grutva, idam vacanam abravlt, 
grutva tu vacanarh tasya; and again the phrases are sliifted, 
tatas tad vacanam grutva, tad etad vacanam grutva (old and 
rare), G. iv, 38, 46 ; grutva tasaiii tu vacanam, M. ix, 35, 52 ; 
idam vacanam uktavan, G. v, 68, 24 ; and in many other ways, 
too tedious to recount. 

Herewith w'e come to the pada phrase, wiiich fills the whole 
half-verse with the same locution, as in palayanaparayanah, 
parasparajighansavah. In the Am. Journal of rhilology, xix, 
p. 138 ff., I cited verses of the Mahabharata which are full of 
such phrases. Such passages are also easily found in the Ilama- 
yana, of which I will give but one instance, vi, 71, where gl. 
67 alone contains four such phrases : tarn apatantam nigitam 
garam aglvisopamam, ardhaeandrena ciccheda Laksmanah para- 
viraha (with others following). Here the whole gloka with 
the exception of the proper name consists of iterata. In the 
Ramayana, too, we find, as often in the Mahabharata, two 
iterata enclosing a verse that is new, as in iv, 11, 18, where 
the independent verse is sandwiched betwreen the iterata 
tasya tad vacanam grutva and krodhat saihraktalocanah, 
which arrangement is found again, ib. 73. In G. iii, 57, 15, 
the hemistich consists of two whole phrases, rosasamraktana- 
yana idam vacanam abravlt. In G. vi, 27, there are nine ite- 
rata in the first eighteen glokas. I mention this that there 
may not seem to be any distinction in tins regard in the two 
epics. Both have many chapters wiiich teem with verbal or 
whole pada-iterata, the later the more. 1 Noticeable are tlieir 

1 The cumulative style is characteristic, naturally, of later sections. So, 
for instance, in the late fourteenth chapter of the thirteenth book, within 
the compass of about thirty glokas, 249 ff., we find sarvabharanabhusitam, 
Barvabhutabhayavaham, yakratulyaparakramah, trigikham blirukutim krtva. 



extent and variety. There is hardly a field in which Yyasa 
and Valmiki do not echo the same words. General descrip- 
tive epithets and phrases that paint the effect of grief and 
anger, or the appearance of city and forest; the aspect of 
battle and attitude of warriors, with short characterization of 
weapons and steeds, are all as frequent as the mass of similes 
found in both epics in the same words. In the last category, 
identical similes are drawn from gods, men, animals, and phy- 
sical phenomena. Again, both poets, as shown above, use 
the same phrases of speech, as they do also of noises, and 
of the course of time; and finally there are many didactic 
verses, almost or quite the same in both epics. 

In the list of parallels given elsewhere 1 1 have incorporated 
such examples as I have noticed of identical or nearly identi- 
cal phrases and verses. Illustrative additions are occasionally 
added, not to add weight to the general effect, for the number 
of cases of actual identity is sufficiently large, but to supply 
material for fuller treatment of tins whole subject eventually. 
The three hundred examples here registered include also some 
cases where verbal identity is not quite complete, such as 

M. iv, 19, 29, 

prabhinnam iva matangam pariklrnarii karenubhih 

G. v, 14, 28, 

karenubhir maharanye pariklrno yatha dvipah 

and I have not perhaps been thoroughly logical in the admis- 
sion or exclusion of such cases ; but in general I have sought 
to establish an equation not only in the thought but in the 
expression of the thought, and for the most part have omitted 
such parallels as did not tend to bring out the verbal identity. 2 

pagahastam iva ’ntakam, dvitiya iva pavakah (to which one text adds vidhu- 
mam iva pavakam) all common iterata of both epics, but far in excess of 
the usual number; as in G. vi, 27 (above). 

1 Appendix A. 

2 I have omitted, for example, such cases as iii, 30, 42, karmana tena 
papena lipyate nunam igvarah ; G. vi, 62, 22, vidhata lipyate tena yatha 
papena karmana (R. vi, 83, 23 quite otherwise), though I have no doubt that 
the tirades against God and duty (G. 15 ff.) in each epic (as in this case) 
belong together. Some few proverbs are also entered. 



Those I have collected were gleaned incidentally from a field 
which I traversed with other objects in view, and I have no 
doubt that these parallels could be largely increased by a 
close and systematic comparison of the two epics throughout. 
The alphabetical arrangement followed is merely for conven- 
ience of reference. I should have been glad to group the 
examples according to their content also, that I might have 
shown more fully the varied fields they occupy, but, as this 
would have taken too much space, the remarks made above on 
this subject and the fomier grouping made in a preliminary 
study of the question two years ago 1 must suffice. 

I will suppose that the reader has now read Appendix A. „ 
lie will have noticed in so doing that, just as the Uttara Rfun- 
ayana, as well as the real poem of Valmiki, is recognized in 
the pseudo-Bharata , 2 so in the expressions asid raja Nimir 
nama, ekantabhavanugatah, and yasya prasiidaiii kurute sa 
vai tarn drastum arhati, we have a direct copy on the part of 
the Uttara Ramayana 3 not only of the early epic but of the 
pseuclo-epic’s episode of the White Country and even of 
the very words employed in the description of the Whites 
(Islanders, to retain the usual name, though oidy country is 
really meant; Kaslunere, I think). There are several such 
passages in the Uttara reflecting the great epic in its earlier 

1 AJP. xix, p. 138 ff., 1898. 

2 Thus the story of Rama pudraghatin, as told in R. vii, 75-76 (G. 82-83), 
killing Qambaka or ^ambuka is recognized with an “ I have heard,” yruyate, 
xii, 153, 67 (where Jambfika takes the place of Qambuka). 

3 So in the praksipta passage after R. iii, 56, where Sita demands signs of 
the god Indra, and he appears with the devaliiigani : “ He touched not earth 
with his feet, winked not, had dustless garments and unfaded garlands,” as 
in Nala 5, 12-24, which the praksipta clearly copies. So, too, in the same 
book, iii, 60, not in G., evidently an artistic improvement on the preceding 
sarga, in pi. 26, Rama says : (drsta 'si) vrksair acchadya ca’tmanarii kim maiii 
na pratibhasase, as DamayantI says (Xala 11, 9: drsto ’si) avarya gulmair 
atmanam kim mam na pratibhasase; and in pi. 17, Rama cries out: apoka 
pokapanuda . . . tvannamanam kuru ksipram priyasamdarpanena mam, as 
DamayantI, 12, 104, and 107 : vipokam kuru mam ksipram apoka priyadarpana 
satyanama bhava ’poka apokah. 



parts as well. Compare for instance tlie division of Indra’s 
sin as related in M. v, 13 with R. vii, 85 and 86. It will be 
necessary only to cite M. v, 13, 12, 

raksarthaih sarvabhutanaiii visnutvam upajagmivan 
and from ib. 13-15, 

tesaiii tad vacanaiii qrutva devanam Visnur abravlt 
mam eva yajataiii Qakrah pavayisyami vajrinam 
punyena hayamedliena mam istva pakacasanah 
punar esyati devanam indratvam akutobhayah 

as compared with R. vii, 85, 18, 20-21, which give exactly the 
same words. 

But this correlation exists not only in the later parts of 
both epics and in the later part of the Ramayana and an 
earlier part of the Bharata. It is just as easy to reverse the 
positions, as for instance in the account of creation at R. iii, 14 
(G. 20) and M. i, 66. This passage is instructive as an ex- 
ample of the way complete passages were roughly remem- 
bered and handed down with shifting phrases, omissions, and 
insertions : 

M. 66, 58, 

dhrtarastrl tu hahsahc ca kalakahsang ca sarvaqah 

R. 14, 19, 

dhrtarastrl tu hansamj ca kalahahsain} ca sarvacah 

M. ib. 

cakravakaiiQ ca bhadra tu janayamasa sai ’va tu 

R. ib. 

cakravakaiiq ca bhadrarii te vijajne sa ’pi bhamiDl 

G. 20, 20, 

dhrtarastrl tv aj an ay ad dhaiisan jalavibarinah 
cakravakan^ ca bhadram te sarasaiiQ cai ’va sarvaqah 

M. 59, 

cukl ca janayamasa cukan eva yaeasvinl 
kalyanagunasampanna sarvalaksanapujita 

G. 21, 

cukl qukan ajanayat tanayan vinayanvitan 
kalyanagunasampannan sarvalaksanapilj itan 



[R. 20, 

(Juki natam vijajne tu natayam vinata suta] 

M. 60, 

navakrodhavaqa, narlh prajajne krodhasambhavah 
mrgl ea mrgamanda ca bar! bhadramana api 

R. 21, 

dacakrodhavaca, Rama, vijajne ’py atmasambhavah 
mrglni ea mrgamandaiii ca harlm bhadramadam api 

G. 22, 

tatha krodhavaqa nama jajne sa ca ’tmasambhavan 
mrglrn mrgavatlm cai ’va cardulliii krostuklrii tatba 
M. 61, 

matang! tv atha (jardull cveta surabhir eva ca 
sarvalaksanasampanna surasa cai ’va bhamim 

R. 22 (and G.) a, do., but acc. ; b, 

sarvalaksanasampanna surasam kadrukam api 

M. 62 = R. 23 almost exactly, and the following verses agree 
much in the same way, until one passage which I will cite 
entire, as follows: 

MahAbhArata(1, 66, 67-68): 

tatha duhitarau rajan 
surabhir vai vj-ajayata 
rohini cai ’va bhadrarii te 1 
gandharvi tu yajasvini 
vimalam api bhadram te 
analam api, Bharata, 
rohinyarii jajnire gavo 
gandharvyam vajinah sutah 
sapta pindaphalan vrksan 
anala ’pi vyajayata 
(70, b) surasa ’janayan uagau 

kadruh putrans tu pannagan 

RAmayaxa (iii, 14, 27-28) : 

tato duhitarau, Rama, 
surabhir devy ajayata 
rohiniih nama bhadram te 
gandharvirii ca yayasvinim 

rohiny ajanayad gavo 
gandharvi vajinah sutan 

(see 31, below) 

surasa ’ janayan nagan, 

Rama, kadru 9 ca pannagan 
(29) manur manusyan janayat 
(31) sarvan punyaphalan vrksan 
anala 'pi vyajayata 

The last verse in R. gives the origin of the four castes 
(Ruling Caste, p. 74, note), where G. has manur manusyan . . . 

1 bhadra tu, in C. 



janayamasa, Raghava. G. has virtually the same text, insert- 
ing llama and omitting the mention of Anala’s birth, giving 
only her progeny. In the last verse G., like M., has sapta 
pindaphalan vrksan (but) lalana (sic) ’pi vyajiiyata. There is 
here the same substitution of Rama and Bliarata observable in 
the late Ivaecit chapter. 1 

In my Proverbs and Tales 2 I have shown that a scene of 
the Ramayana is exactly duplicated in the Harivai^a. An- 

other similar case is found in 
(both full of iterata) : 
Haeivax^a : 

(see verses below) 

vartamane mahaghore 
samgrame lomaharsane 
panavanarh tathai ’va ca 
pankhanam patahanam ca 
sambabhuva mahasvanah 
hatanam svanatarh tatra 
daityanam ca 'pi nisvanah 


and further, 

pastrapuspopahara sa 
tatra ’sid yuddharaedinl 
durdarpa durvigahya ca 

H. 13,666 ff. ; G. vi, 19, 12 ff. 

Tartamane, etc. (— M.). 

tato bherimrdaiiganam 
patabanaiii ca nisvanah 


liatanarii stanamananam 
raksasanarii ca nisvanah 

(see the first verse, above) 
and further, 

pastrapuspopahara sa (v. 1. ca) 
tatra 'sid yuddhamedini 
duspreksya durvipa cai ’va 

R. here (sarga 44) has samutthitam in <jl. 10, but in the 
following, panavanam ca ni(h)svanah, as in H., and hayanaih 
stanamananam (with ca for sa in the first pada of the last 
stanza). The only important variant is in the last verse, 15, 
where, instead of the stereotyped pada of G. and H., stands : 
durjneya durniveqa ca gotiitusravakardama 

1 AJP. vol. xix, p. 149. 

2 ib., vol. xx, p. 35. I showed here a score of proverbs common to both 
epics, most of which had been previously noticed. Another, not noticed, is 
aliir eva aheh padan vijanati na sampayali, R. v, 42, 9 ; ahir eva liy alieh 
padan papyati ’ti hi nah prutam, M. xii, 203, 13. See also the note below, 
p. 83, note 2. 



HB. has a few slight changes, 3, 58, 66 £f., with samutthitam 
like R. (R. indicates the Bombay text only.) 

The identity of R. iv, 40, 20 ff., with the geographical pas- 
sage H. 3, 46, 42 If. = 12,825 ff., can be established on sight: 
G. 19, nadim bhagarathnii cai ’va sarayum kaueikun api = H., 
where R. 20, has ramyam for cai ’va in G. and H. ; but for api, 
R. and H. have tatha. The next stanza, G. 20, mekalaprabha- 
vam gonam, agrees only in this text with H. 44. The next 
verse in H., gomati gokulaklrna tatha purva sarasvatl is in 
G. 24 (in aec.) ; ib. b in G. reads : nadim kalamaslm cai 
’va tamasaiii ca mahanadlm, where HC. and R, both have 
mahi(m) kalamalil(iii) ca ’pi (cai ’va, HB. kalanadi). So 
R. and IIC. give the Magadhas the epithet mahagramah and 
add paundra vaiigas tathfu ’va ca, where G. has magadlian 
dandakulang ca vahgan angans tathai ’va ca (12,831, G. 25), 
and HB., §1. 49, Magadhahgca mahagraman angan vahgtihs 
tathai ’va ca. G. 26, a, b, e are identical with H. 12,830, c, d, 
and 12,831, a ; with a slight v. 1. in HB. 48. There are here 
the usual aberrations from any fixed text, but on the whole 
the two passages are identical. 

Another passage, G. i, 24, 9, 11-12, appears to be one with 
(M. iii, 52, 15 and) 31. iv, 70, 10-12 (after the first verse, it 
agrees with R. 21, 10-12) : 


ma dharmyan nlnagah pathah 1 
esa vigrahavan dharma 
esa vlryavatam varah 
esa buddhya ’dhiko loke 
tapasam ca parayanara (y. I. °ah) 
eso 'stram yiyidham yetti 
trailokye sacaracare 
na cai Va ’nyah puman yetti 
na vetsyati kadacana 
na deva na ’surah kecin 
na manusya na raksasah 
gandharyavaksapra varah 
1 This pada alone appears in iii, 52, 15. iy, 70, 10 has the following verses; 
G. has both. R. omits G.’s 9 entirely. 

Ramayaxa (G.) : 

anrtam ma vacah karslr 
ma dharmyan ninafah pathah 
esa vigrahavan dharma 
esa vedavidarii varah 
esa viryavatam frestho 
divyany astrany aijesena 
vedai ’sa Kucikdtmajah. 
devag ca na vidur yani 
kuto 'nye bhuvi manavah 



Here R. in the Bombay edition has in general the reading 
of M., but it omits the first verse and Kugikatmajah, while it 
has the late astran for astrani, with other variations : 
esa vigrahavan dharma esa vlryavatam varah 
esa vidya ’dhiko loke tapasacj ca parayanam 
eso 'stran vividhan vetti trailokye sacaracare 
nai ’nam 1 anyah puman vetti na ca vetsyanti kecana 
na deva na ’rsayah kecin na ’mara na ea raksasah 
gandharvayaksapra varah sakiihnaramahoragah 

Besides these parallels I have previously 2 compared the 
extended identity of H. 3, 60, 2 ff., and R. vi, 58, 24 ff. ; and 
three passages already noticed by others, where the great epic 
seems to have an older form, viz., i, 18, 13 and G. 1, 46, 21 ; 
iii, 9, 4 and R. ii, 74 (G. 76); i, 175 and R. i, 54 (compare 
Holtzmann, loc. cit.) Other parallels noticed by Holtzmann 
are: the creation, xii, 166 and R. ii, 110; Ganges, iii, 106 and 
R. i, 39 (later) ; Ilvala, iii, 96, 4, and R. iii, 11, 55 ; Rsyagriiga, 
iii, 110 and R. i, 19 (see now Luder’s essay) ; also a couple of 
passages in both later epics, origin of poem, i, 1, 57 and R. 
i, 2, 26; Skanda, xiii, 85 and R. i, 37, which approximate 
closely with i, 136, 1 and R. vii, 65, 10, and a few more less 
striking cases in both later epics. 3 

A review of these parallels, proverbs and tales, shows that 
whereas the former may be said to occur universally, in any 
part of either epic, of the latter (apart from the Rama tale 
itself), as far as formal identity goes, by far the greater part 
is found where either one or both versions occur in later addi- 
tions to the poem (R. i and vii, M. i and xii ff.), thus: 

M. R. M. R. 

i, 1, 57, and i, 2, 23 v, 13 and vii, 85 

i, 18 and i, 46 (G.) v, 141 and i, 2 

1 Here enam is astra(ganam) understood (?). 

2 A JP. xx, p. 34 ff. Holtzmann’s Das Mahabharata, already cited, both adds 
to and is complemented by the matter given there and here. 

3 I do not include parallel tales without parallel phraseology, as, for 
example, the allusion in xii, 57, 9, to the tale of Asamanjas told in iii, 107, 
39 ff. and in R. ii. 36, 19 ff. 



M. R. 

i, 66 and iii, 14 

i, 175 and i, 54 

(ii, 105 and ii, 100, Kaceit) 

iii, 9 and ii, 74 

iii, 58 and vii, 55 

iii, 96 and iii, 11 

iii, 106 and i, 39 

iii, 110 and i, 19 

iv, 70 and i, 24 (G.) 

M. R. 

xii, 127 and vii, 37 
xii, 153 and vii, 76 

xii, 166 and ii, 110 

xiii, 85 and i, 37 

H. R. 


iv, 40 
vi, 19 
vi, 44 
vi, 58 

That is, parallel tales are rare in the older, three times as 
frequent in the later books of each. The additions to one 
epic are thus on a par with the additions to the other in their 
mutual obligations. 1 This illustrates again the facts pre- 
viously observed in regard to the two epics by Jacobi and 
myself respectively, namely that the Uttarakanda has many 
tales of the middle district (Jacobi, R. p. 205), and that the 
early Mahabharata shows familiarity with the customs of the 
Punjab, while the didactic parts show no familiarity with 
the holy land, but all the numerous tales with scarcely an 
exception are laid in Kosala and Videha and on the banks of 
the lower Ganges (AJP., xix, p. 21). In other words, the 
two epics in their later development belong to the same 
locality and probably to about the same time. It is in this 
later development, then, that the two epics copy each other. 2 
The common tales that remain, apart from this phase of the 
poems, are few, and such as may be easily attributed to the 
general stock of legendary tradition. 

1 It must not be forgotten, however, that the Ramayana, apart from the 
first and last books, refers to episodes known only from the Mahabharata. 
For example, when SIta says she is as devoted to Rama “as Damayanti 
Bhaimi to Naisadha,” Naisadham DamayantI ’va Bhaimi patim anuvrata, 
R. v, 21, 12. Then when, ib. 34, 28-30, Rama is described as satyavadi, adi- 
tya iva tejasvi, and kandarpa iva murtiinan (all in one description, as in Kala), 
which is probably the borrower ? 

2 So the later G. agrees more closely with M. in many of the cases in 
Appendix A. But there is no uniformity in this regard, and R. has parallels 
enough to refute the idea that similarity is due solely to G.’s later copying. 



When we have peeled off the outer layer (and in it are 
included with one exception, if it be an exception, all the 
references to Valmlki in the great epic), we have left two 
epics, one of which is a complete whole, the other a congeries 
of incongruous stories grouped about a central tale ; both built 
on the same foundation of phrase and proverb and in part over 
the same ground of literary allusion ; both with heroes of the 
same type (whose similarity is striking) ; 1 and both arranged 
on the same general plan, a court-scene, where the plot is 
laid, a period of banishment in a forest-scene, followed by a city- 
scene , 2 where an ally is gained, and then by battle-scenes. One 
of these epics claims priority, but the claim after all is not 
that the great poet invented epic poetry, but that he first 
wrote an epic in §loka verse in a Kavya or artistic style. As 
the Ramayana is mainly in (jlokas of a more refined style than 
the Mahabhiirata and the Ivavya or artistic element is really 
much more pronounced, and as, further, it is highly probable 
that epic poetry was first written in the mixture of rougher 
eloka and tristubh characteristic of the Mahabharata, this 
claim, so stated, may in general be allowed, without impugning 
the relatively greater age of the other epic. 

Professor Jacobi admits that the metre of the Ramayana is 
more refined, but the explanation he gives is that it was a pro- 
duct of that East where poetic art was first developed. In a 
subsequent chapter I shall show that those parts of the great 
epic which from a metrical point of view agree most closely 
with the Ramayana are the later parts. Here I would merely 
raise the question whether the dictum that poetic art was re- 
fined in the East before the great epic arose, is not based on the 
style of the Ramayana alone ? Products of the same part of 
the country are Buddhistic and Upanishad verses, with which 
agrees the versification of the Mahabharata much more closely 

1 Not merely as being central figures. See for details the article by 
Professor Windiseh, cited in Pas Mahabharata iv, p. 68. The similarity of 
exploits is increased as we take the whole epics, which plainly have influ- 
enced each other in their final redaction. 

2 Owing to Rama’s oath he does not actually enter the city, but he finds 
his ally there, as do the Pandus at Virata’s town. 



than does that of the Ramayana. The Puranas also are eastern 
and their versification is in general rather that of the great 
epic. The distinction then is not sufficiently explained by 
geographical relations. On the other hand the metrical re- 
finement of U. the Upanishads, B. the early Bhilrata, B . 2 the 
late Bharata, R. the Ramayana, and K. Kalidasa is in the 
order U., B., B ., 2 R., K., with B 2 =R. in some cases, which 
looks to a progressive development . 1 

Another moot point in connection with tins geographical 
inquiry is whether the Ramayana was written by a poet who 
really knew anything about Ceylon, where Lanka, the seat of 
action in the Ramayana war, is usually supposed to be. Pro- 
fessor Jacobi has expressed the opinion that Lanka is not 
Ceylon, and that, further, Viilmlki did not know the littoral 
at all, but he was a riparian poet. Unless the allusions in the 
poem are all interpolations, I cannot accept this view. In the 
first place, the language of both poems on this point is identi- 
cal, the images are the same, and they are couched in the 
same words. If, then, they are all later additions to Valnnki’s 
poem, they must be copied from the Mahiibharata ; which opens 
a vista (of later Ramayana imitating an earlier epic) which 
Professor Jacobi would scarcely accept. But accepting some 
copying, there still remains enough sea-scape in the Ramaj'ana 
to show that no poet who did not know ocean could write as 
does Valmlki. In both texts, for example, occurs this splendid 
onomatopoetic description of the rising waves of full flood, 
which, as the poet repeatedly says, accompanies the filling of 
the moon : 

parvasu ’dlrnavegasya sagarasye ’va nihsvanah 

where the swell and filling and very hiss of the combing 
breakers is reproduced with a power that it is hard to ascribe 
to a riparian poet. But I must refer the reader to a special 

1 Valmiki’s work holds indisputable right to the title adikavya, or “first 
elegant poem,” a title which the great epic imitates in claiming to be a 
kavyam paramapujitam, “highly revered elegant poem,” to which claim it 
won a right after the more refined versification of the pseudo-epic had been 
added to it. 



paper on this subject for further illustration of our Yalxnlki’s 
intimate acquaintance with the sight and sound of ocean 1 — - 
or, if not our Yalmlki, to whom shall we assign the double 

Again, from the first dawn of critique it has been urged 
that widow-burning is not practised or known (as sometimes 
stated) in the Ramayana, but it is practised in the Mahablia- 
rata. Yes, in the first book and the twelfth and following 
books, just as conversely, in the Ramayana, the queens an- 
nounce that they are “ devoted ” and will die on the pyre with 
their husband ii, 6G, 12, or lament that being “not suttee” 
they “ five an evil life ” in not thus dying, v, 26, 7. Does this 
not imply widow-burning? And if it be said (with truth) 
that these are interpolations — well and good, but so are Adi 
and Canti interpolations. Both epics ignore the custom, 2 * * * 6 ex- 
cept hi their later form. 

One more observation is necessary in tins summary account 
of the mutual relations of the two epics. I have instanced 
the use of the word marisa in the Mahabharata as typical of 
influences not so often to be seen in the Ramayana. In the 
former, as a constant term of address, it is a link connecting 
this epic with the classical period; and yet it will not do 
to build too much on the fact that this fink is wanting in the 

1 AJP. vol. xxi, p. 378. Among the tributaries of Ayodhya are men- 

tioned the inhabitants of Malabar, and “sea-men,” in R. ii, 82, 8, where the 
senseless kevalali must be corrected to the reading of G. 88, 7, Keralah. 
The sea-men, samudrah, may be merchants or the name of a people. The 
Iveralas, or Malabar people, are here expressly “Southerners.” They are 
mentioned also among the lists of people in R. iv, 40 ff., which takes in 
the whole of India (41, 12, Pundras, Colas, Pandyas, Keralas) and mentions 
the Yavanas and other outer tribes: “Look among the Mlecehas, Pulindas, 
<?urasenas, Prasthalas, Bharatas, Kurus with Madrakas, Kamboja-Yavanas 

(empd.), and the towns, pattanani, of tjakas,” 43, 11-12 (compare M. yi, 87, 10). 
Also Yavadvlpa, R. iv, 40, 31, that is Java, is mentioned. I fail to see that the 
Ramayana, without such a priori excision as may also be applied to the Maha- 
bharata, shows less geographical knowledge or hearsay than does the latter 

1 Elsewhere in the epic, the widow is as much recognized as in Manu, who 
also knows no suttee. Compare Ruling Caste, pn. 172, 371, and a paper On 

the Hindu Custom of Dying to redress a Grievance, JAOS. xxi, p. 146 ft 




Ramayana. Such an example shows only that the Mahabha- 
rata has been in this instance retouched. S imil ar cases are 
found in the Ramayana, one of which I have already cited. 

For example, later Sanskrit poetry describes women 
adorned not only with the niipura or anklet (alluded to in 
both epics), but also with the kancl or gold girdle set off 
with bells. Probable as was the adornment in early times, 
this name for it does not occur in early literature, and so 
far as I know it does not occur in the great epic (frequently 
as women’s adornment is described) till the time of the 
pseudo-epic, where, xiii, 106, 56, and 107, 67 we find kahcinu- 
puragabda, just as we find the same collocation in R., for 
example, v, 4, 11; 18, 20; G. iii, 58, 26 (gugublie kaiicanl 
kanei) ; v, 12, 44. The later epics must have suffered tliis 
experience in many cases, another being offered just here by 
the use of the rare vallaki, xiii, 106, 49, and in vii, 6,665, but 
not here in B. 154, 25, where jharjhara takes its place. Just 
so in G. iv, 33, 26 is found this same vallaki (sic), but it is 
not found in the corresponding verse of R. iv, 33, 21. In 
sum, chance lateness of this sort is evidence only for the epic 
as we have it, tampered with by a thousand diadochoi. It can 
never show that one epic was produced before the other. So 
niryana for “death,” xv, 37, 40, is indicative of the age or 
origin of xv, 37, not of the Mahabliarata ; 1 of R. v (13, 41), 
but not of the epic as a whole. 

So, while we must admit that Valmlki’s mention of Kurus, 
Janamejaya, and Hastinapura, as against his non-mention of 
Pandus and Indraprastha, looks as if he knew not the latter, 
we must remember at the same time that Valmlki’s poem in 
turn has, quite apart from vocabulary, certain indications of 
an age not recognized by the poets of the latter epic, of which 
I will mention particularly two. 2 

1 Here, xv, 37, 43, tathagata seems to mean “ dead,” but it may be taken in 
its usual sense of “ in such a state,” as in K. ii, 109, 34, oddly near the Bud- 
dhist : yatha hi corah sa tatha hi buddhas tathagtam nastikam atra viddhi. 

2 Minor points of lateness (in either epic) are frequently apparent. Those 
in Mbh. are perhaps more common, but not in proportion to its extent. In 
B. may be noticed ships holding one hundred men each and palaces having 



The date of the Allahabad banyan cannot be carried back 
with any certainty to a very early date, though mentioned by 
Hwen Thsang. 1 Now the place where this tree ought to be 
is most elaborately described and praised in the great epic, 
iii, 85, 80 If., but the existence of such a tree is not even 
mentioned; whereas the other fig-tree at Gaya is praised as 
holy beyond words, for, in the epic interpretation of the 
modern aksay bai (bat), its fruit is imperishable. 2 This is 
particularly remarkable as in M. iii, 85, 65, (Jrngaverapur is 
especially famed as the place “ where Rama crossed.” But the 
Ramayana knows the Allahabad tree, ii, 55, 6 and 24. The 
mention of this tree at Prayaga, as against its non-mention in 
the Maliabharata, and the latter’s mention of Rama point to an 
earlier date for the Maliabharata Tlrtlia stories than for R. ii, 
55, and perhaps shows that at tliis time the Rama story was 
known, but not just as we have it. 

The word Sanskrit in its present meaning is found in the 
Ramayana but not in the Maliabharata. The bare statement, 
however, that the word Sanskrit in this sense is not found 
in an older period but occurs in the Ramayana, does not give 
quite all the facts. The great epic knows the word but only 
in its earlier meaning, “adorned,” “prepared,” asamskrtam 
abhivyaktam bhati, iii, 69, 8 ; samskrta and prakrta, 3 “ initiated 
and not initiated,” iii, 200, 88 (with priests who are suvedah 
and durvedah) ; samskrta mantrah, xiii, 93, 56. This is also 
the sense in R. iii, 11, 57, where bhrataram saniskrtam krtva 
itself (in M. iii, 96, 10, chagaih krtva susamskrtam) is joined 

(as in the drama) eight courts instead of three (as in the other epic), R. ii, 
84, 8; 57, 17 and 24; iv, 33, 19. 

1 Cunningham, Ancient Geography of India, p. 389. 

2 This, or “ makes the giver immortal,” is the epic interpretation, not (as 
now) that the tree itself is immortal. Compare iii, 84, 83, tatra ’ksayavato 
nama trisu lokesu viQrutah, tatra dattam pitrbhyas tu bhavaty aksaram 
ucyate. So in iii, 87, 11, and 95, 14 (with iii, 87, begins a recapitulation of 
Tirthas already mentioned) ; vii, 66, 20, where it is (vatah) aksayakaranah, as 
also in xiii, 88, 14. Here is found the proverb on Gaya, as in R. ii, 107, 13, 
with v. 1., and in M. iii, 84, 97, etc., as given in Spruch 1474 ff. 

3 As to this word in R., compare strfvakyam prakrtam jrutva, iii, 40, 5 
(asaram, comm.), with references in PW. s. v. 



with the preceding samskrtam vadan, the former in the Maha- 
bharata version being “ cooking ” (sariiskrtya = paktva) and 
the latter not used, which looks as if the Ramayana version 
were later. Several cases in the Ramayana do indeed show the 
older sense, but there are others, such as v, 30, IT, cited by 
Weber, and again by Muir, Original Sanskrit Texts, ii, p. 157, 
in which samskrta vak means Sanskrit, in that it is the “ culti- 
vated speech.” 1 In this case also the Ramayana is later than 
the Mahabharata, though the latter epic recognizes dialects, 
de§abliasas, iv, 10, 1 ; ix, 45, 103, etc., and seems (in its in- 
troduction) to use the expression bralinil vak or “ holy 
speech,” exactly in the sense of the Ramayana’ s samskrta vak. 
For in this instance a woman recognizes a king because his 
“ form and clothes are regal and his speech is the holy speech,” 
rajavad rupavesau te brahmliii vacam bibharsi ca, i, 81, 18. 
Rut these cases show only that when the Ilvala tale was re- 
written and the much adorned fifth book of the Ramayana 
was composed, samskrtam vad and samskrta vak were used 
nearly in the modern sense ; yet in showing this they indicate 
again that in our estimate as to the relative age of the epics 
nothing can be absolute or universal, but all must be stated 
relatively and partially. If it be said that this judgment 
lacks definitiveness, the reply is that it accords with the facts, 
which do not admit of sweeping statements. 2 

1 Also Jacobi, Ramayana, p. 115 (PW. s. sam-kar). Other cases show 
regard for grammatical nicety in the use of language (Jacobi, loc. cit.). 

2 For the metrical position of the two poems, see Chapter Four. I regret 
that Professor Jacobi’s long-expected book on the epics is not yet out, as it is 
sure to contain much valuable matter. As it is, I have had to rely, in citing 
his opinions, on the work cited above, and a review in the GGA., 1899, p. 
869 ff. 



Sukhad bahutaram duhkham jivite na ’tra samgayah, xii, 331, 16. 

“There is no doubt that there is more sorrow than joy in life.” 

Epic Systems. 

Ik the preceding chapters I have shown that from a syn- 
thetic point of view the epic as we have it, judged solely by 
the literature it recognizes, must be the product of a compara- 
tively late period. In this chapter it is my purpose to sketch 
as briefly as possible the salient features of the great systems 
of philosophy expounded in the later epic. To regard them 
as identical is impossible. To see in them a philosophic chaos, 
out of which are to arise future systems, is equally impossible. 
Some of them belong to the latest epic and they have their 
unity only in the fact that they are all colored by the domi- 
nant deistic view of an age that, having passed from pure 
idealism into dualism, sought to identify the spirit of man 
with that of a personal God and equate this god with the 
two separate factors of dualism; a dualism which was not 
that of spirit and matter but of conditioned being, conscious 
intelligence, as opposed to pure being or spirit (soul), con- 
scious intelligence being itself the only origin of matter, which 
is merely a form of mind. 1 

The importance of a review of this sort lies in the historical 
background it furnishes to the epic, which represents the last 
of six approved systems traceable in it: (1) Vedism or or- 
thodox Brahmanism ; (2) atmanism or Brahmaism (properly 

1 See on this point some pertinent remarks by Dr. Everett in the twentieth 
volume of the Journal of the A( >S., p. 309. It is a common error to speak of 
Samkhya dualism as setting spirit and matter in antithesis, whereas, accord- 
ing to the system, matter is only a development of self-consciousness. 



Brahmanism, but this term connotes a different idea), that is, 
an idealistic interpretation of life ; (8) Sariikhya, the dualism 
spoken of above; (4) Yoga, the deistic interpretation of Sarii- 
khya ; (5) Bhagavata or Pa§upata, different blit both sectarian 
interpretations of Yoga; (6) Vedanta or Illusion-idealism. 
Some of the epic writers support Sariikhya ; some, Yoga ; some, 
the sectarian interpretation; some, the Maya, Illusion-theory. 
Besides these are approved sporadically Vedism and Brahma- 
ism, not to speak of a number of theories not approved. 


In the Gita it is said, 4, 40 : “ The ignorant and unbelieving 
man who has a soul of doubt is destroyed; neither this world 
nor the next exists, 1 nor happiness, for him who has a soul of 
doubt.” The italicized words are those which, at xii, 133, 14, 
are put into the mouth of the Nastika, the negator or repu- 
diator of scripture, spirit, or duties. According to epic inter- 
pretation, one saying niisti, in refusing a gift to a priest, is a 
“ negator ” no less than he who refuses assent to the orthodox 
belief. But ordinarily Nastika is used in the latter sense and 
connotes a dissenter from received opinion in regard either to 
the existence of transcendental things or to the authority of 
hallowed tradition. 2 Such an unbeliever is threatened with a 
sudden enlightenment hereafter : “ If your opinion is that this 
world does not exist and that there is no world beyond, the 
devils in hell will soon change your ideas on that subject.” 3 
Any number of these unbelievers is known, who deny every- 
thing there is to deny. In ii, 31, 70, an unbelieving or heretic 

1 na ’yam loko 'sti na paro na sukharh sampayatmanah. Compare Katha 
L T p., ii, 0, ayam loko nasti para-iti main, punah punar vagam apadyate me 

- Xeglect of Yedic ordinances or denial of Veda is nastikva, par excel- 
lence, according to xii, 270, 67, and xii, 12, 5 (the latter) : vedavadapaviddhans 
tu tan viddhi bhrfanastikan (also anastika, ib. 4), for “rejecting the Veda 
a priest cannot attain heaven,” ib. 

3 Literally, will “ make you remember ; ” yad idam manvase, rajan, na ’yam 
asti kutah parah, pratismarayitaras tvarii Yamaduta Yamaksaye, xii, 150, 19. 



king is mentioned among those who pay tribute (in conjunc- 
tion with a tributary “ city of the Greeks ”) ; while in iii, 
191, 10, it is said that in the golden age to come there will be 
“ people of truth,” where previously had been established the 
schools of heretics ; from which it may be inferred perhaps that 
Buddhists or Jains are meant, as irreligious heretics would 
not have religious orders. 1 The Lokayata or Lokayatika 
(doubtful in i, 70, 46) is perhaps less a Buddhist (like Carvaka, 
who appears only as a pretended Brahman Parivraj, or priestly 
mendicant, and friend of the foe) than a devotee of natural 
science, as Professor Rhys Davids maintains. The doubter’s 
scriptures are not, however, referred to Brhaspati. The code 
of this ill-reputed sage, whom we have seen as a law-giver, is 
often enough alluded to, generally in connection with that of 
Uganas. The worst that is said of Brhaspati’s teaching is 
that it is drawn from a study of the female intellect, which is 
full of subtilty and deceit. But he is here only one of many 
authors of Arthagiistras, xiii, 89, 10. As a teacher he is ex- 
tolled. 2 Materialists and other heretics without special desig- 
nation appear to fill the whole land. Thus in xii, 19, 23, are 
mentioned rationalistic Pundits, hetuinantah, hard to convince, 
who are by nature befogged and stubborn, and deny the exist- 
ence (of a soul). These are opposed to those good men who 
are “ devoted to ceremonies and know the Purvagastra ” 
(nrimahsa?). “These fools,” it is added, “are despisers of 
immortality and talkers in assemblies of people ; they wander 
over the whole earth, being fond of speaking and learned in 
revelation.” 3 Others are cited to illustrate the unbelief that 
consists in a denial of the soul’s unity, ekantavyudasa. These 
believe in a sold possessed of desire and hate. An apparent 
allusion to Jains may be found in the description of the priest 
who “ tramped around Benares astounding the people, clothed 

1 acjramah sahapasandah sthitah satyajanah prajah (bhavisyanti). 

2 xii, 325, 23. His teaching in xiii, 113, is Buddhistic (5 = Dh. P. 132, and 7 
is like Dh. P. 420). On Lokayata, see Davids, p. 169 of op. cit. above, p. 55. 

3 vavaduka bahu^rutah. The denial in nai 1 tad asti must from the context 
refer to the existence of the soul. Tor anrtasya ’vamantarah in B. must, I 
think, be read amrtasya. 



in air, clothed like a madman ; ” 1 bnt we must be careful not 
to identify the characters of the epic too quickly with special 
names. This madman priest, for example, would seem to be 
rather a £ivaite Brahman than a Jain, and digvasas is applied 
to Vidura in his last state and to N ala in Ins distress. 2 In the 
same way, the brown and yellow robe does not necessarily refer 
to a Buddhist, any more than does the statement that one 
goes to heaven who builds a Vihura, xiii, 23, 99 ; for tlxese 
terms are common property. “ What makes you so glorious ? 
asks one woman of another, who replies : “ I did not wear the 
yellow robe, nor bark-garments, nor go shorn or with matted 
hair,” xiii, 123, 8. Here quite possibly Buddhists may be re- 
ferred to; but when I read that Civa’s devotees are of two 
sorts, householders, and those “ whose sign is tonsure and the 
yellow robe,” maundyaiix kasliya§ ca, xiii, 142, 22 ; and see 
that the yellow robe is also worn as a sign of grief, Nala, 24, 
9; R. vi. 125, 34, and that “the wearer of the yellow robe” 
is excluded from yraddha, xiii, 91, 43, I am by no means sure 
that even in the most tempting passage this robe indicates a 
Buddhist, unless, indeed, for some of these passages we may 
assume that £ivaite and Buddhist were already confused. But 
xii, 18, 32, “ those who cast off the Vedas and wander about as 
beggars shaved and wearing the yellow robe,” refers distinctly 
to Buddhists, as I opine. Similarly, the remark “ they that are 
budhas, enlightened, are devoted to Nirvana,” xii, 167, 46, may 
be put beside the buddhas of xii, 160, 33, who “ have no fear 
of return to this world and no dread of another ; ” but in the 
latter section, and in many others, “ enlightened,” budha and 
buddlia, refers to Brahmans; and Nirvana in epic teleology 
usually means bliss, for example the bliss of drinking when 
one is thirsty, or the bliss of heaven. 3 In short, we see here 

1 cankramlti difah sarva digvasa mohaynn prajali . . . unmattavesam 
bibhrat sa cankramlti yatliasukharii Varanasvam, xiv, 6, 18, and 22; com- 
pare 5, 6. 

2 To the author of Das JIbh. als Epos, etc., digvasas necessarily implies 
digambara (as Jain), p. 224. 

3 In the epic, nirvana is used in both of its later senses, bliss and extinc- 
tion, brahmanirvana, bliss of Brahman, like the nirvana, bliss, attained by 



and in a passage cited further on, that Buddhists are some- 
times referred to, but we must not call every beggar a Bud- 
dhist. The late passage xiv, 49, 8-12, shows that when the 
Anuglta was written, probably not before our era, these infi- 
dels were fairly rampant. The list of them is quite appalling 
and we may perhaps believe that the “ believer in nothing ” 
is a Buddhist and the “ shaven and naked ” mentioned in the 
same place is a Jain; while the svabhavam bliutacintakah are 
perhaps materialists. The “ course of right is varied ” and the 
view' of the author is here that of tolerance. Some of these 
philosophers deny a hereafter, some doubt all things, some 
hold the vyamigra doctrine of revolution (often mistranslated 
as evolution) of the universe, and according to the commen- 
tator some are adherents of the atomistic theory, bahutvam. 
Contests of these hetuvadins, rationalists, are not discounte- 
nanced, but enjoyed as a philosophic treat at the king’s court 
or at a great sacrifice, as in xiv, 85, 27, w'here “ talkative philo- 
sophers, eager to outdo each other, discussed many rational- 
istic arguments.” 

With all this liberality there is often no quarter given to 
the heretic, especially the Pasanda, 1 who appears to be pre- 
eminently a despiser of the Vedas. The reason is the natural 
one that he who despises the priest’s authority naturally de- 
spises the priest. “ The reason why I was born a jackal,” says 
a character in xii, 180, 47-48, “ is that I was a Punditkin, pan- 
ditaka, who was a rationalist, liaituka, and Plainer of the V edas, 
being devoted to logic and the useless science of reasoning (a 
telling phrase, repeated in xiii, 37, 12-14), a proelaimer of 
logical arguments, a talker in assemblies, a reviler and opposer 
of priests in arguments about Brahman, an unbeliever, a 
doubter of all, who thought myself a Pundit.” 2 The Pasanda 

drinking. On this subject much that is misleading has lately been published, 
owing to a false historical point of view. But the goal of extinction is also 
lauded. Thus, in xii, 242, 11-12, one attains to that where going he “grieves 
not, dies not, is not born, nor reborn, and exists not,” na vartate. 

1 v. 1. in xii, 218, 4; xiii, 23, 67 (other references in PW.) ; apparently a 
foreign or dialectic word ; especially Buddhists, according to N. 

2 akrosta ca 'bhivakta ca brahmavakyesu ca dvijan . . . murkhah pandi- 



and reviler of the Vedas are closely associated, as in xiii, 28, 
67, and 72, and like those who here “sell or wxite down the 
Vedas,” they go to hell. In short, any denial is usually per- 
mitted save the denial of the Vedas. The more surprising is 
it that elsewhere (see below) the Vedas are openly repudiated ; 
but this is only one of the inconsistencies with which the epic 


What then was authoritative? Characteristic of the con- 
tradictory views presented in the epic is the fact that in one 
place the very authority, pramanam, which is insisted upon 
as the only valid authority, is in another rejected as altogether 
delusive, and this not by heretics, but by the authors of the 
respective essays whose combined publications issued in one 
volume form the pot-pourri of the complete epic. 

The reason for this is obvious. Several forms of religion 
are advocated in the epic and each has its own test. Oldest 
and most widely represented is the biblical test. Over and 
over again we are assured that scripture is authoritative and 
those who will not accept scripture as the pramanam or test- 
stone of philosophy are damned. But beside these vigorous 
expressions of orthodoxy stands the new faith, which discards 
altogether the old scripture as an authority. For sacrifices 
and rites the Vedas are well enough ; they are there authori- 
tative. If one wishes to perform rites one must naturally 
go to the ritual. Such (jastrapramanya and vedapramanya 
rules, 1 admitting the necessity of rites at all, remain valid, 
simply because there are no others. But in all liigher matters, 
as for one who sees no use in rites, the scriptures are but a 
mass of contradictions. 2 

tamanikah (hence reborn, as a krostar). Compare Katha Up. ii, 5, sva- 
yamdhirah panditammanyamanah ; Mund. Up. i, 2, 8 ; Maitr. Up. vii, 9. The 
passage in Anu§asana cited above ia a repetition of all these epithets in 
characteristically free form. Compare, e. g., pi. 13, akrosta ca ’tivakta ca 
brahmananarh sadai ’va hi (here panditamani). 

1 xiii, 84, 20, and 37. 

2 One of the minor epic contradictions is that referred to above, p. 46, in 
regard to the “ two brahmans.” The orthodox, but not too liberal man, says : 



The old view is best represented in the saying that Veda, 
Dharmagastras, and acara, custom, are the recognized author- 
ities in every matter, as in iii, 207, 83; xiii, 84, 20, and 37. 
The confused rule of the Veda is referred to in xii, 19, 1-2: 
“ I know the highest and other t-Tistras and the double injunc- 
tion of the Veda, ‘Do acts and abandon them.’” “Untrue, 
according to casuistic reasoning, is the word of the V eda — 
but why should the Veda speak untruth?” says Vyasa, xiii, 
120, 9, when inculcating the late notion that a small gift is as 
efficient as a great sacrifice in procuring salvation, a theory 
that is certainly untrue in the light of the Veda. “ Logic 
has no basis, the scriptures are divided ; there is not one seer 
whose opinion is authoritative,” pramanam. “ The truth about 
right is hidden in a cave ; the only path is that pursued by 
the majority,” iii, 313, 117. 1 “Deceitful is the Veda,” it is 
said in xii, 329, 6. Both scripture and argument, tarka, are 
useless in comparison with the enlightening grace of God, 
which alone can illuminate the “mysterious hidden communi- 
cation of truth,” xii, 335, 5. Such holy mysteries must, 
indeed, be kept from those who are “burned with books of 
philosophy,” tarkagastradagdha, xii, 247, 18. 

In the matter of the Veda, the new faith discounts its 
value by setting beside it the recent books of later cult, 
exactly as modern sects take as authoritative their own scrip- 
tures. Bhisma’s words, being inspired by Krishna, are “as 
authoritative as the words of the Veda,” vedapravada iva 
(pramanam), xii, 54, 29-30, and Veda, Purana, and Itiliasa are 
all reckoned as authoritative in xii, 343, 20. But the Gita is 
the only authority of the Bhagavatas, Gita, 16, 24. Compare 
also the tirade in xiii, 163, 2-9: “Immediate perception or 
biblical authority, againa, what is convincing proof, karana, 

dve brahman! veditavye gabdabrahma param ca yat, fabdabralimani nisndtah 
pararn brahma ’dhigaechati, xii, 233, 30, “when one is thoroughly conversant 
with the Veda he attains to Brahman bnt the devotee ‘‘even by desire of 
wisdom surpasses the Veda,” api jijnasamano 'pi fabdabrahma ’tivartate, ib. 
237, 8, 

1 mahajana, if this be the meaning here ; apparently only usage is meant : 
mahajano yena gatah sa panthah. 



in these ? Answer : “ There is many a text to increase doubt. 
Rationalists say that perception is the only proof. They are 
children who think themselves wise and believe only in 
denial, na ’sti. Recourse to ‘cause’ amounts to nothing.’’ 
But though philosophy is really interwoven with religion, we 
may leave for the present the Bliagavatas and Civaites to 
their religion which is “ freed from philosophy,” xiii, 14, 198, 
and consists in identifying the All-god with their special 
gods (viii, 33, 51 “ one God of various forms ”), to consider 
the more strictly philosophic view of authority. 

Only one view is held by the real philosopher : “ Through 
inference we learn the truth.” 1 Traditional wisdom, amnaya, 
as was shown above, is not always recognized, though it is 
generally admitted. “ In amnaya are established the Vedas; 
from amnaya come the Vedas. 2 . . . Universal opinion says that 
an amnaya-declaration is truth, and there is no authority at 
all, gastrata, when that which is not authoritative is allowed 
to stand against the recognized authority of the Vedas,” xii, 
269, 33; 261, 9-10. Thus “inference together with scrip- 
ture,” anumana and grata, are the two most substantial tests 
of truth, xii, 205, 19 and 210, 23, lietvagama ; for “ all that is 
Vedic is the word of God,” xii, 269, 10. 3 

The third authority is the one scorned above, perception, 
pratyaksa (xiv, 28, 18, pratyaksatah siidhayaniah, and often, 
as cited below in the course of this chapter). In the mystic 
religion of the Yogin this pratyaksa becomes the intuitive 
insight of the seer and is the only test of truth, answering 
to “ second sight.” 4 The Harivahga inveighs against the 
“doubters and curious speculators” who accept any authority 
save faith, 3, 4, 8 ff. 

1 anumanad vijanlmah purusam, xiv, 48, 6; xii, 206, 23. 

2 The commentator becomes confused, and rendering amnaya by Veda 
renders vedah by smrtayah ! 

3 sarvam arsam vyahrtam viditatmanah (= paramepvarasya). The com- 
mentator cites Bril. Up. ii, 4, 10, nihgvasitam, in support of plenary inspiration 
as here inculcated. 

4 The curious result is thus reached that the crassest materialist and 
most exalted mystic reject all proofs save pratyaksa. Only one means by 
“autopsy” (physical) perception and the other means insight. 



Besides these three, to wit, biblical authority, inference, 
and direct observation, the fourth “ proof by analogy ” may 
be implied in the late conversation of Draupadi, where, after 
a passing reference to the arsam pramanam and pratyaksa, is 
added “ and thy own birth is the proof by analogy,” uparnii- 
nam, iii, 31, 11-33. Elsewhere the epic stands philosophi- 
cally on the Samkhy-yoga basis of three reliable proofs only. 

This result is fully borne out by the terminology. The 
Vedanta philosophy of the epic is not called by that name. 
Nyaya may possibly be known, but it is doubtful whether the 
word ever refers to the system, or the system, except perhaps 
in one or two late passages, is ever recognized. A brief sur- 
vey of the facts will make this clearer. 


If the philosophical system were known as such the use of 
the name would occur as such. But Vedanta seems every- 
where to mean Upanishads or what is the same thing, Aran- 
yakas. 1 No Vedanta system is alluded to, Vedanta may refer 
to Sariikhya in xii, 196, 7 (where it takes the place of the 
latter in antithesis to Yoga, as the commentator thinks), but 
the word more naturally means the teaching of the Upan- 
ishads, as usual. 2 The passages cited above in the chapter on 
literature exhibit the characteristic usage. Thus hi Gita 15, 
15, vedantakrd vedavid eva ca’ ham, where Telang rightly 
takes the reference to be to the Aranyakas. So in viii, 90, 
114, vedanta vabkrtliaplutah, where Kama appeals to Arjuna 

1 So, for example, in yad uktam vedavadesu gahanam vedadarfibhih, 
tadantesu yatha yuktam krama(karma)yogena laksyate, xii, 233, 28 (= tad 
uktam vedavadesu . . . vedantesu punar vyaktam, 230, 11), a mystery (viz., 
gambhlram gahanam brahma, 224, 48). 

2 samkhyayogau tu yav uktau munibhir moksadar?ibhih, sannyasa eva 
vedante vartate japanam prati, vedavada^ ea nirvrttah fanta brahmany 
avasthitah, three hemistichs, of which the first is repeated in the next yloka, 
where alone it seems to belong. Conversely, in Gita 18, 13, the word Sarii- 
khya is taken by the commentator to mean Vedanta, because here we have a 
grouping of five karmahetavah not recognized in Sariikhya. It may be said 
once for all that the commentator is often useless in philosophical sections, 
as he wishes to convert Samkhya into Vedanta on all occasions. 



to observe the law of fighting, since the latter knows the law 
of fighting and is thoroughly acquainted with the holy scrip- 
tures, i. e., he is a moral man (not a Vedanta philosopher). 
So in ii, 53, 1, kings who are declarers of all the Vedas and 
versed in the Vedanta, paryaptavidya vaktaro vedantava- 
bhrth&plutah. Durga is Savitri, vedamata tatha vedanta 
ucyate, “mother of the Vedas and famed (not in philosophy 
but) in the Upanishads,” vi, 23, 12. A Gandharva is “wise 
in the knowledge of Vedanta,” xii, 319, 27, and asks ques- 
tions about Veda and logic, which are answered in Samkhya 
terms (vedya is purusa, for example). The priest who at 
xii, 349, 56 is said to transmit the knowledge of the Gita, 
knows the Jyestha Saman and the Vedanta; and lie who 
knows the names of Vishnu is Vedanta-learned, xiii, 149, 123. 
Again in xiv, 13, 15: “Whoso would kill me (Kama) by 
vedair vedantasadhanaih, power derived from the mysteries 
of the Veda.” I know in fact only two passages where, per- 
haps, Vedanta might be fairly taken as referring to the phil- 
osophy. One of these is in a tristubh verse which has been 
interpolated (out of all syntactical connection) in xiii, 69, 20, 
and even here, late as is the verse, it is perhaps more prob- 
able that the word is to be taken in its usual sense. 1 The 
other is found at xii, 302, 71, where the “island of Vedanta” 
is a refuge to the saints. The “Secret of the Vedanta” cited 
below is clearly “Upanishads.” The Brahma Sutra I have 
spoken of above, p. 16. 

Mlmansa does not occur as the name of a philosophical 
system. I have referred to the Purva^astravids above, but 
the word is obviously too general to make much of, though 
it is used as if it applied to the Purva-mimansa, for the Pur- 
va§astravidah are here, xii, 19, 22, nirata nityam dime 
yajne ca karmani. This implication is not absolutely neces- 
sary, however. The old name for the system, Nyaya, does 
not seem to be used in the sense of Purvamlmansa. 

1 vedantanisthasva bahugrutasya, supposed to be governed by vrttim 
(dvijaya) ’tisrjeta (tasmai) in the next stanza! 




The argumentative group of five, explained according to 
the padartha in xii, 321, 80 ff., consists of sauksmya, sariikh- 
yakramau, nirnaya, and prayojana, wliich recall, especially in 
the definition of the last, the corresponding section in the 
formal Nyaya. The epic gives the following definitions : 

1. Sauksmya, subtilty, is where knowledge, in respect to 
objects of knowledge which are divided, comes from distinc- 
tion and the intellect rests (on this distinction). 

2. Samkliya or samkliya, reckoning, is reckoning the value 
of weak and valid points and arriving at some conclusion. 

3. Krama, order: when it is decided wliich should be said 
first and which last, they call that kramayoga, the application 
of proper sequence in an argument. 

4. Nirnaya, ascertainment, is a conclusion that the case is 
so and so, in cases of duty, desire, gain, emancipation, after 
recognizing them according to their differences. 

5. Prayojana, motive : where inclination is produced by ills 
arising from desire or dislike and a certain conduct is followed, 
that is motive. 

As has been remarked by Mr. K. Mohan Ganguli in his 
translation, this final definition of prayojana is almost identical 
with that given by Gautama i, 24, yam artham adhikrtya 
pravartate tat prayojanam : “ If one sets an object before one’s 
self and acts accordingly, that is motive.” So the epic, 
prakarso yatra jayate, tatra ya vrttis tat prayojanam, as ren- 
dered above. Similarly, the epic definition of nirnaya is like 
that of Gautama in i, 40 : “ The conclusion reached after hear- 
ing what can be said for and against (on both sides) after 
doubting.” The other members of Gautama's syllogism, i, 
32, seem to have no connection with the above. The speech 
to be delivered, it is declared in this passage of the epic, must 
be nyayavrttam (as well as reasonable, not casuistical, etc., 
sixteen attributes in all). 1 

1 No explanation is given of the eighteen merits with which the speaker 
begins. The sixteen attributes may be compared (numerically) with the 
sixteen categories of the Nyaya. 



We may compare further in the late list of Pundits at i, 
70, 42, those with nyayatattvatmavijiiana, possibly “versed in 
psychology according to the Nyaya-tattva ; ” and i, 1, 67, 
nyaya§iksa, Nyaya-system, opposed to Vedadliyatma but also to 
cikitsa, etc. Also xii, 19, 18, referred to above, p. 87: “ Some, 
rejecting unity, attribute to the atman desire and dislike,” a 
Nyaya view. Finally, in xii, 210, 22, nyayatantrany anekani 
(declared by various people), “ systems of logic,” is typical of 
all remaining cases. Nyaya, then, usually means logic, but 
occasionally, in the pseudo-epic, the special Logic-system 
known to us as Nyaya. 1 


This word is used as an adjective, of gunas, etc., in the 
sense of excellent; but the system is unknown in the main 
epic though it is referred to in the passage cited above, in 
i, 70, 43-44, and also in ii, 5, 5 (viikya) pancavayavayukta, 
another proof of the lateness of the Kaccit section, 2 wdiether 
the five avayavas here mentioned be terms implying Nyaya or 
Vakjesika. Kanada’s name appears fust in the Iiarivainja (see 
below, p. 98, and above, p. 89). 

The Four Philosophies. 

In xii, 350, 64 if. (compare 350, 1, pracaranti) it is said 
that there are four current philosophies, jiianani, the Sarii- 
khyayoga, Pancaratra, Vedaranyaka (or Vedah), and Pa§u- 
pata. Kapila declared the Saiiikhya; Hiranyagarbha, the 

1 For the ordinary use, compare tais tair nyayaih, such arguments, passim. 
All speculation is Tarka. Compare the remarkable statement, xii, 15, 26: 
“There are minute creatures whose existence can be argued by tarka (so 
small that) an eyelid's fall would be the death of a number of them.” 

2 The former passage, after mentioning those endowed with nyayatattva- 
tmavijnanaaddsnanavakyasamaharasamavayavifaradaih, vifesakaryavidbhij 
ca . . . sthapanaksepasiddhantaparamarthajnatam gataih . . . karyakarana- 
yedibhih, which may refer to either system. The passages have been cited 
by the author of Das Mahabharata als Epos, etc., p. 226, who admits that the 
five “avayas,” as he call them twice, imply the Vaifesika system. 



Yoga; 1 Apantaratamas is called the Teacher of the Yedas 
(“ termed by some Pracinagarbha ”) ; 9 iva declared the Piitju- 
pata religion ; Vishnu, the whole Pancaratra. “ In all these 
philosophies Vishnu is the nistha, or chief thing.” 2 

Kapila and his System. 

Although it is said, as quoted above, that there is no seer 
whose authority is authoritative, this is merely a teaching of 
temporary despair. Kapila is authoritative in all philosophical 
matters and Ills name covers every sort of doctrine. He is in 
fact the only founder of a philosophical system known to 
the epic. Other names of founders are either those of mere 
gods or disciples of Kapila. Badarayana and Patanjali 3 are 
unknowm even as names, and Jaimini and Gautama appear only 
as sages, not as leaders of speculation, yandilya (otherwise 
said to be known in the epic) is respectfully cited on Yoga, 
not as founder but as recommending Yoga concentration. 4 As 

1 See the note on this verse just below. As Yoga-teacher of Daityas. Qukra 
is mentioned, i, 66, 43. Both Vishnu and £iva are credited with being Yoga- 
lords (loc. cit. by Holtzmann, Das Mbh. im Osten und Westen, p. 110). 

a In the Vasudeva religious philosophy of Krishnaism, as expounded in 
xii, 345, 7 ff., some people, after death, become paramanubhutas, very fine 
sprites, and enter Aniruddha ; then as manobhutas, or mental entities, they 
enter Pradyumna; thence they go to Jiva (Samkarsana). Such people are 
“the best priests and Saiiikhyas and Bhagavatas.” Finally, devoid of all 
unspiritual constituents, traigunyahlna, they enter Paramatman (Ksetrajna, 
nirgunatmaka), or Vasudeva. These are the four forms of God. The name 
of God is immaterial. Budra and Vishnu are one being, sattvam ekam, 
divided in two, xii, 342, 27 (they are synonyms like brhad brahma and mahat, 
337 2, paryayavacakah yabdah ; Vishnu may be called (,'iva and Brahman 
may be called Intellect). 

3 In the Sarvadaryanasamgraha it is said that Patanjali made (atha yoga- 
nuyasanam, i, 1) an anuyasana, or secondary collection (as ana is explained) 
based on earlier Puranic materials. The verse attributed in this connection 
to the Yajnavalkya Smrti (158, 17 ; p. 239 of Cowell’s translation) has caused 
the Petersburg Lexicon to postulate, s. v., another Smrti of the same name. 

I think it is a mere lapsus for Vyasa’s Smrti, for the verse cited (“ Hiranya- 
garbha, and no other ancient, is the declarer of Yoga”) occurs xii, 350,65. 
It has occurred to me that this verse might imply Patanjali, and the “ no 
other” be a distinct refutation of his claim, the epic preferring divine 
authority ; but this is perhaps too pregnant. 

4 prthagbhutesu srstesu caturthayramakarmasu samadhau yogam evai- 
’tac (maduktam vakyam) chandilyah yamam abravit, xii, 254, 14. 



a teacher of unconditioned Brahman, Atreya is lauded in xiii, 
137, 3; and in xii, 319, 59, a list of teachers of the twenty- 
fifth (spiritual) principle is given as having instructed the 
Gandharva Vigvavasu: Jaiglsavya, Asita Devala, Paragara, 
Varsaganya, Bhrgu, Pancagikha, Kapila, (,'uka, Gautama, 
Arstisena, Garga, Narada, Asuri, Pulastya, Sanatkumara, 
£ukra, Kagyapa, seventeen mixed gods, saints, and philoso- 
phers, of whom two are important besides Kapila, n am ely 
Asuri and Pancagikha, his pupils ; while one system (explained 
below) is referred also to Asita Devala. 

There seems to be no reason to doubt that Kapila was a 
real (human) philosopher, and not a mere shadow of a divin- 
ity. The fact that his name is also given to divinities proves 
the opposite as little as does his deification, for it is customary 
to deify sages and for divinities to have sages’ names. A per- 
fect parallel to the use of Kapila in this way is afforded by 
Kanada, which, as far as I know, occurs first as an epithet of 
<piva as supreme god, in the Harivanga 3, 85, 15-16 : 

yam ahur agryam purusam mahantam 
puratanaiii sariikhyanibaddhadrstayah 
yasyS ’pi devasya gunan samagraiis 
tattvang caturvingatim ahur eke 
yam ahur ekam purusam puratanam 
Kanada-namanam. a jam mahegraram 
daksasya yajfiarii vinihatya yo vai 
vinagya devan asuran sanatanah 

Kapila’s treatise is repeatedly declared to be oldest, but he 
is not only the oldest, he is the supreme seer, identical with 
Agni, with <yiva also, and with Vishnu. He is said to have 
got his wisdom from (Jiva. 1 

1 “ Of the treatises declared by metaphysicians that by Kapila is the ear- 
liest,” xii, 351, 6 ; agnih sa Kapilo nama, samkhyayogapravartakah, iii, 221, 21. 
Hall gives a later v. 1., samkhyafastrapravartakah, Samkhyasara, p. 18, where 
most of the epic allusions are collected. As supreme seer, xii, 350, 05 ; gira, 
xii, 285, 114, where the commentator interprets Samkhya as Vedanta (as 
often) ; xiii, 17, 98, and xiii, 14, 323, giva as kapila. Kapila is identified with 
Vishnu in iii, 47, 18; Gita, 10, 26, etc.; with Prajapati in xii, 218, 9-10, where 



I have noticed only one passage, xii, 269, 9, where Kapila 
is presented in the light of adverse criticism from the point 
of view of orthodox Brahmanism. On seeing a cow led 
out for sacrifice, Kapila, filled with compassion, cried out 
O ye Vedas! an exclamation of reproof against the Vedas, 
as inculcating cruelty to animals. At this he was attacked 
by the inspired cow with a long discourse, challenging 
him to show why the Vedas should be regarded as authori- 
tative in any regard, if not in regard to the slaughter of 

Kapila appears in this tale as a teacher of unorthodox 
non-injury and maintains to the end (so that his view is 
presented as really correct) that not the sacrifice of animals 
but the “ sacrifice (worship) of knowledge ” is the best. 
Elsewhere also we find the same antithesis between the old 
orthodoxy and the new science of thought, which not only 
disregards Vedic ceremonies but condemns them (xiv, 28, 
7 ff.). 

The best evidence of the authority of Kapila is given not 
by express statement but by implication in the praise of other 
systems, which, an important point, are by the same implica- 
tion looked upon as distinct from that of Kapila, although his 
name is used to uphold them. Thus Kapila’s own system is 
called generally the Samkhyayoga, or specifically the Kapi- 
lam. 1 The Saihkhyayogins are said to be the models even in 
teaching of other tendency, as in xii, 347, 22, and nothing 
better can be said of the Bhagavatas, here extolled, than that 
their system is “equal to the Samkhyayoga,” not, be it 

he is called the supreme seer, incorporate in Paficafiklia (the first pupil of 
Asuri, who in turn was a pupil of Kapila). In xii, 337, 8, Kapila is £alihotra- 
pita smrtah, father of £alihotra, the veterinary sage (above, p. 12). Kapilah 
praha: pritaf ca Bhagavan jnanam dadau mama bhavantakam, xiii, 18, 4. 
The Harivahfa, 3, 14, 4, and 20, speaks of Kapila as the “teacher of Yoga, 
the teacher of Sariikhya, full of wisdom, clothed in Brahman, the lord of 
ascetics.” Compare the supreme spirit as Kapila, xii, 340, 68. 

1 “lie learned the whole Yoga-sastram and the Kapilam,” xii, 326, 4; 
Virinca iti yat proktam Kapilam jnanacintakaih sa Prajapatis eva ’ham, xii, 
343, 94 (Kapila, 95). Also Samkhya krtanta, Gita, 18, 13. 



observed, the same, but as good as the system of Kapila. 1 
Amid a list of heroes in xiii, 75, 24-25, we find placed beside 
battle-heroes, gift-heroes, moral-heroes, etc., only Samkhya 
and Yoga heroes, enrolled to represent philosophy. 2 As be- 
tween the two, the implication contained in the words at 
Gita 5, 5, “ the Yoga gets as good a place as the Samkhya,” is 
that it is the Samkhya which is the norm. Samkhya is cited 
alone as the one system of salvation in i, 75, 7 : “ Salvation he 
studied, the unequalled system of Samkhya.” In contrast 
with Veda and Vedanga, it is the one type of philosophy: 
“He became learned in the Atharva Veda and the Veda, in 
the ritual also, and a past-master in astronomy, taking the 
greatest pleasure in Samkhya,” xiii, 10, 37 ; “Vedas, Angas, 
Samkhya, and Purana,” xiii, 22, 12. 

The two systems are often separated. Yogapradarginah 
stands parallel to Samkhyanatlarginah, xii, 314, 3-4. “The 
rules both of Samkhya and Yoga” are mentioned, xii, 50, 
33. Narada “knew the difference between Samkhya and 
Yoga,” ii, 5, 7 . yaunaka is “ rapt with metaphysics, adhyatma, 
skilled in Yoga and in Samkhya,” iii, 2, 15. The difference is 
explained in the Gita as: “The double point of view, nistha, 
of the Samkhyas, who have jftanayoga; of the Yogins, who 
have karmayoga. Sometimes Samkhyajuana on the one hand 
is opposed to Yoga alone on the other, xii, 315, 18.® Some- 
times the 9astra is that of the Yoga, as opposed to jilana of 
the Samkhya, xii, 319, 67 ; yogagastresu, 340, 69, etc. Never- 
theless, they are, says the Gita, essentially one system. And 
so often we find that V edic practices and the existence of God 
are claimed for Samkhya and Yoga, as if they were one system. 
The same is true of the practice of austerities or asceticism. 
“ The many names of God are declared in the Rig Veda with 

1 Samkhyayogena tulyo hi dharma ekantasevitah, xii, 349, 74. 

2 So in viii, 33, 49, Yoga and Samkhya (atmanah) represent philosophy. 

3 Compare xiii, 149, 139: yogo jnanam tatha samkhyam vidyah pilpadi- 
karma ca. In the passage cited above, the interesting aristani tattvani are 
grouped with yoga and samkhyajnana (as objects of research). They are 
explained elsewhere, xii, 318, 8, as “signs of death,” appearing to one if he 
cannot see the pole-star or his reflection in another’s eye, etc. 



the Yajur Veda, in Atharva (and) Samans, in Purana with 
Upanishads, in astronomy also, in Samkhya and in Yoga- 
castra, and in Ayur Veda,” to give the bizarre group of xii, 
342, 8. “Both gods and demons practise austerity, tapas, 
which has been argued out, yuktitah, of Veda and Sam- 
khyayoga,” xii, 285, 192. 1 

Samkhya and Yoga. 

But it must be noticed that the claim for the identity of 
Samkhya and Yoga comes from the Yoga side, which is deistic 
and seeks to make the Samkhya so, exactly in the way the 
Vedanta commentator seeks to make the Yoga passages Ve- 
dantic. The distinctive mark of the Yoga, as given above 
from the Gita, 3, 3, is, if we translate it in the natural original 
sense, application to work as opposed to application to under- 
standing; in other words the Yoga laid stress on religious 
practices, the Samkhya on knowledge. 2 It may be that Yoga 
also, like Samkhya, was originally atheistic and that deistic 
Yoga was a special development. Notliing could be falser, 
however, than the supposition that the Yoga and Samkhya 
differ only in method, or the epic assumption that both are a 
sort of Vedanta inculcating belief in Brahman as the All-soul. 
Even the Gita recognizes the distinction between the two 
schools in saying that the system that recognizes the All-soul 
(“ one entity eternal, undivided, in all divided existences ”) 
is better than the one that recognizes “ separate and distinct 
entities in all existent beings,” 18, 21-22, clearly referring 
to the fundamental difference between Brahmaism 3 and Sam- 

1 It may be observed of the terminology that as Yoga means Yogin as well 
as the system, so Samkhya means system or a philosopher of that system. 
Typical of the pseudo-epic is the circumstance that here Samkhyayogau are 
personified as two beings along with Narada and Durvasas, xiii, 151, 45. 

2 Compare the use in xiii, 84, 40, where it is asked : kena va karmayogena 
pradanene ’ha kena va (can I be purified), i. e., “ by application to holy works.” 
Compare krsiyoga, xiii, 83, 18. 

3 As Vedanta is commonly used of pamkara’s interpretation, I employ 
Brahmaism to connote a belief in the All-soul without necessarily implying 
a concomitant doctrine of Illusion, Maya. 



khyaism. The practical difference is that formulated at xii, 
317, 2 ff., where it is said : “ There is no knowledge like the 
Samkhya, no power like the Yoga; these are both one in 
practice, ekacaryau, because both destroy death. Foolish 
people regard them as distinct, but we recognize them as one. 
What the Yogas see is seen by Sarhkhyas ; who sees Samkhya 
and Yoga as one sees truly,” a passage copied from the Gita, 
5, 4-5, and repeated with varied readings in xii, 306, 19. 

Though the pseudo-epic is so like the Gita, its relative late- 
ness, I may observe in passing, is shown inter alia by the use 
in this passage of yogam as a neuter noun, xii, 317, 27, etad 
dhi yogam yoganam, 1 as in xiii, 17, 19; one of the many little 
points ignored in the unhistorieal synthetical method. 

This passage, in its admission under cover of fools’ opinion, 
shows clearly that the two systems could be regarded as iden- 
tical only by insisting on the objective of each. Botli sys- 
tems gave emancipation, therefore they were one. But one 
way was that of pure science or knowledge, the other was 
that of pious work (yoga, tapas) added to this science, a practi- 
cal divergence that existed quite apart from the question 
whether the goal was really the same. 

But the epic in other passages, despite its brave pretence, 
is not content with Samkhya science or even with Yoga work. 
On the contrary, the religious devotees named above throw 
over both systems. It is true they keep the name, just as 
these philosophical systems themselves pretend to depend on 
the Vedas, or as European philosophers used to claim that 
their systems were based on orthodoxy. But this only shows 
how important and fully established were these philosophi- 
cal systems when the sects arose that based salvation on 
faith and the grace of a man-god, while still pretending to 
philosophy. They could not unite, for the true Samkhya did 
not teach Brahmaism, but kevalatvam, or absolute separation 
of the individual spirit from everything else, an astitvam 
kevalam, or existence apart from all, not apart in Brahman. 

1 Repeating yoga esa hi yoganaru in 307, 25. 



No less irreconcilable with the earlier belief is the later 
sectary’s view of action, pravrtti, as due to God. For the older 
sage was intent on escaping action, which the system regards 
as due not to spirit but to the inherent quality of its antithe- 
sis, Prakrti. But in the religious substitution of a personal 
Lord, I c; vara, as synonymous with the Supreme, it is taught 
that “ the Lord created pravrtti as a picturesque effect ” (after 
electing nivrtti for himself) ! 1 Here the roots of the Karma 
doctrine are cut by the new faith of the quasi monotheism 
which is reflected in the later pseudo-epic. 2 

Fate and Free-Will. 

Another side of speculation presents a varied field of belief. 
Is there such a thing as free-will? The later epic fixes 
responsibility in turn on the Lord, man himself, purusha, luck, 
hatha, and Karma, xii, 32, 12, ff. ; where Karma is finally rec- 
ognized as the only agent, as otherwise God would be re- 
sponsible for sin ; and if man were the sole agent there could 
be none higher than man. As luck would absolve a man, 
only Ivanna is left, associated with Time in a sort of dual 
fatalism, kamiasutratmaka. Obviously Fate, ‘as Time is here, 
really undermines the theory of Karma quite as much as does 
the interposition of the Lord or any other foreign factor. So 
in xii, 224, 16 ff. and 226, 13 and 21 ff., we find first the re- 
flex of the Upanishads and Gita, “he who (in imagination) 
slays and he who is slain are both ignorant," and then : “ The 
deed causes the deed ; but the deed has another creator, F ate, 
Time. Fate or what will he will he is the cause.” “Sorrow 
lies in thinking ‘ I am responsible ' ; for I do that which the 
ordainers ordained when I was born." 3 

1 pravrttidharman vidadhe krtva lokasya citratam, xii, 341, 99. 

2 This is the “fourfold God,” worshipped by the Ekantins as bavin* one, 
two, three, or four forms, identified with Krishna, his son, grandson, and 
brother, as named above, p. 97. lie is maker and non-maker, and takes 
Prakrti’s function in “sporting:” yathe ’cchati tatha rajan kridate puruso 

3 So 224, 31 ; 226, 8 ; 227, 34 and 35 : kalah pacati . . . kalah kalayati pra- 
jah; 226, 12: “Whatever state one obtains he must say bhavitavyam,” “it"' 
was fated,” i. e., independently of Karma. For kala from kal, cf. Gita, 10, 30. 



Elsewhere Fate is the Divine power, daiva, opposed to 
human effort and to nature, svabhava, the latter having the 
implication of the Karma doctrine. Each of these factors is 
upheld by one or another theorist, while others claim that they 
all work together, xii, 233, 19, repeated at 239, 4—5. In other 
places the same Fate that is elsewhere made responsible is 
scorned, daivaiii kllba upasate, “only eunuchs worship Fate ; ” 
and “there is no Fate, all depends on one’s own nature;” 
the Karma doctrine, svabhavatah, xii, 139, 82 ; 291, 13. 1 

Samkhya is Atheistic. 

In the “ one-soul ” doctrine just referred to, God himself is 
energy, karyatman, the soul of all, the saviour, “the Light 
which Yogins see,” the Ego, eternal, without characteristics 
of any sort, aham ca nirgunah, xii, 47, 54, 63, 69-70; xiv, 
25, 7. He exists “alone with wisdom,” till he makes the 
worlds, each succeeding mon, xii, 340, 71-72, just as sunrise 
and sunset follow each other, ib. 75. On the other hand, the 
epic declares with all plainness that the Samkhya system is 
devoid of a belief in a personal supreme God. In xii, 301, 1 ff., 
the question is raised, What is the difference between Sam- 
khya and Yoga? The answer is: “Saihkliyas praise the Sam- 

1 According to xii, 239, 20, Time is the origin and controller of all things, 
prabhavah . . . samyamo yamah, and all things produced by duality exist 
according to their own nature, svabhavena. The nature of the individual 
spirit is often rendered by this word, as such a spirit is conditioned by its 
former acts. Below is cited a case where it is a factor of the body, distinct 
from organs, mind, and spirit. An interesting critique of heretics leads up to 
xii, 238, 3 ff. (where the word connotes nature as understood by Buddhists 
and materialists) : yas tu pay an svabhavena vina bhavam acetanah pusyate 
sa punah sarvan prajnaya muktahetukan, yesarii cai ’kantabhavena svabha- 
vat karanam matam, putva trnam isikam va, te labhante na kimcana . . . sva- 
bhavam karanam jnatva na freyah prapnuvanti te, svabhavo hi vina jaya 
mohakarmamanobhavah, “ He is a fool who teaches that nature alone exists, 
or that cause of change is inherent in nature alone” (nature is without in- 
telligence and, 5I. 9, only intelligence gives success ; hence nature without 
intelligence would result in nothing; the final opinion given in 5I. 6 on 
svabhava and paribhava). C. has a curious v. 1. (for putva, etc.) crutva 
nmam rsinam va. 



khya system; Yogas the Yoga system. The pious Yogas say, 
How can one be freed when one is without a personal God 
(anlqvarah) ; while the Samkhyas say that one who knows 
truly all earthly courses becomes unaffected by objects, and 
would clearly get released from the body in this way alone. 
This is the exposition of release given by the veiy intelligent 
Samkhyas. But one should take as the means of release that 
explanation which is given agreeably to his own party. . . . 
The Yogas rely on immediate perception (of truth), while 
the Samkhyas determine according to their code. For my 
part, I approve of both, 1 for either system followed according 
to its code would lead to the highest course (emancipation). 
Purity, penance, compassion toward all creatures, and keeping 
vows, are found equally in both (systems), but the (philo- 
sophic) exposition is not the same in both.” The last words, 
dar§anam na samam tayoh, “ the exposition is not the same,” 
can point here only to the essential difference just indicated 
by the speaker, namely, that one admits and one denies God. 
And it is to be noticed that this is the end of the explanation. 
There is not the slightest hint that the anlgvara or atheistic 
Samkhyas believe in God (a personal Lord, I^vara). 

It must also be remembered that the very term here used 
to describe the Saiiikhya belief, far from being admitted as 
one that connotes a belief in Brahman, is reprehended, not 
only in the pietistic question above (which may fairly be put 
categorically as “ it is impossible to be saved if one does not 
believe in a personal God”), but also in the Gita, which 
links together as a “creed of devils” the denial of “reality, 
basis, and personal God,” asatyam apratistham te jagad ahur 
anlgvaram, Gita, 16, 8, an expression which would have been 
impossible had the anlqvara doctrine been accepted as simply 
a formal modification of deism, implying a belief in a back- 
ground of Brahman. 

I do not think that anl§vara can possibly mean here “ not 

1 The Yoga has the immediate perception of the mystic : pratyaksahetavo 
yogah sarhkhyah jastravinijcayah, ubhe cai 'te mate tattve mama (Bhis- 
masya), $1. 7. 



having the senses as master,” as it does in xii, 247, 7, where 
it is opposed to indriyanam vagyatma ; a passage mistranslated 
by the author of Nirvana, p. 96, as “ Without the Lord one 
attains the place of immortality,” though it clearly means: 
“Not having (the senses as) a master one attains the im- 
mortal state, but being subiect to the senses one obtains 

In the theistic religion, the personal God not only supplants 
the old explanation of spirit, but even takes the place of Pra- 
krti, the unmanifest unknown Source of the Samkhya, and 
creates everything, as does egoism in the pure dogma of the 
Samkhya, as “the name made by egoism, which is synony- 
mous,” ahamkarakrtam eai ’va nama paryayavaeakani, xii, 
340, 62. So to the sectary the name is ever indifferent. 
As to-day he accepts Christ as Iris own divinity under another 
name, so he did of old. The passage in the Gita is well 
known, which establishes the principle. In xiii, 14, 318, it 
is said: “In the Samkhya system the All-soul is called Puru- 
sha,” i. e. the Samkhyas recognize only Purusha, but we say 
that their Purusha is our All-soul. The twenty-fifth, Puru- 
sha, is thus identified with wisdom, vidya, xii, 308, 7 ff. In 
a preceding section, 303, 119, Hiranyagarbha is intellect, and 
is called Virinca, Aja, etc., “ called by many names in the 
Samkhya (Jastra.” 

Yoga as Deistic and Brahmaistic. 

The ancient Yogin tales in the epic show that there are 
important differences between the older and later view of 
Yoga. To stand on one leg for years and keep quiet long 
enough for birds to nest in one’s matted locks was the “ disci- 
pline” of the primitive Yogin as he is represented in these 
tales. But the Yogin of the later epic regards all such practices 
as crude and unsatisfactory. His discipline is an elaborate 
course of breathings and mental confinement in bodily postures 
described as customary in the Yoga (Sstras. So many breath- 
ings at such a time and so many at another, minute attention 
(in a sitting posture) to concentration and meditation, the 



whole paraphernalia of Patanjali, exercised for a “limited 
time,” 1 not a word about standing on one leg for years. The 
difference is more than superficial, however. The one-leg 
Yogin strove for one thing only, supernatural powers. Tale 
after tale recounts what powers he gained by these exercises, 
and these powers were his goal. He was deistic but he had 
no thought of “entering Brahman,” only of controlling the 
powers terrestrial, celestial, and elemental. On death his 
goal is to be a spirit free and powerful, enjoying good tilings. 
On the other hand, the Yogin of the pseudo-epic discipline 
learns all these powers, but “ he who practises them goes to 
hell,” because his goal was not to he a thaumaturge but to be 
released. Both experienced the apunarbhavakama, “longing 
not to be bom again,” but the first desired bala, or Yoga 
“ lordship,” aigvarya, and all his efforts were directed to that 
end ; while the last desired lordship only as a means soon to 
be rejected for something higher, release, moksa, or kevalatva, 
isolation, 2 and eventually the recognition of ekatva, unity, of 
intellect, mind, senses, and universal soul, atmano vyapinah, 
xii, 241, 2-3. 3 

The Bralimaistic Yogin is an advance on the deistic Yogin. 
The latter recognizes only isolation, kevalatva. So under 
the influence of Vishnuism a lecture which teaches Brahman 
isolation appears revamped as pantheistic Brahmaism. 4 

In xii, 317, 16 ff., the Yogin meditates on the eternal Lord- 
Spirit and Brahman, tasthusam purusam nityam . . . I§anam 
bralnna ca, the Yogin being in concentration and trance, saih- 
yama, samadhi: “Like a flame in a windless place, like a 

1 xii, 241, 22 ff. evarn parimitam kalam (six months) acaran asino hi 
rahasy eko gacched aksarasamyatam. Cf. pratibha, apavarga, 317, 14. 

2 The chapter xii, 289, shows that moksa may he simply isolation or inde- 
pendence and does not necessarily connote absorption. 

3 The whole Yogakrtya is comprised here in this union as “ the highest 

4 The compilers are not averse to this practice ; it is a common Hindu 
method of improvement. Either the text is rewritten and interpolated or it 
is allowed to stand and another section is prefixed or added of the same con- 
tent differently treated. The rule is that the improvement precedes the 



mountain peak (compare kutastha), he beholds Brahman, 
which is like a fire in great darkness.” Then “ on abandoning 
his body without a witness,” this Yogin, after attaining in life 
his powers over the breathings and elements, rudrapradhanas, 
and wandering about with the “ body of eight characteristics,” 
enters into the Lord- Spirit who is isolated, kevalam yati, for 
“this is the Yogin’s Yoga; what else would have the sign of 
Yoga? ” 1 So ends the chapter, without a suggestion that the 
Yogin is to be identified with Vishnu. 

In the imitation and improvement of tins passage, thrust 
before it in the text, the Yogin’s release does not end matters, 
though Vishnuism is inserted rather clumsily, as will be seen 
from an analysis of the whole section, 301, 11 ff. “ Cutting 
off the five faults by Yoga, people freed of sins obtain that 
place (or condition), tat padam, like as big fishes cut through 
a net and get the water (the fish is not identical with the 
water, tat padam is place or condition, freedom). Even as 
strong animals, mrgah, cut the net, so they would get a clean 
road when they are freed from all their bonds. Endued with 
strength, Yogas, on cutting thus the bonds made by greed, go 
the clean way that is highest and auspicious. . . . Those with- 
out power are destroyed, those that have power are released, 
mucyante balanvitah. . . . On acquiring Yoga-power one can 
oppose the many objects of sense, vyuhate visayan, as an ele- 
phant opposes a great stream. By Yoga-power made inde- 
pendent, ava§ah, Yogins enter Prajapatis and seers and gods 
and the elements, as their lords. Not Yama nor the End- 
maker (differentiated here, often as one), though angered, 
nor Death, fearful in prowess, not all these lord it over a 
Yoga of unmeasured energy. A Yoga could make himself 
many thousands when he has got his power, and with these 
could wander over earth. Such an one could take the objects 
of sense and then perform hard austerity and again reduce it, 
as the sun does his beams of light, tejogunas. The Yoga who 
holds to the power and is lord of bonds obtains in release, 
vimokse, the fullest lordship, prabhavisnutva. These powers 
1 etad hi yogam yoganam kim anyad yogalaksanam, 317, 27. 



obtained through Yoga have been obtained by me. For elu- 
cidation I will now tell thee again, O King, also about the 
subtile powers. 1 Hear from me, O Bharata, the subtile signs 
of the soul in concentration, samadhana, and in respect to con- 
templation, dharana, 0 lord. As an archer by being attentive, 
apramatta, with concentration hits the mark, so the Yogin, 
properly intent, doubtless obtains release, moksa. As a man 
intent, yukta, with intent mind would go up a ladder, steadily 
fixing his thoughts on the vessel full of oil (in his hands), so 
the Yoga here, intent, O King, steadily makes spotless his 
soul (till) it looks like the image of the sun. 2 . As the steers- 
man with concentration, samahita, would guide a ship on the 
ocean, so by applying self-concentration with intentness, atma- 
samadhanam yuktva yogena, he that knows the true, tattva, 
gets a place hard to attain, durgam asthanam, after leaving 
his body here. As a charioteer with concentration yoking, 
yuktva, good horses, quickly brings the knight to the desired 
place, degam istam, so, O King, the Yogin with his mind con- 
centrated in contemplation quickly gets the highest place, 
param sthanan, just as the arrow when released, mukta, finds 
its mark. The Yogin who stands steadily seeing self in self 
destroys sin and gains the unalterable place, padam, of those 
who are pure. The Yogin who properly joins, yunkte, with 
his soul (self) the subtile self in the navel, throat, head, heart, 
chest, sides, eye, ear, and nose, quickly consuming his Karma, 
good and bad, though mountainous (in size), having recourse 
to highest Yoga is released, if he wishes.” 

This is the end of the discourse for the present. Nothing 
is said of the Yogin’s emancipation being other than a release 
from bonds. The conversation turns to the question of food 
and means of restraint of the senses, the hard path of auster- 

1 These words are perhaps the mark of interpolation here. 

2 sneha-purne yatha patre mana adhaya ni?calam, puruso yukta arohet 
sopanam yuktamanasah, yuktas tatha ’yam atmanaihyogah parthiva nifcalam 
karoty amalam atmanam bhaskaropamadarfanam. In 317, 22, tailapatram 
yatha purnam karahhyam grhya purusah sopanam aruhed bhitas tarjyamano 
'sipanibhih samyatatma bhayat tesam na patrad bindum utsrjet tathai Vo 
’ttaram agamya ekagramanasas tatha, etc. 



ities which makes the subtile soul shine forth, but he who 
follows it “is released from birth and death, ill and weal.” 
“This,” it is then said, “is what has been set forth in various 
Yoga-^astras ; in the twice-born is admittedly the liighest 
Yoga practice,” krtyam, <jl. 57. 

Thus far the 9 lokas and the final stanza seems to show that 
this is the end. But to this are tagged on five tristubh stan- 
zas, with which the chapter now concludes: “That highest 
Brahman-made Brahman and Lord Vishnu, the boon-giver, O 
great-souled one, and Bhava, and Dliarma, and the six-faced 
(god), and the sons of Brahm&n, tamas, rajas, sattva, and high- 
est Prakrti, and Siddhi the goddess wife of Varuna, and all 
energy, tejas, and patience, and the pure lord of stars hi the 
sky with the stars, all the all-gods, the snakes, and manes, 
and all mountains, the terrible seas, all rivers with forests and 
clouds, Yagas and nagas, troops of genii, spaces, the angel 
hosts, males and females — one after the other attaining, the 
great great-souled Yogin would enter soon after he is released. 
And this narration, O King, is auspicious in that it rests on 
the god who has great vigor and intelligence. Such a great- 
souled Yogin, overpowering all mortals, acts, having the self of 
Narayana” (according to the commentator, makes all things 
as being identical with Narayana). 1 

It is true that a view which ignores every indication of in- 
terpolation may insist that literature is to be treated without 
critique, overlook the patchwork, and concentrate emphasis 
on this last narayanatma to offset the whole teaching preced- 
ing, which is that the soul gets isolation, not absorption into 
Brahman. But even then Narayana is not philosophical 
Brahman. In the following chapter, which is a new discus- 
sion, 302, 55, the Kapilah Samkhyah are also led to emancipa- 
tion, in which teaching atman rests on Narayana, Narayana 
rests on emancipation, but emancipation has no support (the 
same word as above of the narration which rests on Narayana), 
moksam saktam tu na kvacit ; though the Saihkhya philoso- 

1 yogi sa sarySn abhibhuya martyan narayanatma kurute mahatma, 301, 

62 . 



phers are finally conducted through an unfinished sentence 
eighteen glokas long 1 to Narayana, who bears them to the 
Highest Soul, when they become fitted for immortality, and 
return no more, §1. 78. 

These are chapters of a sectarian cult, which seeks to in- 
clude in its embrace all systems of philosophy, 2 and does so 
vi et armis. The more precious and reliable are those expo- 
sitions which show the systems still but slightly twisted from 
their original form. This last is a system called Vedanta, 
802, 71, as I have already remarked, but in point of fact it, 
i. e., this last chapter, not the preceding exposition, is an ex- 
position of Yoga twisted into sectarian Brahmaism. The 
soul eventually enters Vishnu, who is unconditioned Brahman, 
and does not return ; but it enters by jiva and videha mukti, 
in Yoga style. That is, before death the real soul enters 
Vishnu, leaving behind in a man not soul but only mind and 
senses. Shortly after, however, one is really “released and 
gets peace.” This, it is said, is the Samkhya system which 
is identical with eternal Brahman (302, 96-101; compare 106, 
amurtes tasya . . . samkhyam murtir iti grutih). The Samkhya 
system, which is at first said to be faultless (§1. 4), is in §1. 13 
declared to have faults as well as virtues, the same being true 
of Veda and Yoga; that is, this teaching is put forward as an 
improvement on the old, though the accepted base is the 
Samkhya. It is pretended that the teachers teach as do the 
Kapilas, who are endued with knowledge and “clarified by 
ratiocination,” karanair bliavitah gubliah, gl. 17. 

Difference between Samkhya and Yoga. 

As has been shown above, the epic itself teaches that the 
great difference between the two systems is that the Samkhya 
does not believe in a personal God, while God is the supreme 

1 xii, 302, 24-52. Compare 5-17 also one sentence. These interminable 
sentences are marks of the late style of the pseudo-epic. 

2 In 5 I. 108 it is said that this Vedanta (yl. 71) Samkhya embraces all the 
knowledge found in Samkhyas and Yoga (samkhyesu tathai 'ya yoge), the 
Purana, the great Itihasas (pi.), Arthayastra, and the world (Lokayata ?). 



belief of the Yogin. A further difference is found by the 
commentator in the words of xii, 240, 8, where it is said : 
“ Vishnu in stepping, ^akra in power, Agni in the digestive 
organ (etc.) wishes to enjoy,” bhoktum icchati, a stanza 
wedged between the statements that bodies come from earth, 
etc., and that ears, etc., are organs of sense. What is appar- 
ent is that experience is here shifted from pure spirit to the 
corresponding divinity. 1 

So far as I know, the difference of opinion is nowhere in the 
epic stated to involve a distinction between the two systems, 
and in this chapter the subject of active and experiencing 
spirit is not further touched upon. I doubt, therefore, the 
validity of the commentator’s explanation as applied to the 
epic, but his words are worth citing: “In the Yoga system 
the spirit is not active but experiences only, while in the 
Siimkhya system the spirit neither acts nor experiences. In 
this passage the poet repudiates the first doctrine, and ex- 
presses approval of the second ” (by naming devas as “ enjoy- 
ers,” and thus showing that it is only a false imagination of 
the spirit when it thinks itself an “ enjoyer ”). 2 

According to the epic, all activity resides in Prakrti, the 
Source alone, while experience resides in spirit but only as the 
latter is conditioned by its environment, prakrtisthah, so that 
when it is in the body the highest spirit is called enjoyer and 
active, but it is not really so, kurvann api na lipyate, na 
karoti na lipyate. This is the explanation of the Gita 3 
(which denies that there is any speculative difference between 
the two systems), and is found often enough elsewhere. 4 So 
God as a conditioned being, spirit, enjoys the gunas, as in 
xii, 340, where the twenty-fifth principle, though “ without 

1 As in Mait. Up. vi, 10, bhokta puruso bhojya prakrtih, “enjoy” is some- 
times sensuously rendered, “ Spirit is the eater, Prakrti the food.” Ordinarily 
“ enjoy ” is experience. 

2 yogamate, atma bhoktai ’va na tu karta; samkhyamate tu, na bhokta 
na ’pi karte ’ti; tatra ’dyam dusayati, etc. 

s Gita, 3, 27; 5,7; 13,20, etc. 

4 Compare xii, 247, 1-2: “The spirit supervises modifications (he knows 
them, they do not know him), he does what is to be done (only-) in conjunc- 
tion with the senses and mind, the sixth ” (like a charioteer, as above). 



characteristics,” is gunabhuj or enjoyer of gunas as well as the 
superior creator of gunas, gunasrasta gunadhikali, cl. 28. 1 So 
Civa is sastibhaga (below). “ Like a lamp giving light know 
the jnanatman, knowledge-spirit, Purusha, to be in all crea- 
tures. It makes the ear hear; it hears; it sees. The body is 
the cause (of perception), but this (soul) is the doer of all 
acts,” xii, 210, 40. Here the last clause, sa karta sarvakar- 
manam, means that soul acts only as modified by Prakrti. In 
xii, 222, 17 If.: “Whoso thinks himself an actor, faulty is his 
judgment. Activity is nature only, the only factor,” svabhava 
eva tat sarvam (one becomes vitrsna, <jl. 30, when one knows 
the difference between the Source and its modifications). In 
xii, 304, 45, the Source does every act, and it alone enjoys, 
acnati. Opposed to this is the Brahmaistic view, which holds 
that “the inner soul, antaratman, alone smells, tastes,” etc., 
as an entity separate from elements (below). 

A practical difference may be found in the attitude of the 
two systems toward austerities, though it is stated that this 
exercise is common to both. Nevertheless it cannot be sup- 
posed that the “ knowledge-philosopher ” admitted as much 
tapas as did the Yogin, whose practical discipline was almost 
wholly a “ razor-edged path ” of austerity. The practice is 
occasionally reprehended, as in xii, 221, 4, where it is said 
that fasting is not meritorious, as it is injurious to the soul’s 
discipline, atmatantropaghatah, a view which is of course con- 
tradictory to the mass of teaching in the epic, for example, ib. 
233, 23, where penance is the means of “ attaining to the being 
that creates the universe.” The “ difference between Sariikhya 
and Yoga,” as admitted and explained in the late passage xii, 
237, 29 ff ., is mainly a practical one, in that “the Sariikhya 
keeps aloof from objects of sense, controls the senses, and is 
alike to all creatures, friendly to all, indifferent to all things, 2 
injures no creatures, and so attains to Brahman ; ” whereas 
that Yoga is released “ who, transcending supernatural power, 
ceases” (from activity). The Yoga is thus described in one 

1 The twenty-fifth, not the twenty-sixth principle, is here God. 

2 sarvabhutasadrn maitrah samalostajmakaneanah, 38, a standing epithet. 




verse: yogaigvaryam atikranto yo niskramati mucyate, 287, 
40. The dependence of the Samkhya on knowledge alone is 
here merely implied, though the following image of the saving 
“ ship of knowledge ” makes it clearer, but the whole passage 
is a late attempt to interpret Samkhya by another norm. 1 

One further practical difference between the systems is 
pointed out by the commentator at xii, 241, 34, where, after 
asceticism is described, it is said that a man of low caste or a 
woman seeking virtue “ may attain the highest course by this 
path” (of the Yoga). The commentator takes pains to re- 
mark that this applies only to the Yoga, and not to the 
Samkhya. A little farther on, in 247, 16, where the same 
system is still taught, but on the intellectual side, not on the 
ascetic side, it is, expressly stated that the (Tistra should be 
told only to men of the higher castes, Snatakas. 2 

It is expressly charged against the Pagupata sect that it is 
subversive of caste : “ I, Rudra, formerly for the first time 
invented the mysterious Pagupata religion, beneficent to all, 
facing in all directions, one that takes years or only ten days 3 
to learn, one which, though blamed by the unintelligent (be- 
cause it is) here and there opposed to the rules of the (iastra 
and those of the Orders, varnagramakrtair dharmair viparltarii 

1 brahmanam abhivartate, a late carelessness, repeater! with ca ’dhigacch- 
ati, si. 30 and 41. The four-faced Brahman and the highest Brahman, re- 
spectively, is the commentator’s ready explanation (“masculine by Vedic 
licence”). The same sort of thing is found in another later passage, where 
a double carelessness appears, brahmanam adhigatva (sic) ca, iii, 83, 73. 
Part of the above description is a copy of the Gita, nirmamap ca ’nahariikaro 
nirdvandvaf chinnasam 9 ayah nai ’va krudhyati na dvcsti, 237, 34, as in Gita, 
5, 3; 12, 13 (= 2, 71) ; 18, 53, brahmabhuyaya kalpate. 

2 See below the passage inculcating pure Yoga (the twenty-sixth prin- 
ciple), where it is said, xii, 319, 89, that it is a doctrine of emancipation for 
all, and knowledge is to be got from all, for all castes are Brahmans, all are 
born of Brahman, and all castes are equal ; and compare ib. 188, 10 if., na 
vifeso 'sti varnanam, etc. In 251, 21, atmajnanam idam guyham, as in the 
earliest Upanishads. A “ God without characteristics ” is responsible for 
the democratic equality of the “no caste” view. So Givaism teaches that 
castes are only indications of position, brahmah svabhavah is everywhere 
equal, and all men are children of the one God who created them, xiii, 143, 

3 Instead of ten days, says the commentator, the Gaudas read “ five days.” 



kvaeit samam, is nevertheless appreciated by those of per- 
fected wisdom, gatantas, and is really superior to the Orders ” 
(atyagramam, xii, 285, 194-195). In the preceding stanza, 
this Pagupata is contrasted with the gods’ and demons’ relig- 
ion of austerity, the latter being “drawn from the Vedas and 
Sariikhya and Yoga by logic,” 1 another mark of difference in 
the views urged in the epic, not, as often, concealed under a 
pretended unity, but openly stated. 


I would say a word here in regard to the sects recognized 
in the epic, though, except for their philosophy, I do not in- 
tend to touch further on them. The epic commentator sees in 
the epithet pancamahakalpa, applied to Vishnu, a reference 
to the scriptures, agamas, of five diverse sects, Sauras, (Sktas, 
Ganegas, (laivas, and Vaisnavas. The epic in reality recog- 
nizes only the first and last two, for the allusion to shadow- 
worship (which the commentator explains as a Left-hand rite) 
though interesting, does not imply necessarily a body called 
(Jaktas, and Ganegas are unknown, the god himself belong- 
ing only to the pseudo-epic introduction, and very likely in- 
terpolated there, as has been shown by Dr. Winternitz. Even 
Durga seems to be a late addition to the epic as she appears 
hymned. But the C'aivas are known as having a religion 
called Pagupata (above) and the Vaisnavas and Sauras are 
known in two late passages, xviii, 6, 97 and vii, 82, 16, under 
these names. I suppose only the synthetic method would 
claim that the whole epic recognizes the titles of sects so 
sporadically mentioned. The older Vishnuite sect-name is 
Pancaratra or the more personal “devotees of the Lord,” 
Bhagavatas, and Bhagavadbhaktas, even these being rather 

1 Rudra says to Daksa : bhuyag ca te varam dadmi tam tvam grhnisya 
surrata, prasannavadano bhutva tad ihai ’kamanah grnu ; vedat sadangad 
uddhrtya sarhkhya-yogac ca yuktitah tapah sutaptara vipulam dugcaraiii 
devadanavaih, xii, 285, 191-192 ; and then as aboye, in contrast, the Pagupata 
system, which has overthrown the older systems (Rudra destroys Daksa’s 



rare. The last, for example, is found in i, 214, 2 (with 
bhaiksas or cauksas). The same passage that calls Vishnu 
pancamahakalpa gives him the titles of Praelnagarbha (below) 
and Kaugika and identifies him with the Atharva§iras Upani- 
shad, xii, 339, 113-125. Though the god is here Vishnu, I 
venture to think the last epithets were originally applied to 
()iva. The “white men” of the White Island, or rather 
country (dvipa = the dig uttara or more exactly uttarapagci- 
mena, “in the Northwest,” 336, 8-10; 337, 21 ff.) must be 
Kashmere Brahmans, who are often almost as white as Euro- 
peans and whose religion was the worship of Civ a (as a god of 
culture and letters) in monotheistic form, which is here per- 
verted. The location “Northwest” and “far North” can 
scarcely be anywhere else than Kashmere, where alone “ north- 
ern white men,” §vetah pumansah, 336, 10, were to be seen. 1 

The Different Schemata. 

The philosophical schemes elaborated in the epic show three 
distinct groupings, which must belong to different systems. 
These are the Samkhya, the Yoga, and a third system, which 
follows a different series of topics. All three differ essentially 
from Vedism and Brahmaism, as this latter, in turn, differs 
from what we call Vedanta. Both of the latter are repre- 
sented, making six systems, as said above ; but of these there 
are full schemata or topica in three cases at least, 2 indicating 
what for convenience I shall call scholastic differences, the 
three schematizing systems being here termed schools. It is 
unnecessary to point out that no one set of teachers, much 
less the one poet of the unhistorical method, would have incul- 
cated six systems, or elaborated three schools, especially as the 
topics of two of these schools imply a fundamental difference 
between them. 

1 The “ Sea of milk ” in the Puranas is said to surround a Himalayan 
mountain, Kraunca. The second (earlier) account of the “ white men ” in the 
epic is quite Samkhyan, God is Purusha, etc. 

2 Compare also the rather rare recognition of pure Vedanta Mava- 
Brahmaism, and above in the first chapter the philosophy copied from the 
Upanishads without identification of soul with sectarian god. 



Common to all three schools is the distinction between 
the First Cause or Source as manifest and unmanifest. The 
manifest, or known, is all that is born, grows, ages, and dies, 
while the unmanifest, or unknown, is “the opposite,” 1 that 
is, it is devoid of these four marks, laksanas. Further, Sam- 
khya and Yoga both admit two selves, atmans, it is said, which 
are declared “in the Vedas and in the Siddhantas.” 2 The 
first is that born with the four marks, that is, those of the 
manifest, and has four objects (caturvarga, virtue, pleasure, 
gain, emancipation). This is the manifest self, born of the 
unmanifest; it is awakened, buddha, but has not the highest 
intelligence, cetana; it is the conditioned sattva soul, in dis- 
tinction from the pure knowing soul, ksetrajna, though both 
are attached to objects of sense. “ Both systems admit twenty- 
five topics,” a statement to be reviewed below. 

The Unmanifest is that which cannot be known, avedyam, 
which has no padanyasa, leaves no track, and is therefore 
beyond knowledge, xii, 205, 18; avedyam avyaktam, xii, 319, 
42. Kapila calls it the apxn, adya, and says he uses the term 
First Cause, Source, Prakrti, merely to escape a regressus 
ad infinitum. It is therefore merely a name, samjnamatram. 
It is used of the That : “ One could never reach the end of 
causation, nai ’va ’ntam karanasye ’yat, even if one went 
unceasingly like an arrow from the cord, yatha bano gunacyu- 
tah, and swift as thought. Nothing is more subtile than the 

1 So in xii, 217, 9-10, it is said that Prakrti creates and has three gunas, 
while spirit’s marks are “ the opposite ” (for the threefold gunas are only his 
“ turban,” gl. 12). 

2 xii, 237, 27, 31, siddhantesu. Siddhanta is mentioned also in i, 70, 44. 
In the present passage the commentator takes the Vedas and Siddhantas as 
Purvamimansa and Uttaramlmansa. Another late expression in this section 
describes the effulgent jlva-yoked car as having all the Tantras as its goad 
(sarvatantrapratodah, xii, 237, 11, straddles the padas), where the commentator 
says C'astra, and is probably right, as we have Nyayatantras mentioned, which 
are doubtless works on logic. Compare with the passage above, xii, 206, 28, 
avyaktatma puruso vyaktakarma so 'vyaktatvam gacchati hy antakale ; xii, 
199, 125, caturbhir laksanair hinam tatha sadbhih sasodajaih purusam tarn 
atikramya aka^am pratipadyate (the six are ills and the sixteen are breaths, 
organs, and mind, according to the commentator), but the four are here said 
to be cetas and three proofs. 



unmanif est That (§1.18) ; nothing is coarser. Finer than fine, 
greater than great is That, the invisible end of all things,” 
xii, 240, 28 (29 = ()vet. Up. iii, 16 ; Gita, 18, 18). It is a term 
used in both philosophies, and is simply equivalent to the 
invisible unknown First Cause. From its synonym Prakrti, 
First Cause, it may be called simply the Source. So also 
Brahman is avyaktam. Usually this term is defined in such 
negatives as in neti neti, a superabundance of which appears 
in this definition : “ Brahman has not been explained by 
mantras; with the world of experience it has not anything 
in common; it has not sound, touch, not form; it is not com- 
prehended; not manifest . . . not female, not male, not neuter 
(as in 251, 22), not being, not not-being, not being-and-aot- 
being . . . not perishable,” 1 an imitation of older matter. 

This “Unknown,” which forms the common basis of the 
great philosophical systems, in the Saiiikhya connotes potential 
egoism, becomes known first as Ego or self-conscious intellect, 
and out of this egoism is developed the whole created uni- 
verse ; over against which stands the pure unconscious spirit, 
the real Ego. This, in outline, is the whole plan of the Saiii- 
khya philosophy, which admits nothing outside of pure Ego 
and self-conscious Ego, and ascribes all apparent other to 
modifications of egoism. There are here twenty-four prin- 
ciples over against the pure spirit Ego as the twenty-fifth. 2 

On the other hand, besides these, the Yogin’s system super- 
adds one exalted spirit as Supreme Spirit, or God, the twenty- 
sixth principle. 

The Pa§upatas and Bhagavatas have a different system of 
categories, but teach that the Supreme Spirit as a personal God 
becomes manifest ; in the latter sect, as a god-man. 

Common to the three schools is the belief in the three con- 
stituents of the Unmanifest, called gunas; but these are some- 
times treated as constituents and sometimes as attributes. 

1 na san na ca ’sat sad-asac ca tan na . . . tad aksaram na ksarati ’ti viddlii. 
In 251, 22, Brahman is asukham as well as aduhkham, “ not joy, not sorrow.” 

2 Prakrti is devoid of the highest intelligence, aeetana, and only when 
supervised by spirit creates and destroys. Purusha has millions or 1,400,000 
courses, xii, 315, 12 ; ib. 2 ; 281, 36. 



The Gunas. 

The Unknown becomes known as a result of energy, tejas 
or rajas, rousing itself and rousing conditioned being, sattva, 1 
out of the equilibrium which is maintained between these two 
and inertia (dulness, darkness, tamas). These are the three 
constituents of the conscious Ego, and consequently of all 
things except pure spirit. That is to say, energy, inertia, 
and existence (conditioned being), characterize all things, 
and life begins with energy moving sattva as well as itself. 
A moral interpretation of these strands, gunas, as they are 
called, makes being, as compared with the other two, repre- 
sent the true and real and good ; inertia, the stupid and bad ; 
while energy may be good or bad, but is never the best, as 
that is devoid of all activity (quietism). 2 These gunas, con- 
stituents, are, to use a term taken from their grammatical 
application, themselves gunated or characterized by the pres- 
ence of certain qualities, a meaning often found employed in 
the case of guna. Thus in xii, 334, 2, one abandons fourfold 
faults, eightfold tamas and fivefold rajas. What is of most 
importance, however, from the historical rather than the philo- 
sophical point of view, is that in these groups there is no 
uniformity in the teaching of the epic. Thus in xii, 314, 21 ff., 
not five, as above, but over twenty faults are given as charac- 
teristics, gunas, of rajas. In the same way, sattva has in xii, 

1 Sattva (compare satyasva satyam) is being, but not absolute being, which 
is free from consciousness of self. We may best render the “ three strands ” 
or inherent constituents of creation (everything except pure spirit) by energy, 
inertia, and conscious-existence, which exist potentially in the undeveloped 
and actually in the developed universe. I am aware that the gunas are 
translated differently by high authorities, but must for the present refrain 
from further discussion of the interpretation. 

2 Compare Gita, 17, 26: “ Sat is employed in the meaning of existence and 
of good” (commentator wrong). The avyakta (unknown undeveloped) is 
gunated as much as is vyakta, only the equilibrium not being disturbed the 
gunas are merely potential, avyaktam trigunam smrtam, xiv, 39, 24. In re- 
gard to “ darkness,” it must be remembered that in the older philosophies, 
darkness, tamas, is not a quality but a substance (only the Xyaya regards 
it as absence of light). See the argument in the Aulukya chapter of the 



342, 13, eighteen gunas, while in 314, 17 ff., nearly double 
this number are given it, including most of the former group 
but placed in a different arrangement. Again in xii, 302, 14- 
16, sattva has ten (unexplained) gunas ; rajas, nine ; tamas, 
eight; buddhi, seven; manas, six; nabhas, five; but then, 
again, buddlii has fourteen ; tamas, three ; rajas, two ; sattva, 
one. 1 This merely means that each strand has certain attri- 
butes. 2 The same list, for instance, is given in the Anuglta, 
xiv, 38, 2 ff., as indications of sattva. It seems unnecessary 
to enumerate these varying characteristics. The gist of them 
all is found in Gita, 14, 9 ff. : sattva belongs to pleasant 
things, rajas to activity, tamas to apathy. So in xii, 194, 30, 
a touch of joy is characteristic of sattva, and “ if anything is 
joined to joy there is the condition, bliava, of sattva” (only 
five are given here) ; while in 35 there are five lingas or signs 
of energy, rajas, and in 36, five gunas of tamas (= 286, 25 ff., 
with v. 1. = 248, 19 ff.) As tejas, energy, is attributed to 
Brahman, the term falls into comparative desuetude, being 
replaced by the less moral rajas, while tejas is left as a 
virtuous characteristic: dliutapapma tu tejasvl . . . nin lsed 
brahmanah padam (said of tire good man), and Brahman is 
tejomayam, xii, 241, 9 and 13. So tejas is a good quality, 
Gita, 16, 3. 3 

In this conception, sattva is as much of a bond as are the 
other two gunas. Knowledge and pleasure are the attach- 
ments with which it binds the soul; while rajas binds with 
action and tamas with heedlessness, laziness, sleep, the signs 
of inertia, Gita, 14, 6-8. 

1 The eighteen gunas of sattva, to give an example, are pritih prakiifam 
udrelco laghuta sukham eva ca, akarpani/am asnmrambhah santosah craddudha- 
nata, ksamd dhrtir ahinsd ca faucam akrodha eva ca, arjovam samata satyain 
anasuya tathai ’va ca (those in italics reappear in the longer list, 314, 17-20). 

2 The Hindu conception is not quite uniform in regard to the gunas, but 
there is, 1 think, no reason for confounding essential constituents with attri- 
butes. Joy and sorrow are not the gunas themselves but their objective signs 
in the moral world. The true opposites are tejas and tamas, light and dark- 
ness, as energy and inertia physically, and as goodness and badness morally. 

8 But rajas often keeps its pure tejas sense, as in xiv, 30, 0, rajah parya- 
yakarakam, rajas is energy. 



The Source, Prakrti, is the combination of the three gunas, 
represented as a female productive power. As a lamp lights 
thousands so the Source modifies herself into the many gunas 
(characteristics) of spirit. She does it of her own will and 
desire, and for the sake of sport. 1 

According to the proportion of gunas in a creature, it has 
a high, middle, or low place, xii, 315, 3-4; Gita, 14, 18. Evi- 
dently, therefore, the Yoga-god must be without gunas, so 
nirguna is predicated of him and of Brahman, nirgunasya kuto 
guntih, xii, 306, 29, as say the gunadaiyinah, but as God must 
be everything he is also “ with gunas ” as well as “ without 
gunas,” a contradiction which is on a par with God’s being 
being and not being being and being neither being nor not- 
being, the common tangle of metaphysics. 2 * * * * * In fact, religious 
philosophy is hopelessly at sea, not only in regard to the 
question of a conditioned God hut also in regard to the gunas 
of the spirit. It is universally admitted that energy and 
inertia must be dispensed with in order to a full attainment 
of pure spirithood, xiv, 51, 25. But when spirit has sattva 
alone or is in sattva alone, sattvam asthaya kevalam, is it one 
with this being or not? Some say, “ and they are wise,” that 
spirit and sattva have unity, ksetrajnasattvayor aikyam, but 
this is wrong. Still, they cannot exist apart. There is unity 
and diversity, as in the case of the lotus and water-drop, the 
fish in water, the fly in the Udumbara plant, ekatvaniinatvam, 
xiv, 48, 9-1 1. 8 In xiii, 108, 7, sattva must be “washed out” 

1 prakrtir guniin vikurute svacchandena ’tmakamyaya kridarthe tu, xii, 

314, 15-115 (prakrtis tatha vikurute purusasya gunan bahun). 

1 God is nirguna and gunatman and nirguna alone and triguna, etc., xii, 

339, 3 ff. ; xiii, 137, 3. Guna-made are all existences, Gita, 7, 13; God is not 

in them, they are in him, ib., 12. They do not affect God, xii, 340, 22 (in 20 it 
is said that those devoid of rajas and tamas attain to God, presumably retain- 

ing sattva; but elsewhere sattva must also be lost, e. g., 335, 30); viddhi 

bhavan mada?rayan, xiv, 54, 2 ; avyaktat utpanno mahan atma adir gunanam, 
40, 1. 

8 Here Telang is obliged to render sattva as goodness and as nature, ac- 
cording to the verse, e. g., unintelligent sattva, 49, 9, and 12, where the spirit 
enjoys sattva. Sattva, however, is always conditioned existence or a condi- 
tioned being, abstract or concrete. It is the highest, because it may be free 



of the soul of pure Yogins, along with rajas and tamas. In 
these cases we have simply an attempt on the part of theology 
to utilize the terms of atheistic philosophy, which naturally 
leads to confusion. For the terms (applicable to Prakrti) of 
Samkhya are incompatible with the philosophy which substi- 
tutes God for both Purusha and Prakrti. 

When the gunas are called atmagunas, as in xiv, 12, 4, it is 
to distinguish them as mental from the bodily constituents, 
gunah garirajah, with which they are compared. As the three 
constituents of the body, gitosne vayug ca (= kapha, pitta, 
vata) give a healthy condition when in equilibrium, so the 
three atmagunas, when equal, produce a healthy condition. 
Here the three are merely essential elements in a tridhatu or 
threefold entity. Thus elements are called, as the constit- 
uents or factors, cLhatavah, inherent in the Source, dhatavah 
pancabhautikah, iii, 211, 9 ff., just as the essential constituents 
of a king’s concern are called gunas, xv, 6, 6. 

Plurality of Spirits. 

The passage just cited from the Anuglta on “unity and 
diversity ” reflects an important section in (’anti. Here, xii, 
316, 3 ff., a difference is established between Unmanifest 
Prakrti and spirit, the former being affected by gunas, inca- 
pable of escaping from them, and inherently ignorant; the 
latter being both pure and contaminated, because he is asso- 
ciated with the Unmanifest. Causing creation he is called 
creator. Because of his observing as a spectator and of his 

from rajas and tamas, but is itself, though “ good,” not " best.” This is what 
is in the Hindu’s mind, but the distinction between this existence and that of 
God or Brahman is much like that between the highest knowledge of man 
and that non-knowledge knowledge of God. Both are attempts to release the 
infinite from the limitation of any definition. To say He is is to put Him 
in a class, hence we cannot say He is, but of course we cannot say “ He is 
not.” He is pure knowledge but this is a limitation ; hence He knows with- 
out knowing and exists without existing, totally indefinable. The difference 
between the early Upanishad and epic philosophy in respect of conditioned 
Atman, is that only the latter uses technical Samkhya terms, just as the later 
Upanishads use them. 



being without a second, ananyatva, and of his false opinion 
(of himself), abhimana, Yatis (Yogas) regard hhn (the same 
spirit) as both eternal and non-eternal, manifest and unman- 
ifest : “ This is what I have heard said ; but those who have 
the religion of compassion and abide by knowledge alone, 
say that there is unity in the Unmanifest but a plurality of 
spirits.” Here the last authorities are clearly the Samkhyas, 
who are characterized in the epic not only as “devoted to 
knowledge,” but as especially moral and compassionate. 1 The 
section concludes: “Purusha, spirit, and the Unmanifest 
(masculine) are different. The latter is called eternal but is 
not eternal. Spirit’s connection with the Unmanifest is that 
of the grass blade in its sheath, the fly and the Udumbara, 
the fish in water, the fire in the pan, the lotus and water-drop ; 
there is connection but not identity. This is the Samkhya 
view, the best estimate, parisamkhyana.” 

So in xii, 351, 1, the question is raised in regard to one or 
many spirits, only to be answered with the statement that 
there may be many spirits, but they all have the same birth- 
place. The answer is really assumed in the question, 2 so that 
the passage is of interest chiefly as showing a full recognition 
of the fact that Kapila taught (as above) the doctrine of mul- 
titudinous spirits without a common source. This is brought 
out more distinctly in the following statement, viz., that Vyasa 
(the Yoga) teaches that all spirits have a common source, 
although Kapila and other metaphysicians have declared 
Castras in which a plurality of spirits is inculcated : “ In 
the discussion (of this subject) by Sariikliya-Yogas there are 
many spirits assumed in the world and (these pliilosophers) 
will not grant that one spirit (exists as the sole source). (But 

1 ib. 5I. 11 : avyaktai ’katvam itv ahur nanatram purusas tatha sarvabhu- 
tadayavantah kevalam jnanam asthitah. It is worth noticing how frequently 
the Samkhyas are called “those who hare compassion and knowledge,” a 
Buddhistic inheritance apparently, though this is a suggestion liable to seem 

2 bahayah purusa brahmann utaho eka eva tu, ko hy atra purusah fresthah 
ko ya yonir ilio ’cyate, “ Are there many spirits or only one 1 Which is the 
best 2 or which (spirit) is the source 1 ” 



this is a mere assumption) and, as a sole source of many 
spirits is declared (to exist), so will I explain that spirit which 
is superior to conditions (or has superior characteristics) to 
be the All. . . . This hymn [Rig Veda, x, 90], the Purusha- 
Sukta expounded in all the Vedas as right and true, has been 
considered by (Vyasa), the lion among sages. (Tistras with 
rules and exceptions, utsargenapavadena, have been proclaimed 
by sage metaphysicians beginning with Kapila. But Vyasa 
has proclaimed spirit-unity , purusfiikatvam, and his teaching 
in brief will I declare.” 

Nothing could show more clearly the absurdity of denying 
the variegated beliefs reflected in the epic, or the ancient 
foundation of the Kapila, not in Brahman but in a plurality 
of spirits devoid of a common source. In Vyasa we have a 
revolt against Kapila, not in absolute rebuttal, but in a denial 
of his chief principles and in an attempt to show that the 
time-honored system could be interpreted in accordance with 
a belief in a personal God. 1 

Another point of importance is the decision with which the 
heretical view is attacked: “Unity is a proper view, separate- 
ness is an incorrect view,” ekatvam dargantuii nanatvam adar- 
ganam ; again: “The view that the Supreme Soul is one 
with the individual soul is the correct view; the view that 
they are separate is an incorrect view,” anidarganam (the com- 
mentator says there is another reading anudarganam, which 
he interprets as a following or later view, xii, 306, 35-37). 2 

1 Here the author of Nirvana, p. 07, suppresses the fact that Vyasa’s view 
is placed in antithesis to Kapila’s, and, leaping over the intervening verses, 
says that Samkhya-Yoga in this passage teaches only a common source of 
souls. It is indeed said at the end of the text that Samkhya-Yoga is Vishnu- 
ism (see just below), but no notice is taken of the fact in Nirvana that the 
special passage under consideration presents the matter quite differently. 
The passage above almost seems to imply that Vyasa is to be regarded as 
a philosophical teacher especially, perhaps as the author of a philosophical 
work (Holtzmann opposed, iv, p. Ill); possibly of the Vyasagrantha of i, 
70, 45 (commentator opposed). In any case, Vyasa’s teaching, though not 
that of Badarayana, claims to improve on Kapila’s view. 

2 Compare Katha, iv, 11: (He perishes) “who sees, as it were, separateness 
here,” ya iha nane ’va pafyati (the separateness is here that of any part of 



Of course the SamkhyarYogas, being the models, are cred- 
ited with the view expressly said to be not theirs. So in the 
exposition above from xii, 351, after Vyasa has been distinctly 
opposed to the SamkhyarYogas and his view is explained to 
be that the different souls (created by Brahmin) at last are 
absorbed into their one source, the “ subtile entity appearing as 
four ” (Aniruddlia, etc.), it is calmly said that this is Samkhya 
and Yoga, xii, 352, 12-13, 23. But occasionally this flat self- 
contradiction is avoided, as it is in the second passage cited 
above, by saying that while Samkhya-Yogas generally hold a 
view not quite orthodox, the wise among them think other- 
wise. Thus: “That twenty-fifth principle which the Sam- 
khya-Yogas as a whole, sarva§ah, proclaim to be higher than 
intellect , buddheh param, the wise declare is a (personal) 
Lord, conditioned and not conditioned, identical both with 
Purusha and with the Unmanifest . . . and this is also the 
opinion of those ivho being skilled in Samkhya- Yoga seek after 
a Supreme ,” paramaisinah, xii, 306, 31-33. In other words, 
such Samkhya-Yogas as admit that the twenty-fifth topic is 
a Supreme Being say that he is our personal God. 

The Twenty-fifth Principle. 

In the passage cited above, xii, 306, 33, the spirit is denomi- 
nated Pancaviihjatika, the twenty-fifth principle. This is the 
last Samkhya topic. But: “The wise say that the twenty- 
fifth creation is a topic and that there is something apart from 
the topics and higher.” Here stands the implication of the 
twenty-sixth principle, in contradiction to the preceding, as 
appears still more plainly in the next section, where 307, 43 
ff., it is expressly said : “ Counting up the four-and-twenty 
topics with Prakrti, the Samkliyas recognize a twenty-fifth 
principle which is apart from the topics ; this twenty-fifth 
principle is said to be the soul without Source or un-Prakrti- 
soul, aprakityatma, when it is enlightened, budhyamanah ; 
and when it thus recognizes self, it becomes pure and apart, 
Brahman from the whole). On the Yoga anudaryanam, see the note above. 



yada to budhyate 'tmanam tada bhavati kevalah. This is the 
correct view according to the topics. Those knowing this 
attain equableness. From direct perception one could under- 
stand Prakrti from guna and topic and so one can judge from 
things without gunas. There is something higher than the 
destructible. They who do not agree to this have a false 
view and do not become emancipated but are bom again in 
manifest form. The unmanifest is said to be the All. But 
the twenty-fifth principle is not part of this ‘ all,’ asarvah 
pahcavihcakah. They that recognize him have no fear.” 

Here there is not an indication of any principle higher than 
the Samkhya twenty-fifth, except as the commentator reads 
Brahman into the word self as “ soul,” but the word is used of 
jlva in the preceding verse, and of Brahman there is not a word. 
The “ thing to be known ” is the “ twenty-fifth principle ” as 
opposed to the Unmanifest, winch is here the “ field ” of 
knowledge. The view of a Lord-principle is distinctly op- 
posed: “It is said that the Unmanifest comprehends not only 
the field of knowledge (as has just been stated in §1. 38) but 
also sattva and Lord; the Samkhya-system holds, however, 
that the twenty-fifth principle has no Lord and is itself the 
topic that is apart from topics ” (that is, the twenty-fifth prin- 
ciple is the supreme principle), 307, 41-42. 

This whole chapter, xii, 307, 26 ff., gives as close an ap- 
proach to Samkhya as is found in the epic. It is called, §1. 
42, the Samkhyadargana, parisamkhyanudargana. That is 
to say, 

Samkhya is Samkhyana. 

Even in the Anuglta, xiv, 46, 54-56, we read: “The or- 
gans, the objects of sense, the five gross elements, mind, 
intellect, egoism, the Unmanifest, and Spirit (these are given 
in nominative and accusative) — on counting up all that 
properly, according to the distinction of topics,- tattva, one 
gets to heaven, released from all bonds. Counting them over, 
one should reflect on them at the time of one’s end. Thus one 
that knows the topics is released, if one abide by the ekanta, 



doctrine of unity.” So in xii, 316, 19, samkhyada^anam 
etat te parisamkhyanam uttamam, “the Samkhya system is 
the best enumeration ; ” evam hi parisamkhyaya samkhyah 
kevelatam gatah, “the Enrunerators by thus enumerating 
attain separateness.” In the same way the Yogin gradually 
emancipates him self by parisamkhyaya, enumerating the steps 
of abstraction, xii, 317, 16. The same thing is found in Gita 
18, 19, where gunasamkhyana or “ enumeration of gunas ” is 
equivalent to Samkhya. Even more strongly is this shown 
when Yoga and Samkhyana are antithetic, like Yoga and 
Samkhya, as in xii, 814, 3 ff., where the samkhyanadar^nah 
are opposed to yoga-pradarginah ; and in xiii, 141, 83 : yukto 
yogam prati sada prati samkhyanam eva ca. 

The Samkhya Scheme. 

As I have shown above, this system stops with the twenty- 
fifth principle. This fact sometimes appears only incidentally, 
as when in xiv, 48, 4, we read : “ By ten or twelve suppres- 
sions of breath one attains to that which is higher than the 
twenty-four.” 1 In its environment this verse is as significant 
as it is grotesque ; but it is simply carried over from an older 
account: “Turning the senses from the objects of sense by 
means of the mind, one that is pure and wise should with ten 
or twelve urgings urge the soul to that which is beyond the 
twenty-fourth principle,” xii, 307, 10-11. Here, at the outset 
of the chapter discussed above, it is evident that no twenty- 
sixth is contemplated. The conditioned soul is to be urged to 
associate itself with the pure soul and abstain from the other 
elements which condition it. This pure soul is declared to 
be the “inner self standing in the breast,” antaratma hrda- 
yasthah, §1. 19, which in Yoga contemplation appears like a 
bright fire. “ It has no source, ayoni ; it stands in all beings 
an immortal thing, and is not seen, but may be known by 
intelligence, buddhidravyena drijyeta. He makes the worlds, 

1 The commentator says ten or twelve, va ’pi may mean and, i. e., twenty- 
two. He gives the exercises. 



standing beyond darkness, and he is called tamonuda, vita- 
maska, the smiter of darkness,” 24. So much for the Yoga 
doctrine, where the inner soul is that “which surpasses the 
twenty-fourth,” and is then treated (as given above) as neuter 
tad or masculine, but without recognition of the Lord-Soul as 
twenty-sixth. 1 Then follows the Samkhya-jnana (parisam- 
kliyanadar§anam), 307, 26 ff. : “It is the system of the Pra- 
krtivadins and starts with highest Prakrti, which is the 
Unmanifest. From this is produced the Great One (neuter), 
intellect, as the second; from the Great One, egoism, as the 
third; and the Samkhyatmadarginah say that the five ele- 
ments come from egoism. These together are the eight 
(forms of) the Source, called the eight sources (because pro- 
ductive). The modifications are sixteen. There are five 
gross elements, vi§esah, and five senses (or the sixteen are 
the five gross elements and ten organs with mind). 2 These 
(twenty-four) are all the topics, tattvas, as explained in the 
enumeration of the Samkhyas. Inversely as it created them 
the inner soul, antaratman, also absorbs them, as the sea 
absorbs its waves. The Source is a unit at absorption and 
a plurality at creation, ekatva, baliutva. The Source itself 
has the principle of productivity, prasava. Over this field 3 

1 This section, like the one cited above (to which it is a parallel), ends with 
yoga esohi yoganam. The next verse (though in the middle of a chapter) has 
the tTpanishad mark of a closed account, yogadarganam etavat (as in Katha, 
etavad anudarganam). The soul appears as a smokeless fire, vidhuma, as in 
Katha, iv, 13, adhumaka ; it is anubhyo anu, as Katha, ii, 20, etc. The point 
of view is wholly that of Atmaism to the very end without a trace of Vishnu- 
ism. It is, however, an intruded section, for the opening of the chapter 
marks a repetition, the questioner saying : “ Now- you have told me all about 
oneness and separateness, but I should like to hear it all again ” (just as the 
Anugita is marked). 

2 So the commentator explains 5I. 29-30, eta prakrtayag ea ’stau vikarag 
ca ’pi sodaga, panca cai 'va vigesa vai tatha pance ’ndriyani ca, etavad eva 
tattvanam samkhyam ahur manlsinah. But see below. 

8 Instead of “ field ” we find also the “ pasture” : “ When the senses (in- 
driyani pramathini, as in the Gita) return from the pasture, gocarah, and 
rest at home, then shalt thou see the highest self with the self, the great all- 
soul” (self), xii, 251, 6. The principle of productivity, prasava, is synony- 
mous with Prakrti. Thus we have prakrti j a gunah (Gita), and prasavaja 
gunah, xiii, 85, 105. 



stands the Great Soul as the twenty-fifth, called the kse- 
trajila, field-knower, also the male, Purusha (avyaktike pra- 
vigate, 38). The field is the Unmanifest, the knower of the 
field is the twenty-fifth principle.” Then follows the extract 
given above. It is clear that here the twenty-fifth principle 
(Purusha) is not a lower principle than a twenty-sixth (not 
recognized at all). Still more remarkable is the following 
exposition : 

In xii, 311, 8 if. : “ There are eight sources and sixteen 
modifications. Metaphysicians explain the eight as the Un- 
manifest, the Great One (mase.), egoism, and earth, wind, air, 
water, and light. These are the eight sources. The modi- 
fications are (the five perceptive organs) ear, skin, eye, tongue, 
and nose ; the five (great elements), sound, touch, color, taste, 
smell ; the five (organs of action) voice, hands, feet, and two 
organs of excretion. [These differences, vigesah, are in the 
five great elements, mahabhiitas; and those organs of per- 
ception are savigesani, that is, differentiated.] Mind, say the 
metaphysicians, is the sixteenth.” The bracketed stanza 1 in- 
terrupts the description (as in the scheme above) with a 
statement of the “ differences ” appertaining to the gross 
elements (as distinct from the fine elements, which have 
only one characteristic apiece, and are avigesa). 

Both these schemes 2 give the Aphorism's list, whereby the 
tattvas of the Samkliya (the Yoga is here expressly included, 
§1. 8) appear as follows : — 

r The Unmanifest 

forms of 


J I 


Five (fine) elements (not here named col- 
k lectively; called tanmatras elsewhere). 

1 ete vigesa rajendra mahabhutesu pancasu buddhindriyany athai ’tani 
savip-sani, Ittaithiia, 311, 14. 

2 Compare xiv, 40, 1 S., where the same creations appear. 




r 5 Organs of Perception (buddhindriyas, §1. 

, 14> 

5 Organs of Action (not here named collec- 
tively; called kannendriyas elsewhere). 

1 Mind. 

5 Gross elements (vigesas, mahabhutas). 

But to the scheme at xii, 311, there is appended the following 
incongruous account, thus, 9I. 16 ff. : “From the Unmanifest 
is produced the Great Soul, mahlin atma, wdiich the wise say 
is the first creation, and call the pradhanika. From the Great 
One is produced egoism, the second creation, which is called 
buddhyatmaka, that is, identical with intellect. From egoism 
is produced mind, bliutagunatmaka, identical with the ele- 
mental constituents, called ahaiiikarika, that is, egoistic, the 
third creation, sargah. From mind are produced the great ele- 
ments, mahiibhutah (sic), 1 the fourth creation, called manasa, 
mental. The fifth creation comprises sound, touch, color, 
taste, and smell, which is called elemental, bhautika. The 
sixth creation is the ear, skin, eye, tongue, nose, called bahu- 
cintatmaka, that is, identical with much thought (matter is 
only a form of mind). The seventh creation is the group 
of organs (of action) after the ear, called organ-creation, 
aindriya. The eighth creation is the up-and-across stream 
(of breaths) called arjavaka, that is, upright. The ninth is 
the down-and-across, also called arjavaka. These are the nine 
creations, sargani, and the twenty-four topics, tattvani, de- 
clared according to the system of revelation (grutinidarQa- 
nat).” So this scheme ends without hint of a twenty-sixth 
principle, but with productive mind and a substitution of 
atman, soul, for intellect. 

A more striking substitution is found in xii, 204, 10-11, 
where, instead of the received order as given above, the list 
from Source to the senses is as follows: 

1 As remarked above, organs and elements are called indifferently indriyah 
or indrivani, mahabhutah or mahabhutani, as shown here and elsewhere. So 
in this passage, sargah and sargani. Compare tattvan, above, p. 98. 

modifica- < 



The Great Unknown, or Unmanifest, avyaktam, makat 
Knowledge, jnana 





In the following section, 205, 16 ff., intellect active in mind 
is mind. It is mind which is freed from the gunas and, ib. 9, 
mind, as a form of knowledge impeded by the gunas, pro- 
duces intellect, which must be withdrawn into mind again for 
one to attain the highest. In these cases, there can be. from 
a synthetic point of view, no unsystematic interpretation of 
intellect and knowledge and mind, but a loose 1 exploiting of 
Sarhkkya in terms of Brahmaism, because elsewhere the Saiii- 
kliya scheme is fully recognized. So carelessly are the terms 
employed that, while in one part of the exposition knowledge 
is Brahman and mind is a part of it, related to it as jiva is to 
Atman, in another part we are told that this knowledge comes 
from something higher, the Unmanifest. Again. Brahman is 
not the Unmanifest but in the Unmanifest. xii, 319. 1. There 
is no substitution for egoism in the above, for this is recog- 
nized in another stanza which enumerates as the “group 
called bhutas,” (created) spirit (!), Source, intellect, objects 
of sense, the organs, egoism and false opinion, 205, 24. 2 Here 

1 These para ladders (compare Gita, 3, 42 ; Ivath. iii, 10) are found every- 
where and often contradict the regular schemes: “Soul is higher than mind, 
mind than senses, highest of creatures are those that move : of these the 
bipeds ; of these the twice-born ; of these the wise, of these those that know 
the soul, atman ; of these the humble,” xii, 298, 19 ff. ; “Objects arc higher 
than senses, mind higher than objects, intellect higher than mind, the great 
Atman higher than intellect,” xii, 247, 3 ff. (in 249, 2 paro matah for mahan 
parah) ; “ The unmanifest is higher than the great ; the immortal is higher 
than the unmanifest: nothing is higher than the immortal” (ib.l. The stages 
in xiv, 50, 54 ff., are space or air, egoism, intellect, soul, the unmanifest, and 
spirit ! 

2 This is called the samuho bhutasamjnakah, or “ group of so-called 
created things,” which is noteworthy as containing Purusha, spirit, and abhi- 
mana, false opinion, as a distinct factor. 



the source of the Source and of Purusha alike is Brahman, a 
view utterly opposed to the passages cited above. 

The Anugita, which, as already indicated, also has the 
schemes above, continues in xiv, 42, with a parallel to xii, 
314, on the relation of the elements to the individual, as 
organ, to the object, and to the special deity concerned with 
each action. At the opening of the eighth chapter of the 
Gita adhyatma is called the individual manifestation. It is 
literally that connected with the self or soul, and is often 
used as a noun in the sense of metaphysics (xii, 194 and 248, 
etc.). 1 2 In xii, 314, 4 and 14, it is said that an explanation as 
the Samkhyas represent it, yatha saihkhyanadarcinah, is given 
of the manifestations according to the individual, vyaktito 
vibhuti, which differs somewhat from that in the Anugita. 
The scheme is as follows, starting with the elements and 
with akacja, air, as the first bhuta in the latter account: 



Light Water 






eye tongue 


organs of 




touch (ob- 
ject of) 

color taste 







Sun Soma 







upastha hands 


organs of 




nanda (gukra) doing, 






Prajapati Indra 






Intellect * 








or thinking 




Rudra, or In- 

or Brahman 


1 Compare the use of these terms in BAU. iii, 7, 14. On adhyatma in 
this sense, compare also xii, 331, 30, adhyatmaratir asino nirapeksah . . . 
atmanai ’va sahayena ya? caret sa sukhi bhavet. 

2 buddhih sadindriyayicarini, “ directing the six senses ” (usually a function 
of mind, which is here pancabhutatmacarakam), xiv, 42, 20, and 31. The 
function of intellect is here mantavyam, which in C'anti is given to mind. 
Rudra in the preceding group in Anugita is replaced by buddhi in £anti, 
where buddhi is both adhyatma and adhidaivata. The adhidaivata of intel- 
lect is spirit, ksetrajna, in ^iinti : Brahman, in the Anugita. It is apparent 
that we have here (a) rather late matter, (b) worked over by two sets of 



This scheme is unknown in the older Upanishads. Even 
egoism thus appears first (with some variations) in Pragna, 
iv, 8 (Deussen). Compare xii, 240, 8, above, where Fire is 
the divinity to digestion, not to voice, and Sarasvatl is assigned 
to the tongue. When, as often happens, no egoism is men- 
tioned, it is because the intellect (“ the twelfth ” as it is called 
in the very passage which gives thirteen above, xiv, 42, 16, 
and in the Pancagikha schemes given below) is held to imply 
egoism. The frequent omission, however, seems to point to 
the fact that there was originally no distinction, or, in other 
words, that intellect was primarily regarded as necessarily 
self-conscious as soon as it became manifest at all. 

The Twenty-Sixth Principle. 

Clearly as most of the schemes given above reveal the fact 
that the twenty-fifth principle, or in other words pure Ego, 
was regarded as the culmination of the group of systematized 
categories, the intrusion into this scheme of a new principle, 
overlapping the twenty-fifth, is here and there made mani- 
fest. This new principle is the one denied in the Sariikhyan 
scheme, namely that of a personal Lord, Tgvara, which is 
upheld in the contrasted Yogin scheme. This twenty-sixth 
principle is explained in xii, 308; after the speaker says he 
has disposed of the Samkhya system. Here the male condi- 
tioned spirit bewails his intercourse with the female Source, 
and the fact that associating with her he has not recognized 
that he has been “ like a fish in water,” a foreign element in 
combination with matter, and consequently is reborn again 
and again, §1. 24-26; but now he becomes enlightened, 
buddha, and will reach unity, as well as likeness with the 
Lord-spirit, the indestructible, 27-40. The twenty-sixth 
principle is thus recognized not only as the one eternal prin- 
ciple, but as a personal spirit, ayam atra bhaved bandhuh, 27. 
Then follows another exposition, winch is based on the system 
of Narada, received by him from Vasistha, who in turn re- 
ceived it from Hiranyagarbha, 309, 40. This system is both 
Yoga and Samkhya, the systems being double but the teach- 



mg being identical (yacl eva gastrarii Sariikhyanam yogadar- 
ganam eva tat, 308, 44), the claim usually made when Yoga is 
advocated. A huge Castra is that of the Samkhyas, “ as say 
viduso janah,” and one “to which, along with the Veda, 
Yog ins have recourse.” In other words, the Yoga teaching 
is based on Veda and on the Sahikhya as a precedent system. 
Then follows the admission : “ In it (the Samkhya system) no 
principle higher than the twenty-fifth is recognized,” (asmin 
gastre) pahcavingat paraiii tattvam pathyate na, naradhipa, 
whereas: “The Yoga philosophers dechire a budhyamana or 
individual spirit and a buddlia or Lord-Spirit to be hi accord- 
ance with their principles, the hitter being identical with the 
former, except that it is fully enlightened,” gl. 48. 

Here also is a perfectly clear and frank statement, which 
may be paraphrased thus : “ In older Saiiikliya philosophy the 
highest principle recognized is that of the pure individual 
Ego; in the Yoga philosophy this Ego is identified as indi- 
vidual spirit with the fully enlightened Lord.” Hence Yogas 
(and not Samkhyas) speak of budhyamana and buddlia as 
two but identical, budhyamanaiii ca buddham ca prahur yoga- 
nidarganam, gl. 48. Elsewhere the twenty-fifth principle is 
itself the Lord: aliam purusah paficavingakah. 1 * * * * * 

After this introduction the speaker, Vasistha, proceeds to 
describe this Yoga philosophy in detail. The Lord-Spirit 
“divides himself into many 7 ,” atmlinam bahudha krtva, and 
becomes the different abuddhas, or imperfectly enlightened 
spirits conditioned by Prakrti. Thus he becomes conditioned, 
gunan dharayate, and “ modifies himself ” -without true knowl- 
edge of himself, vikurvano budhyamano na budhyate. In 
this condition, then, he becomes creator and absorber of what 

1 Compare xii, 340, 43, personal God is the twenty-fifth. He is the witness 

devoid of gunas, and of kalas, ib. 23 ; “ the twenty-fifth, beyond the twice 

twelve tattvas,” ib. 24. In this passage the Unmanifest is resolved into Puru- 

sha, 340, 30-31. This is worth noting as being in direct contradiction of the 

theory of unchanging eternal Prakrti, as enunciated in xii, 217, 8: “Both 
Purusha and the unmanifest Source are eternal, without beginning and with- 

out end.” In 335, 29-31, Source is both born and indestructible. Compare 

H. 3, 85, 16, as cited above, p. 98. 



he has created. The conditioned cannot understand the 
unconditioned; it is the Un-understanding, apratibudliyakam 
(sic, 309, 4). The conditioned spirit can understand the 
Unmanifest but “he cannot understand the stainless eter- 
nal buddha, which is the twenty-sixth principle/’ sadvingaiii 
vimalam buddham sanatanam, though the latter “ understands 
both the twenty-fifth and the twenty-fourth principles,” 309, 
7. “This twenty-sixth principle is pure unmanifest Brah- 
man, Avliich is connected with all that is seen and unseen,” 
ib. 8. “When the conditioned spirit recognizes the pure 
Highest Intelligence, then he becomes clear-eyed, avyakta- 
locanah, and free of the Source” (tadii prakrtiman, sic, read 
apra?). The twenty-sixth is this Highest Intelligence; it is 
“the topic and that which is apart from all topics,” (,‘1. 10 and 
13. “The conditioned spirit attains likeness with the twenty- 
sixth principle when it recognizes itself as the twenty-sixth,” 
sadvingo 'ham iti prajnah, gl. 16. “ That separateness of spirits 
which is part of the exposition of Sariikhya is really (ex- 
plained by) the conditioned spirit when not fully enlightened 
by the (fully) enlightened twenty-sixth,” sadvingena pra- 
buddhena budhyamano 'py abuddhiman, etan nanatvam ity 
uktarii saihkhyagrutinidarganat, gl. 17. The continuation of 
this teaching points out that unity with Bralunan is attained 
by the individual spirit only when it no longer has any con- 
sciousness (of self), yada buddhya na budhyate, §1. 18. 

In this passage the attempt to reconcile the doctrine of the 
Samkhya individual spirits, nanatvam, “ than which there is 
nothing higher,” with the doctrine of unity, ekatva, is as plain 
as a reasonable historian could expect to find it. “ Thus it 
is,” the account concludes, “that one must understand the 
(two theories of) separateness and unity,” nanatvaikatvam 
etavad drastavyam gastradarganat, gl. 22. And then occurs 
a very pretty lapsus. The images of the fly encased in the 
plant, magakodumbare, and the fish in water, matsyoclake, 
are constantly employed in Samkhyan pliilosophy, as shown 
above, to illustrate the fact that spirit is different from the 
Source, though externally united. Our good Vasistha, how- 



ever, brings these images in to illustrate the difference, anyat- 
vam, between the individual spirit and Brahman : “ Tlie 
difference between the fly and plant, between the fish and 
water, is to be understood as the combined separateness and 
unity of these two,” as if, from the historical connotation of 
these images, they were essentially different, whereas according 
to the exposition they are essentially one. But this is of a piece 
with the use of vikurvanas, a Samkhya term applied to the 
modifications of the Source, when used above, of Brahman. 

This Yoga doctrine, as explained above, is to be taught 
(not to the man that bases his philosophy on the Veda, na 1 
vedanisthasya janasya . . . pradeyam, but) “to any one that 
desires it for the sake of wisdom and receives it with sub- 
mission,” §1. 32. 

The Yoga doctrine as here represented stands midway 
between Samkhya and Brahmaism. The former side has been 
fully illustrated. In regard to the latter it will have been 
noticed that while the personal Lord-Spirit is a form of 
Brahman, and Brahman in turn is identified with the pure 
essence of every individual spirit, it is merely said that 
Brahman is connected with the visible as well as with the 
invisible, drgyadrgye hy anugatam, 309, 8. The Bralmian here 
represented is not the All, but a pure Supreme Spirit into 
which fractional spirits, parts of Brahman when he “made 
himself many,” are reabsorbed. Of the identity of the objec- 
tive world with this Bralmian there is no word ; neither is 
there any hint that the objective world is illusion, except that 
at the beginning of the preceding section, 308, 2 ff., the gen- 
eral opinion, ahuh, is cited that “the Unmanifest is igno- 
rance,” avidya, as opposed to the twenty-fifth principle as 
wisdom, vidya. 2 Elsewhere “ the Source is knowledge,” jnana, 
but also aveclyam avyaktam, as opposed to (jneyo) vedyah 
purusah, 319, 40. 

1 But na°, v. 1., N.,“to one wise in the Veda it may be imparted or to,” 
etc. Those excluded are given in the following verses as liars and other evil- 
doers, a long list. 

2 But ib. 7, the Source as unmanifest is vidya ; the highest is Vidhi (com- 
pare pradhanavidhiyogasthah of £iva, xiii, 14, 423), the Creator. 



This doctrine of the twenty-sixth principle belongs only 
to the later part of the pseudo-epic. The passage given 
above is found virtually repeated in xii, 319, 56, and 70 ff. 
Here as Prakrti the chief-tiling, prailhana, does not know 
spirit, so spirit does not know Supreme Spirit. “The one 
that is different (spirit), seeing and yet not seeing, looks 
upon the twenty-sixth, the twenty-fifth (pure spirit) and 
twenty-fourth. But the twenty-fifth also does not recognize 
the twenty-sixth, who recognizes him, and having a false 
opinion of himself thinks that no one is higher than he ” (so 
316, 4). And further: “The twenty-fourth should not be 
accepted by wise men (as the twenty-fifth), any more than, 
because of mere association, the fish should be identified with 
the water it has entered (74). The twenty-fifth on realizing 
that it is different (from the twenty-fourth) becomes one with 
the twenty-sixth and recognizes (the latter). For though 
The Best appears different from the twenty-fifth, the saints 
regard this as due to the conditioned nature of the twenty- 
fifth and declare that the two are really identical. Therefore, 
being afraid of birth and death, and beholding the twenty- 
sixth, neither Yogas nor Saihkhyas admit that the twenty- 
fifth is the indestructible.” 

Here again, with the new notion that jiva is destructible (in 
Paramatman) there is the attempt to foist on the Samkhya 
the belief which has been formally denied to them. Similarly 
in the Aniruddha theology, of the personal Lord Govinda, 
who is said to “create the elements,” xii, 207, 7 ff., it is said: 
“From him whom Samkhya and Yoga philosophers declare as 
Highest Soul, Paramatman, and who is called the Great Spirit, 
mahapurusa, is derived the unmanifest, avyaktam, of which 
he is the base, pradhanam. From the unmanifest Lord, 
Igvara, came the manifest, and he is Aniruddha, called the 
great Sold. As egoism he created Brahmdn and the elements, 
and then the gunas,” xii, 341, 28-33. 

In this copy of the preceding passage there is also no 
notion of Vedanta as implying Maya or illusion. Significant 
is the fact that the present teaching is represented in the fol- 



lowing stanzas, 319, 84-86, as being newly inculcated, and 
especially designed for those who desire emancipation, in con- 
trast to the Samkhyas and Yogas, who are content with their 
own doctrines, dliarma. 

It is thus clear that Sarhkhya is merely a name to appeal to, 
and stands in this regard on a footing with Yeda, an authority 
claimed for the most divergent teaching. 

O O 

Maya, Self-Delusion. 

The “illusion” theory of the universe is a development 
from the simple idea of delusion, often self-delusion. The 
ordinary (non-pliilosophieal) epic maya is a trick of delusion. 
Gods indulge in it to overcome then- enemy. The illusion- 
god par excellence, Vishnu as Krishna, thus deludes his 
enemies by making them think the sun has set when it has 
not, or by parallel magic tricks. 1 This, in my opinion, 2 is the 
only meaning in the older Upanishads, Indro mayabhili puru- 
rupah, Brh., ii, 5, 19 (from the Rig Veda), “Inclra multi- 
form through tricks of delusion ; ” na yesu jihmam anrtam 
na maya ca, “ in whom there is naught crooked, nor untrue, 
nor any trick,” Pragna, i, 16. Magic seems to be the mean- 
ing (parallel with moha) in Miiitrl, iv, 2, where occurs the 
indrajala-maya of Mbh. v, 160, 55. 

In Gita 7, 14-25, maya is a divine, dam, delusion caused 
by the gunas, gunamayl, characterizing people wicked and 
foolish ; in 4, 6, it is a psychic delusion, atmamaya, which 
causes the unborn God by means of Prakrti to appear to be 
born (not, be it noticed, which causes the not-soul to appear 
to be real). It occurs in one other passage, 18, 61, where it 
is the equivalent of moha in the preceding stanza (as in 
Maitrl Up., above). In all these passages, although it is 
possible to read into maya the meaning given it by Oaihkara, 
for example, yet the simpler meaning suffices of either trick 

1 This is called indifferently maya (chadma) or yoga, v, 100, 54-58; vii, 
146, 68, etc. 

2 In this interpretation of maya I am forced to differ from that of Deussen, 
who holds that maya is Vedantic Illusion (i. e., the not-soul appears through 
divine Illusion to be real) even in the earliest scriptures. 



or delusion (false understanding) applied to tlie relation of 
individual soul and God, and this is probably tlie meaning, 
because maya as illusion plays no part in the development of 
the scheme. Guna-made delusion is the regular Samkliya 
Prakrti-made ignorance; it is not Prakrti's self. 

The expression used above of Krishna’s maya that it is 
“ divine,” has no special philosophical significance. The same 
phrase is applied to Duryodliana's water-trick, daivun mayam 
imam krtva, ix, 31, 4. When, too, Krishna in the Gita says 
that he is born by atmamaya, it must be remembered that in 
describing the parallel situation in the llamfiyana, where 
Vishnu is born as llama, the word chadman, disguise, cover, 
is used as the equivalent of maya, G. vi, 11, 32. 

In a very interesting critique of the new doctrine of moksa, 
that is, salvation without Vedie sacrifices, an orthodox objector 
is represented as saying: “This doctrine of salvation has 
been brought out by miserable idle pundits; it is based on 
ignorance of the Veda and is a lie under the guise of truth. 
Not by despising the Vedas, not by chicanery and delusion 
(mayaya) does a man obtain great (Brahman). He finds 
Brahman in brahman ” (Veda). 1 

Similarly, when DraupadI philosophizes in iii, 30, 32, her 
opening words show that she reveres as the chief god the 
Creator, who, like other creatures, is subject to transmigration, 
32, 7, and is in no respect an All-god, though a later rewrit- 
ing of the scene mixes up Bhagavat. Igvara, and Prajapati. 2 
This god, she says, has deluded (moha) her husband’s mind 

1 As the section is occupied in advocating the one-soul (All-soul), aikat- 
mya, doctrine, it is clear that mava is here merely delusion or deceit, xii, 
270, 50-51. The words of the text are : yriya vilunair alasaih panditaih sam- 
pravartitam, vedavadaparijiianam satyiibhasam iva ’nrtani . . . na vedanam 
paribhavan na jathyena na mayaya mahat prapnoti puruso brahmani brahma 
vindati, xii, 270, 17, 19. Kapila, to whom the remark is addressed, admits 
“the Vedas are authoritative,” vedah pramanam lokaniim, 271, 1, but, 42, 
insists that, though “ everything is based on the Veda,” the cruel animal sacri- 
fices therein enjoined are objectionable (as cited above), and upholds the 
thesis that “knowledge is the best means of salvation,” jnanam tu pararna 
gatih, 271, 38 — this by the bye. 

2 The revision appears clearly at the end in Draupadl’s conversion. Com- 
pare the comments, AOS., Proceed., March, 1894. 



and in deluding men generally, mohayitva, the Lord shows 
the power of his delusion, miiyaprabliava, which deludes them 
by atmamaya (the same expression as that of the Gita, cited 
above), making them kill each other as blind instruments of 
his will, which act without volition, just as a stone breaks 
another in the hands of a man. Man proposes, but God dis- 
poses 1 by means of a trick, chadrna krtva, 30, 36, “ playing 
with men as children play with toys.” “ Fie, fie,” says her 
husband, “ don’t speak so of the Lord, through whose grace 
the faithful gets immortality,” 31, 42; “for these things are 
divine mysteries (devaguhyani, rewards of good and evil), 
since the divinities are full of secret tricks,” gudhamaya hi 
devatiih, 31, 35-37. The Qastras and faith, not magic, maya, 
or sinful works, give faith in Krishna, v. 69, 3-5. 

Again, in the account of the Pancakalajnas, the visiting 
Hindus, who look with awe on the service paid to the One 
God, say that they could hear the hymn, but could not see 
the god, because, as they suppose, they were “ deluded by the 
god’s maya,” mohitas tasya mayayii, xii, 337, 44-48. God in 
the following is called the maliamayadhara, as he is also called 
by the rather modern epithets caturmaliarajika, saptamaha- 
bhaga, 2 xii, 339, 3 ff. Here maya is truly illusion, as it is said 
in 340, 43—45 : “ God is he by whom this illusion (of visible 
God) was created,” maya hy esa maya srsta yan mam pagyasi, 
Narada; but it is not illusion embracing the world of objective 
things, even in this late account (careless enough, for example, 
to construe iti vai rnenire vayam, 337, 38). There is at least 
no passage in the epic which says bluntly that “Prakrti is 
maya,” as does ()vet. Up. iv, 10. On the contrary, the great 
mass of epic philosophy, though it teaches that the sinner is 
deluded “ by Vishnu’s hundred mayas,” 302, 59, teaches also 
that this delusion is merely a confusion of mind in respect of 
the relation of the pure soul to the conditioned soul. It does 
not teach that those things which condition the soul are an 

1 anyatha manyante purusas tani tani ca . . . anyatha prabhuh karoti 
yikaroti ca, iii, 30, 34. 

2 He is also called akhandala, which in xii, 337, 4, is still an epithet of 



illusion, but that they are eternal substance, either in them- 
selves or as parts of Brahman. Take for instance the long 
account in xii, 196 to 201. It is not suggested that the sin- 
ner divest himself of illusion. He goes into moha, that is he 
becomes confused, and again he enters Brahman, 197, 10 ; or 
“enjoys bliss,” ramate sukham (“if he does not wish the 
highest, because his soul is still tinged with desire, ragatma, 
he attains whatever he desires”). 1 Knowledge is Brahman, 
and hence one must be free of all delusion to be Brahman 
indeed, and truly immortal, 2 but the objective world is seldom 
an illusion of Brahman. Moreover, the avidya of God is 
clearly an afterthought. According to one section in ('anti, 
God creates the world “ at the point of day ” through avidya 
or ignorance. First mahat was born, “ which quickly became 
mind ” (where mind and not intellect is vyakta, manifest), 
which is “ characterized by desire and doubt.” 3 This same 
account in its first form is found in 232, 32, without avidya : 
“ The Lord, Igvara, sleeps during the cataclysm sunk in med- 
itation, dhyana ; but, when awakened at the close of night, he 
transforms the eternal, vikurute brahma ’ksayyam, and pro- 
duces the Great Being, whence mind, one with the manifest.” 
The following section simply picks up this account, repeats 
it in almost the same words, but slips in avidya to explain the 
expression “ creates.” The alteration is the more marked as 

1 Some very grotesque conceptions are expressed here. In 200, 25, the jiva 
soul goes to Atman ; or goes to heaven and lives separately. When as a flame 
the spirit ascends to heaven, Brahman like a courteous host says “ Come, stay 
with me,” makes it (or him) conscious and then swallows him ! 

2 “ Sorrow is the end of joy as night is the end of day, joy is the end of 
sorrow, as day is the end of night” (these succeed each other and each has its 
end) ; “ only knowledge ends not, for knowledge is Brahman,” xiv, 44, 18, 
20-21 ; 47, 1. Not till 52, 9, i. e., after the Anugita, is finished, is Maya a factor 
here. Previously there is only the ghoramoha or horrible misunderstanding 
of truth, xiv, 45, 4, etc. In xviii, 3, 36, Indra’s ruaya is an optical delusion. 

3 xii, 233, 1 ff. Here is to be noticed a contradiction in epic psychology. 
Mind in this passage has prarthana and sisrksa, that is it desires, whereas 
elsewhere desire (the unexplained “ seventh,” xii, 177,52) is an attribute of 
egoistic intellect. Desire is born of imagination, samkalpa, xii, 177, 25 ; it is 
destroyed by avoiding this, 302, 56 ; but, “ remove mind from samkalpa and 
fix it on self,” 241, 17. 



many tests make no division of chapters here. In either case 
the account of creation goes right on, first, 232, 32, stated as 
(l9varah) : 

pratibuddho vikurute brahma ’ksayyam ksapaksaye 
srjate ca mahad bhutam tasmad vyaktatmakam 

and then as : 

brahmatejomayaih oukrarii yasya sarvam idam jagat 
ekasya bhutam bhutasya dvayarii sth&varajangamam 
aharmukhe vibuddhah san srjate ' vidyaya jagat 
agra eva mahad bhutam acu vyaktatmakam manah. 

As the seven creators 1 mentioned in the following stanza, 
233, 3, are explained as intellect, mind, and the five elements, 
it is clear also that egoism as a distinct factor is omitted. The 
seven cannot create apart, so they unite and make the body 
which the “ great beings,” bhutani mahanti, enter with Karma. 
The adikarta, First Creator, is Prajapati, who acts without 
Maya, §1. 13. 2 In short, while sometimes recognized, Maya 
is generally unknown in the epic, because the epic lacks unity, 
being now and then Vedantic, but generally Yogaistic. 

Panca9ikha’s System. 

In the presentation above I have analyzed the three differ- 
ent religious philosophies advocated in the pseudo-epic ; the 
Saihkhya, which holds to spirit and Source as distinct immor- 
tal entities; the Yoga, which adds the Supreme Spirit; and 
the personal religion of Narada and others, which makes of 
the Paramatman or Supreme Spirit a modified form of Brah- 
man known as Aniruddha, etc., and identified with Krishna. 
In xii, 352, 13, the Paramatman doctrine is declared to be the 

1 manasa, “ mind-creatures,” the same epithet as that applied to the eternal 
Deva in xii, 182, 11. Compare BAU. ii, 5, 7 ; Gita, 10, 6. 

2 saryabhutany upadaya tapasap caranaya hi adikarta sa bhutanam tain 
era ’huh prajapatim. The commentator explains “by means of Maya” 
(BAIT, ii, 5, 19), but there is not even the suggestion of the Maya doctrine here. 
The etymology in pi. 11 (te . . . parfraprayanam praptas tato purusa ueyate) 
seems to be owing to a confusion with puripayam purusam iksate, Prap. y. 5. 



opinion of some Pundits only, in distinction from that of the 
knowledge-philosophers, who are said to hold to unity of soul. 
However this passage may be interpreted, 1 it is evident that 
it distinctly sets over against each other the Yoga and Brah- 
man interpretation. Paramatman is identified with Vishnu 
the “ unconditioned, All-soul spirit.” The religion taught 
is expressly opposed, as sometliing higher, to Samkhya and 
Yoga (§1. 7-8), and by comparison with other schemes is of 
Pahcaratra character. A preceding section states that the 
same religion is identical with the doctrine taught to Arjuna 
in the Gita, 349, 8, and (as already noticed) it is here called 
“ the Krishna religion,” Satvata dliarma, which has mysteries, 
abstracts, and an Aranyaka (ib., 29-31). It was handed down 
through the seers, and a priest who was acquainted with the 
(Jyestha) Simian (and) Vedanta. His name was Jestlia (sic). 
Then it disappeared, to be promulgated again in the Harigltah, 
ib. 46 and 53. In it, Vishnu as God is adored in one, two, 
three, or four forms (the usual group is meant, Aniruddha, 
Pradyumna, Saihkarsana, Vasudeva). 2 The disciples are called 
“ those devoted to one God,” ekantinas, and it is hard to find 
many of them (durlabhah, 349, 62, compare Gita, 7, 19). 
They are identified with the Pancaratras (so 336, 25), a sect 

1 The words seem to indicate the antithesis not of three but of two beliefs : 
evam hi paramatmanarii kecid icehanti panditah, ekatmanaih tatha ’trnanam 
apare jnanacintakah, tatra yah paramatma hi sa nityam nirgunah smrtah, sa 
hi Narayano jiieyah sarvatmapuruso hi sah. The commentator, however, 
may be right in taking atman to refer to Sariikliyas and ekatman as bralima- 
bhinnam (Vedanta), though the single subject would make it more natural 
to take ekatmanam atmanam as “one spirit which is alone.” Vishnu here is 
the manta mantavyam, “ the thinker and the thought,” and the eternal fore- 
cause, pradhana, gl. 17-18. In gl. 22, God plays, kridati, in his four forms (as 

2 £iva, on the other hand, has eight forms (the Puranic view), which, accord- 
ing to the commentator (though murti may imply the incorporations, Rudra, 
Bhairava, Ugra, Igvara, Mahadeva, I’agupati, farva, Bhava), are the fire ele- 
ments, sun, moon, and Purusha, iii, 49, 8. Such divisions are often unique 
and apparently arbitrary. See below on the eight sources. “ Indestructible 
Brahman” (like Sattva) is eighteenfold according to (xii, S42, 13) II. 3, 14, 
13, astadagavidham (or nidham). Eight and a thousand (only pseudo-epic) 
are qiva’s names, against Vishnu’s even thousand. The “ worlds ” are eight 
(see below), or seven, or twenty-one, according to the passage. 



the teaching of which is here identified not only with that of 
the Sarhkhya-Yoga, but also with that of Vedaranyaka, ib. 
349, 81, and with the religion of the “ white men ” and Yatis, 
gvetanam Yatinam ca, ib. 85. Compare 336, 19, the white 
men’s religion, and Satvata Vidhi, declared by Surya. 

The difference between religion and philosophy is obliter- 
ated in India, and the Pahcaratra, sect is exalted as a develop- 
ment of the Bhagavadbhaktas, as the latter are represented 
in the Gita, clearly an indication of posteriority ; while their 
philosophy is rather contrasted than identified with that of 
the Samkhya. 

Three expositions are given, which embody the same ter- 
minology, and may be called the Pancagikha system. 

Panea 9 ikha Kapileya (interpreted as a metronymic !) ap- 
pears in xii. 218, 6 ff., and 320, 2 ff. His punch-name is 
elaborately amplified in the former passage, where, 218, 10 ff., 
he is an incorporation of Kapila and the first pupil of Asuri. 
In Panetfsrotas, where there is a Kapila mandala, he holds a 
long “session,” satra, having “bathed in the pa/Icwrsrotas ” 
(five rivers of the mind ? cf. £vet. 1, 5), and being versed in 
the Puflraratra (doctrine), and being called in consequence 
not only paacaratraviQarada, but also 

pancajSah pancakrt panca-gunah pancacikhah (smrtah), 

epithets which are duly interpreted by the omniscient Nlla- 
kantha. He also (below) has the epithet Paiiearatrah, which 
is the only one that need concern us, as the interpretation of 
the others is mere guesswork. Panca§ikha is regarded, then, 
as the teacher of the new sect of Pancaratras. 1 

His doctrine rests on the ancient foundation of “disgust 
with birth, disgust with acts, disgust with all things,” sarva- 
nirveda, and is, in short, the religion of ennui, which consists 

1 The seven Citrafikhandins are referred to as the author of the Vinca- 
ratra Qastra in 336, 27 ; 337, 3, yastram citragikhandijam. These are the 
seven Prakrtis, personified as the seven old sages, whose names are given below, 
p. 170, to whom is added Manu to make the “eight sources,” 336, 29. In 
the hymn at xii, 339, the god is called Pancakala-kartrpati, Pancaratrika 
Pancagni, Pancayajna, Paiicamahakalpa (as also Citrayikhandin). 



in a little more than mere indifference. The literal meaning 
is that one “ finds oneself out of,” or is sick of, the round of 
birth and death. Nirvana is attained by nirveda. 1 This dis- 
gust and the rejection of that untrustworthy delusion, ana§va- 
siko mohah, which leads to religious practices and the hope of 
rewards, xii, 218, 21-22, is the starting-point of the system, 
which, synthetically considered, should culminate in Ivrishna- 
Vishnu, as the be-all and end-all, as in other cases. 

The analysis of the system is preceded by a most interest- 
ing and historically important review of certain fallacies, as 
follows. The unbeliever says : “ One who relies on tradition 
(the scripture) says that there is something beyond after the 
destruction (of the body), as being obvious and seen by all ; 
but such an one is refuted by the fact that death of self is 
negation, deprivation, of self, anatma hy atmano mrtyuli. 
Death is a weakness induced by age. Through delusion one 
imagines a sold, and this is erroneously regarded as the 
“something beyond” (or higher). For practical purposes 
one may assume what is not true (that there is no death of the 
soul), just as one may say that “ the king never dies,” ajaro 
'yam amrtyug ea raja ’sau. But when something is asserted 
and denied and no evidence is given, on what should one base 
a judgment? Direct observation (evidence of the senses) is 
the base of received teaching and of inference. Received 
teaching is destroyed by direct observation, and (as evidence) 
inference amounts to nothing.” 

The last sentence reads in the original, 218, 27 : 

pratyaksam hy etayor mulaiit krtantaitihyayor api 
pratyaksena 'gamo bhinnah krtauto va na kimcana 

The commentator takes krtanta as anumana and aitihya as 
equivalent to agama ; though in 240, 2, anagatam anaitihyam 
katham brahma ’dhigacchati (where the commentator says that 
agata is pratyaksa and anumana), “ How can a good man 

1 Compare xii, 189, 16-17: “One cannot know the unknown (if faith he 
lacking); keep the mind on faith; hold it to the vital air; the vital air to 
Brahman; nirvana is attained by nirveda;” Gita, G, 23, nirvinnacetasa yogo 
(yoktavyo niycayena ca) ; Mund. Up. i, 2, 12, brahmano nirvedam ayat. 




attain to Br ahma n not known to tradition nor revealed in the 
Veda?” 1 and in G. v, 87, 23, aitihyam anumanam ca prat- 
yaksam api ca ’gamam, ye hi samyak parlksante, it is distin- 
guished from the latter. The word agarna is of sufficient 
importance to note the epic’s own definition given in xii, 270, 
43 : agamo vedavadas tu tarka§astrani ca ’gamah, “ Received 
(scriptural) teaching includes the words of the Veda and 
philosophical codes ; ” a remarkable definition in view of the 
fact that some of the latter are heterodox, and that agama is 
currently used as equivalent to right tradition. The tarka- 
vidya is elsewhere differentiated from logic, anviksikl, though 
both are called useless, xiii, 37, 12, when not extolled, as 
often ! 

The next stanza continues : “ Enough of making assump- 
tions based on this or that inference. In the opinion of (us) 
unbelievers there is no other ‘ spirit ’ than the body.” 

For clearer understanding of the historical value of this I 
must give the exact words, 218, 28 : 

yatra yatra humane 'smin krtam bhavayato 'pi ca 
na hyo jlvah qarlrasya nastikanam mate sthitah 

Here krtam bhavayatah in the meaning of bhavanaya'lam (N.) 
is even more careless than the following genitive with Qari- 
rasya ; but both are indicative of the slovenly style which 
belongs alike to the Puranas and the pseudo-epic. 

The unbeliever (according to the commentator - ) continues 
with a stanza almost unintelligible in its Sutra-like concise- 
ness, which can be given only by the original : 

reto vatakanikayaih ghrtapakadhivasanam 

jatih smrtir ayaskantah suryakanto 'mbubhaksanam 

“ The seed in the banyan-flower (accoimts for the delusion of 
soul) ; butter (is only another form of grass) ; rum (is but 
fermented rice). Memory (and other ‘ psychic ’ functions are 
identical with the) creature born. 2 (The ‘ soul ’ is like the) 

1 Just below, 240, 3, the expression manasaj ce ’ndriyanarh ca aikagryam 
may be noticed as a repetition phrase of iii, 200, 25. 

2 I take adhirasana in the sense of adkivasa, home : (consider) the origin 
of ghee and fermented (liquor) ; N. paraphrases, adhivasitat (add in pw.). 



magnet (which moves iron not by psychical but by physical 
potency). 1 The burning-glass (makes lire, and so the fiery, 
active, soul is but a physical phenomenon). (The fire's) 
devouring of water (is typical of the so-called appetite or 
desire of the soul),” or, in other words : Desire and enjoyment 
are no proof of a superphysical entity, any more than in the 
case of a fire gratifying its tliirst for water. 

The denial of the soul-doctrine next calls forth the follow- 
ing refutation : 

“ A passing away (of something not physical occurs) in the 
case of a dead being. Supplication of the gods (proves the 
existence of incorporeal entities). (There would be besides) 
hi the case of the dead a cessation of acts [the Karma doctrine 
would have to be given up]. 2 This is the proof. (Then 
again) things incorporate cannot be causes, hetavali, for there 
is no identity of that which has form and that which has no 
form,” 218, 30-31. 

After this, other sceptics, who the commentator rightly (as 
I think) says are Buddhists, 3 are introduced with a new argu- 

Jatih smrtih, “birth and memory,” would seem to imply that memory argues 
a former birth, as in Patanjali’s Sutra, iv, 9. This would be an argument on 
the other side, as if the stanza were writ to prove the opposite. I follow X , 
though inclined to think that the words really ought to be put into the mouth 
of the believer (tree, butter, memory, etc., show soul). See the next note 

1 But compare the (orthodox) view as explained in xii, 211, 3: “As sense- 
less iron runs toward a magnet ; so conditions born because of one’s nature 
and all else similar” (are attracted toward the soul). The passages seem 
curiously related, as just before stands, yl. 2, yatlia ’yvatthakanikayam antar 
bhuto mahadrumah nispanno dryyate vyaktam avyaktat sambhavas tatha, 
“ birth from the unmanifest is as when a great tree born in a flower coming 
out is seen clearly.” Compare BAU. iii, 9, 28 ; £vet. Bp. i, 15, etc. 

2 This, like the appeal to the existence of divinities, is a presumption of 
what is to be proved. Of course, the unbeliever believes neither in metem- 
psychosis nor in gods, but he is not allowed to say any more. In xii, 304, 47, 
the argument for the existence of the Source and the spirit is that both are 
inferable from effects (as seasons are from fruits, 306, 27). In the latter pas- 
sage, the spirit “inferred by signs,” liiigas, is called paneavirifatima (takara- 
lopa arsah !). 

3 Interesting, both as showing how the epic repeats itself and Buddhism, 
are xii, 175 and 277 (where several padas are identical with those in the 
Phammapada), and xiii, 113. The ahinsa doctrine is carried on here in xiii, 
114, 6, which repeats xii, 246, 18, with a varied reading that shows the futility 



ment against the existence of soul : “ Some say the cause, 
karana, of successive rebirth is ignorance, avidya, desire, con- 
fusion of mind, and the practice of faulty acts; ignorance 
being the field watered by thirst, and acts being the seed 
planted in it, all of winch cause rebirth. They say that 
(ignorance) is concealed (in the body) and is burned away, 
and that, when the mortal part is destroyed, another body is 
born from it and they call this the destruction of being. But 
(in answer to this), how can it be just the same man in this 
(new body), since he is different in form, in birth, in good, 
and in aims '? For (if there is no soul) all would be discon- 
nected. (Further) if this is so, what pleasure would there be 
in gifts, wisdom, or the power gained by religious practices ? 
For another entity would get the fruit of what this man prac- 
tises, since one man by means of another’s nature, prakrtaih, 
would be made wretched or blessed here on earth. (In this 
matter) the decision in regard to what is invisible (must rest 
on) what is visible. If you kill a body with a cudgel would 
another arise from it? Even so the separate consciousness 
Avould be a different consciousness, not the original one. 
This destruction of being (spoken of above, satvasariiksaya) 
would be repeated like seasons and years ; [there would 
indeed be no end to it, for if it is argued that destruction 
of consciousness ever results in a new consciousness, then 
destruction of being would result, not, as the Buddhists teach, 
in annihilation, but in new being; so there would be no 
escape from rebirth. If one says, however, that there is a 
conditioned soul, it can be only a physical bond of unity] like 
a house, growing gradually weaker through repeated aging 
and dying (consisting, as such a ‘soul’ must) of (mortal) 
senses, thoughts, breath, blood, flesh, bone, all of which perish 
and revert in due order to their original bases. And, further, 
(such a theory) would refute the practice of the world in 

of relying on the commentator, who thinks that the elephant in the following 
stanza of £anti is Yoga! Yatha nagapade 'nyani padani padagaminam, sar- 
vany eva ’pidhlyante padajatani kaunjare, evaiii sarvam ahinsayam dharmar- 
tham apidhlyate (in xiii, evam lokesv ahinsa tu nirdista). 



respect of obtaining advantage from gifts and other religious 
acts, since both the words of the Veda and the practice of the 
world (show that acts are performed) for this purpose (of 
gain). There are many proofs to be found in the mind, 
but what with the iteration of this and that cause uo clear 
light is obtained, but men doubt and turn to some one expla- 
nation, till their intellect becomes fixed on one point and rots 
there like a tree. So all creatures, made wretched through 
(desiring) useless objects, are led away by received teaching, 
agamaih, like elephants led by their keepers. Thus, desiring 
objects that bring endless pleasure, the dried-up many get 
instead a greater sorrow on being forced to abandon the bait 
and enter the power of death. - ’ 

The argument is the familiar one that a man gets sorrow 
through desiring heaven, for after Iris Ivanna is exhausted he 
sinks down again to a lower level. So heaven is a bait which 
attracts men; but as it is only a temporary pleasure followed 
by pain, one suffers from it all the more (nessun maggiore 
dolore die rieordarsi). All this implies unconscious existence 
as the best goal. 

To this it is said, 219, 2, in the woids of the great Upani- 
shacl: “If there is no consciousness after death, 1 what differ- 
ence does it make whether one lias wisdom or not, or is careful 
or not ? ” Then Paneagiklia replies with a long exposition of 
his system, 219, 6 ff., of which I give the chief points : 

It is not a system of annihilation, ucchedanistha, nor one 
of the soul’s separate existence, bhavanistha. The (visible) 
man consists of body, senses, and perception, c-etas. The 
foundations are the five elements, which are independent and 
make the body. The body is not of one element, but of five. 
The aggregate causing activity is knowledge, heat, and wind. 2 
From knowledge come the senses and their objects, separate 
existence, svabhava, perception, cetana, and mind : from wind 
come the two vital breaths; from heat come gall and other 

1 yadi na pretva samjna bhavati ; compare tany (bhutani) era ’nuvinav 
yati, na pretya samjna 'sti ’ti, BAU. ii, 4, 12. 

2 219, 9 ; compare below. 



bases, dhatus. The five senses, indriyas, hearing, touch, taste, 
sight, smell, derive from the mind, eitta, and have its charac- 
teristics. Eternal cetanit is threefold when united with dis- 
cernment, vijiiana. This they call sukhadnhkha and the 
opposite. Sound, touch, color, taste, smell, the forms (mur- 
tavah, containing these as objects), make a group of six 
constant constituents, gunas, to make knowledge perfect. 
Dependent on these are acts and visarga (?), and judgment in 
regard to the meaning of all topics. This they call the highest 
seed, gukra ; it is intellect, the great undeteriorating (sub- 
stance). This collection of attributes is not soul but is 
not-soul, anatman. The true teaching is contained in Renun- 
ciation-(?astras, which enjoin renunciation of all. Having ex- 
plained the six jnanendriyas, organs of knowledge, Paiicacikha 
explains the “organs of action, which are five, with bala, 
power, as the sixth,” 5 I. 20. There are twelve organs, five 
organs of knowledge with mind as sixth, and five of action 
with power as sixth. The eleven organs (with mind) one 
should renounce by means of the intellect. Ear, sound, 
and mind (eitta, in 23 and 34 ; manas in 22) are necessary in 
hearing . 1 Thus for all the senses there are fifteen gunas 
(3 x 5). There are also the three gunas called sattva, rajas, 
tamas. Ear and sound are forms of air (space) ; so with the 
five others. In the ten senses there arises a creation (entity) 
simultaneous with their activity; this is (the eleventh), mind, 
eitta. The intellect is the twelfth. In deep sleep, tamase, 
there is no annihilation (of personality), although there is 
concerned no such creation simidtaneous with the senses (the 
co-operation being a popular fallacy). (In deep sleep) in 
consequence of one’s former waking experience, and because 
one is conditioned by the three gunas, one imagines that one 
has material senses, although one can perceive only subtile 
senses. But though one imagines this, one does not really 

1 Compare Gita, 18, 18 (threefold urgers to action), knowledge, object, 
knower, jiianam jneyam parijnata trividha karmacodana ; threefold action, 
organ, act, agent, karanam karma karte ’ti trividhah karmasamgrahali; in 
14, the five karanani or karmanah hetavah are object, adhisthana, agent, 
organ, action, and the daiva (said to be Sarhkhya, but interpreted as Vedanta). 



co-operate (with the senses. Hence it may be inferred that a 
soul exists independent of mental processes). But the deep- 
sleep consciousness is a finite and darkened pleasure. Even the 
result one derives from traditional teaching, agama, though 
not sorrowful, is also merely darkness, revealed lies, as it were. 1 
Spirit, ksetrajna, is the being, bhava, standing in mind; it 
is immortal, flowing as a stream to the ocean. For the de- 
struction of existence, satvasariiksaya (the expression used 
above) is (in Upanishad language) as when rivers run into 
other rivers and to the ocean, losing their individuality, 
vyakti (equivalent to form) and name. Consequently, when 
the individual spirit, jlva, is united (with the ocean of being) 
and embraced on all sides, how could there be consciousness 
after death? (219, 43). As the creature that spins out of 
itself, wrapping itself in its web-house, stays there over- 
powered, so is the soul ; but when freed, it abandons its misery, 
and then its woe is destroyed, like a clod falling on a rock. 
As the deer leaves its old horn, and the snake its skin, with- 
out looking behind, and a bird leaves the falling tree and flies 
away unattached, so the freed soul abandons its woe, and 
leaving pleasure and pain, without even a subtile body, goes 
the perfect way (47-49 repeats 45). 2 

For a Sarhkliya philosopher Panea§ikha teaches very extra- 
ordinary things, the most advanced Brahmaism, which fails 
only of being Vedanta in its lack of Maya. Three sets of 
philosophers are here refuted, — the materialist, the Buddhist, 

1 The commentator reads atha tatra ’py upadatte tamo 'vyaktam iva 
’nrtam, gl. 38, which is perhaps better “hidden falsehood.” The meaning is, 
as explained above, that the joy given by Vedic teaching is a perishable 
heaven resulting in sorrow (darkness) and the teaching is not the highest 
truth. Compare, on the other side, the same reproach, Mait. Up. vii, 10, 
satyam iva ’nrtam pafyanti. 

2 Compare Praf. Up. v, 5; Mund. Up. 1, 7 and iii, 1. The first image is 
clearly not that of a spider (which is not destroyed by its web), but of a 
silkworm, though the commentator (and PW.) take urnanabhi as a spider, 
which comparison is common. Compare xii, 286, 40, urnanabhir yatha sutrahi 
vijnevas tantuvad gunah (as in BAU. ii, 1, 20). But the silkworm is also 
common. Compare xii, 304, 4, koyakaro yathatmanam kitah samavarundhati 
sutratantugunair nityarn tatha 'yam aguno gunaih dvandvatu eti ca nir- 
dvandvah, etc. 



and the orthodox Vedist. The terms used are those of the 
Silmkbya, jlva and ksetrajila rather than atrnan (sthito manasi 
yo bhavah sa vai ksetrajna ueyate, §1. 40), but this spirit is 
only part of Brahman. 1 

Another point to be noticed is the absence of tanmatras. 
Before passing to the numerical analysis of the Paiicaratra 
scheme into thirty elements, I would point out also that as in 
Gita, 7, 4, so ib. 13, 5-6, there are gross elements, egoism, 
intellect, and mind (= 8), but also ten organs and five objects 
of sense plus avyakta (= 24 topics), to which are here added, 
Gita, 13, 5-6, desire, aversion, pleasure, pain, and also body, 
perception, courage (sariighata, eetana, dhrti) or thirty-one 
elements of “modified Prakrti.” 

The Thirty-one Elements (Pancacikha). 

Here there is a formal group of particles called kalas, not 
sixteen but thirty, but one (God) super-added makes thirty- 
one topics, the same number ascribed by tradition to the 
Pagupatas. A most minute description is given in xii, 321, 
96-112. This scheme is as follows : 2 

In order to act, the organs “ await the outer constituents,” 
gums. In perception, color, eye, and light are the three 
causes, and so in all cases where are found knowledge and 
the object of perception, (similar) causes of knowiedge exist ; 
between knowledge and the object intervenes the guna, con- 
stituent, mind, wherewith one judges. [The organs and mind 
make eleven.] 3 The twelfth is intellect , another constituent, 
wherewith one decides in the case of doubtful things to be 

1 The attribute of Jagatprakrti applied to Narayana in the Paiicaratra 
hymn, xii, 339, 89, “the god who is the Source of the world,” gives the 
vital difference between this teaching and that which inculcates a Prakrti 
distinct from pure soul. 

2 I italicize below without extended comment the points of contact with 
the scheme just given. 

3 This must be supplied from the context. In the scheme at xiv, 42, 16, 
“mind must be recognized as belonging to both, and intellect is the twelfth,” 
only ten organs are recognized, as here, and bala as a separate organ is 



known. The thirteenth constituent is sattva. (It is real) 
for one is argued to be an individual having much or little 
sattva (hence it is a real constituent, a guna). The four- 
teenth constituent is egoism (when one says ‘ I am an agent ’), 
with which one gets the notion of mine and not-mine. Then 
there is a fifteenth constituent, which is different from the 
others and is called the totality of the mass of separate factors, 
prthakkaliisamuliasya samagryam (i. e., the general disposi- 
tion). The sixteenth, a different constituent, is a sort of 
complex, saihghata iva (because it consists, says the commen- 
tator, in the union of the three factors of ignorance ; the six- 
teenth is therefore avidyii, or ignorance itself), wherein are 
combined the Source and the individual manifestation, vyakti, 
which are respectively the seventeenth and eighteenth con- 
stituents, gunau. The nineteenth is the unification of doub- 
lets (opposites), such as pleasant and disagreeable, age and 
death, etc. The twentieth constituent is Time, the origin and 
destruction of all things. This complex, saihghata, of twenty, 
and in addition the seven constituents consisting of the five 
gross elements added to [the origin and relation of j being and 
not-being, (making twenty-seven, is to be added again to) 
three more constituents, vidhi, fu/cra , baht (cause, seed, power ). 1 
That is called the body in which these twenty and ten are all 
together. The Source (fore-cause) of these kalas, factors, one 
philosopher recognizes to be the Unmanifest; another, dull of 
insight, recognizes (as such) the Manifest. Metaphysicians 
recognize a Source of all beings, whether it is the Unmani- 
fest or the Manifest or a double or quadruple source. This 
unmanifest Source becomes manifest by means of the kalas 
(the factors just enumerated). The individual is the Source 
so made manifest. From conception to old age there is an 
uninterrupted momentary splitting up of the factors (par- 
ticles) of the body, although too minute to be observed (in 
detail). But this passing away and coming into existence of 

1 According to the commentator, these are right and wrong as originating 
false ideas, vasana; that which incites to wrong ideas; and the effort leading 
to the attainment of wrong ideas. But see the scheme above. 



the separate particles goes on from stage to stage just like the 
course of a lamp’s light. There is, therefore, no connection 
between the individual existent creature and his members. 
All creatures are born by the union of particles, kalas, as it 
were, 1 just as fire is produced by the union of sunlight and 
fire-stone , mani, or by sticks (rubbed together). 

This exposition is given for a practical purpose, as is seen 
in the last paragraph. One should recognize no own , as all 
creatures are one, distinct from the physical parts. The 
“ body of particles,” as it is called in xii, 322, 25, reverts to 
the unmanifest Source, but the self or soul is but part of the 
same soul in any other body of particles. The doctrine is 
none the less that of Panca§ikha because it is taught by 
Sulabha to Janaka, though it is the latter who professes him- 
self the disciple of Paiicagikha, “the venerable beggar who 
belonged to the family of Paragara,” xii, 321, 24. For Janaka 
does not really understand, and so Sulabha is enlightening 
him. Paiicagikha is here said to be a Samkhya leader. There 
is an imitation and would-be improvement in this late dis- 
course (the metre shows the lateness) of Gita, 3, 3, loke 
'smin dvividha nistlia. Here §1. 38, the “ point of view,” is 
made treble, trividha nistha drsta ; not that emancipation is 
got by knowledge or action, as in the Gita passage, but by 
the third (and best view), that of Paiicagikha, who “rejected 
both these two,” 321, 40. The doctrine is that the vaigesikam 
jiianam or most excellent way, gl. 23, leads one to live a life 
of renunciation. All depends, says the king, on whether one 
is bond or free ; the pure and good devotee may still be active ; 
asceticism is not requisite; a king is as good as a beggar. 
“ The bond of royalty (says the king in conclusion), the bond 
of affection, I have cut with the sword of renunciation, which 
has been sharpened on the anvil of emancipation,” ib. 52. 
But his antagonist intimates that he has not learned the true 
religion, which is renunciation in deed as well as in thought. 
As a system, the doctrine of Paiicagikha is said to be sopayah 

1 The commentator says that “this expression, (kalanam)iva,has no mean- 
ing, and is merely used to fill up the verse/’ 321, 124. 



sopanisadah sopasangah 1 sanigeayah, gl. 163, a detailed philo- 
sophical exposition. 

In xii, 276, 4 ff., there is a third exposition, oddly combined 
with the Samkhya schedule, while at the end it shows resem- 
blance to that just given. It is referred to Asita Devala, who 
in xiii, 18, 18, is said to have received glory from Oiva (Civa 
is Sariikhyaprasadah, xiii, 17, 63), who “gives the goal of 
Saiiikhyayoga,” xiii, 14, 198. In this scheme Time creates 
the five gross elements. Impelled by Being and Soul, Time 
creates beings out of these elements, which with Time make a 
group, ragi, of six. To these are added bhava and abhava, 
making the “ eight beings, bhutani, of beings.” When de- 
stroyed, a creature becomes fivefold (elements) because of 
these. The body is made of earth, bhumimayo dehah ; the ear 
comes from air (space) ; the eye from the sun ; the breath 
from the wind; the blood from water. The five senses are 
the “ knowledges ” (organs of knowledge, jiianani). Sight, 
hearing, smelling, touch, taste, are five, distributed fivefold 
over five. Their constituents, tadgunah, are color, smell, 
taste, touch, and sound, apprehended in five ways by the 
five senses. These, their gunas, the senses do not know, 
but the spirit knows them (this is a correction of the state- 
ment that objects of sense are apprehended by the senses). 
Higher than the group of senses is eitta, perception ; higher 
than citta is mind ; higher than mind is intellect ; higher than 
intellect is spirit. A creature first perceives, eetayati, differ- 
ent objects of sense. Then pondering, vicarya, with the mind, 
he next determines, vyavasyati, with the intellect. One that 
has intellect determines objects of sense apprehended by the 
senses. Perception, the (five) senses as a group, mind, and 
intellect are, according to metaphysicians, the eight jhane- 
ndriyas, organs of knowledge. There are five organs of action 
and bala is the sixth organ of action , gl. 22. Sleep-sight is the 
activity of the mind viien the activity of the senses is sus- 
pended. The states, bhavas, 2 of sattva, tamas, and rajas 

1 upasaiiga for upasangah 1 N. defines as dhvanangani yamafiini. 

2 This word means being as entity (and so is equivalent to guna, constitu- 



(joy, success, insight, virtue, being the causes of one being 
endowed with sattva), which are associated with activity, 
whatever their cause of activity, vidlii, are retained (in sleep) 
by memory. There is an agreeable and constant immediate 
passage between the two states, bhavayoh (that is the passage 
is immediately perceptible between waking and sleeping). 
The organs and the states are called the seventeen constitu- 
ents, gunas. The eighteenth is the eternal incorporate one 
in the body, dehl garire (spirit). 

Here fourteen organs are added to the three gunas, sattva, 
etc., for there are “ eight organs of knowledge ” and six of 
action (elsewhere there are only five organs of knowledge). 
Of the group of seventeen I have already spoken, and note 
here only the intrusion of citta between senses and mind. 
The account proceeds not very lucidly : There concorporate 
constituents bound up in body in the case of all incorporate 
creatures cease to be concorporate on the separation of the 
body ; or the body made of five elements, paficabhautika, is a 
mere (temporary) union, samnipata. The one and the eigh- 
teen gunas with the incorporate one and with heat, usman 
(the internal heat of the stomach, says the commentator), 
make the complex, samghata, of twenty composed of five ele- 
ments, which (twenty) the Great One, malian, with wind sup- 
ports. The death of each creature is caused by this (wind). 
On destruction, the creature enters the five elements, and 
urged by its good and evil, assumes a body again ; and so on 
from body to body, urged by Time the ksetrin (spirit) goes, as 
if from one ruined house to another. 1 

The vincjo samghatah paneabhautikah or complex of twenty 
composed of five elements in this passage is the same with the 
vingakah samghatah of the preceding, 321, 109. But there 

ent) or existence and so state of being. It often adds nothing to the meaning. 
For example in xiii, 141, 85, “ bhava of self ” is the same with self : atmany 
eva ’tmano bhavam samasajjeta vai dvijah, “ put self in self.” 

1 vifirnad va (= iva) grhad grham. The analysis above, 276 (5), 30 : eka<? 
ca da£a ca 'stau ca (= 19) gunah, saha ^arlrina (dehin in §1. 28) usmana saha 
(besides heat)'vin§o va samghatah paneabhautikah, mahan saiiidharayaty etac 
chariram vayuna saha. Compare the first scheme above. 



Time is the twentieth, and the twenty are the bodily gunas. 
Nevertheless, the employment in each, not only of the group 
of twenty but also of bala and vidhi, as found above, points to 
a common basis. 1 In none is there a trace of Vishnuism. 

The Secret of the Vedanta. 

The united systems of philosophy called “ Secret of the 
Vedanta” and exploited in xii, 194, 248 ff., and 286, which 
in the following pages I shall designate as A, B, C, present a 
curious mixture, which on careful analysis show clearly that 
they are three different versions of an older Saihkhya tract, 
which is worked over into Bralimaism. There is no clear 
recognition of egoism, though the commentator so interprets 
the “ maker of bliutas ” in C 9, and, as I have said above, I 
think it doubtful, both from these and other passages, whether 
the earlier Saihkhya recognized Intellect as other than self- 
conscious. One of the present three schemes introduces the 
Bhutatman as deus ex maehina. They all differ slightly and 
have the Pancagikha terminology to a certain extent. In their 
threefold form they offer an instructive example of liow the 
epic copies itself. They all begin with the same request to 
the instructor to give a metaphysical, adhyatma, lecture. The 
first and last versions represent Bhlsma as teacher and Yudlii- 
sthira as pupil ; the other, Vyasa as teacher and (”uka as 
pupil of the same lecture. The two Bhlsma lectures do not 
agree so closely with each other throughout (though more 
alike at first) 2 as do the Vyasa and second Bhlsma version, 

1 Compare with this samghata or vital complex the jlvaghana, Prajn. v. 5. 

2 The closer agreement begins with A 9 as compared with 13 9 and C 10 ; 

“ sound, ear, and holes, this triad is born of air ; touch, action, skin, are born 
of wind ; color, eye, digestion, are called the threefold light, tejas.” Here B 
and C have “vital airs” for skin, and jyotis for tejas. In the next group, 
where A has taste, kleda, tongue, B and C both have sneha. Again “ mind as 
the sixth ” organ appears in A 11 but is omitted in B 11 and C 12, to reappear 
in B 17, C 15. In all these versions, body, with smell and object, is of earth 
alone, bhumigunah, loc. cit. Besides these triads, B and C give sound, gliosa, 
($abda) from air, smell alone as bhuuiiguna in B, all composite matter, sam- 
ghata, as eartli-guna in C ; breath (C) or touch (B) from wind, etc. 



which lie nearer together in place. It will be necessary to 
treat these chapters rather fully if we wish to get a clear idea 
of the manufacture of epic philosophy. 

Coming, then, to details, the glokas are intermingled in such 
a way that part of one cloka in one discourse is part of another 
in another version. Thus, after the introductory stanza, which 
nam es the five elements with but trifling variations, A has : 
“ Whence they are created thither they go, again and again, the 
great bhutas, from other bhutas, like waves of ocean ; and as 
a tortoise, stretching forth limbs, retracts them again, so the 
Bhutatman again withdraws the bhutas he has created.” In 
B, the expression “ like waves of ocean ” comes in the first 
stanza, replacing the expression “ origin and destruction ” in 
A. In C, as regards this expression, the reading is as in A, 
but the important lines of the tortoise and Bhutatman appear 
here thus : “ As a tortoise here, causing liis limbs to stretch 
forth, retracts them, so the smaller bhutas in respect of greater 
bhutas ; ” while B has : “ As a tortoise here, stretching forth 
limbs, retracts them again, so the great bhutas, mahanti bhu- 
tani, modify themselves in the smaller ” (younger) ; and this 
is repeated, ib. 14, in a stanza omitted in the other versions 
with the momentous alteration : “As a tortoise here, his limbs 
outstretcliing, withdraws them, even so the Intellect , having 
created the group of senses, withdraws them.” 

The next change is in A 8, where, after stating that the 
“ maker of bhutas ” put the gross elements differently in all 
beings, the teacher here adds “but the jlva spirit does not 
see that difference,” which in the other versions appears with- 
out mention of jlva, with visayan in C for vaisamyam. Of 
the new group of eight sources found here, I have spoken 
elsewhere. All the versions have the following stanza A 17, 
B 16, C 18: 

gunan (A, C, gunair) nenlyate buddhir, buddhir eve- 
’ndriyany api (C, ca) 

manahsastani sarvani (A, bhutani), buddhy (A, tad) 
abhave kuto gunah, 

that is, Intellect directs the gunas ; the senses are intellect 



and their constituents could not exist without it. A and C 
make the intellect subservient to the gunas ! C, as if to ex- 
plain the gunas, inserts “ tamas, sattva, rajas, time, and act,” 
while in 13 it has a verse (mingling cases), “ sattva, rajas, 
tamas, kala (nom.), and karmabuddhi (nom.), and mind, the 
sixth, in these (bases) the Lord created.” B, too, has an 
addition : “ Mind, intellect, and nature, svabhava, these three 
are born of their own sources ; they do not overpass the gunas 
on arriving at that which is higher than the gunas ” (13, na 
gunan ativartante). So in 316, 2, gunasvabhavas tv avyakto 
gunan nai ’va ’tivartate. But in 249, 8 if., the continuation 
of B, the intellect, identified with the bhavas (states produced 
by gunas) does overpass them, “ as the sea does the shore.” 
The image here is so conventional, saritam sagaro bharta 
mahavelam ivo ’rmiman (compare A, 23 if. ; C, 23 if.) that 
there is no doubt what has happened. The constant unchang- 
ing epic simile is that one remains, not over-stepping, “as 
the sea does not overpass its shore.” In other words, there 
is in this passage an intrusion of the Yoga idea 1 that the soul 
can overpass the gunas (compare Gita, 14, 21, and xii, 252, 
22), and so the ancient simile is introduced without its nega- 
tive, making the absurdity shown above. 2 

B alone adds, in 249, 3, “the intellect is soul,” atman, 

1 Compare xii, 205, 17 : “ Mind abandoning gunas attains freedom from 
gunas” (above). Gunas and bhavas are here the same thing, for the latter 
are the result of the presence of the former. They (or the eight sources) 
“ carry the universe but rest on God,” 210, 28, 36. This is a Lord-system, 

though “ Lord ” is a form of ignorance : “ elements, senses, gunas, three 
worlds, the Lord himself, are all based on egoism,” 212, 18-19. 

2 svabhava, nature, is distinct from sadbhava. One is temporary, the 
other is eternal, xiv, 28, 22; Gita, 8, 3. The three texts in describing the 
modification of intellect “called mind when it desires,” A 20; B (219), 2; 
C 20, have slight variants ; “ that with which it sees is eye, hearing it is 
called ear,” A 19 ; B 4 ; C 19, where B and C have frnvati, etc., but A 
the verb throughout. In A 13 (and the corresponding verses B 18, C 19) 
“ the mind doubts,” sampayam kurute, “ the intellect decides,” adhyavasa- 
naya. Compare 249, 1, raano visrjate bhavam buddhir adhyavasayini, hrda- 
yam priyapriye veda, trividlia karmacodana. “ The intellect is the chief 
tiling in that which is to be made” (B 15), suggesting egoism, but C 14 has 
krtsne and A has no subject at all. 



which is in line with the tendencies at work here. So in 249, 
20, there is a stanza which must be compared step for step 
with the parallel passages : “ Soul, atman, puts forth intellect, 
but never (read na ’pi) gunas ; the gunas do not know soul, 
but soul, sa, knows gunas always, and it is the observer and in 
proper order occupies itself with them. Know that this is the 
difference between intellect and spirit (ksetrajna for the pre- 
ceding atman), one creates gunas, one does not create gunas ; 
both being different but joined by the Source, united as a fish 
to water, or fly to udumbara, or as sheath to grass-blade. 
Intellect truly creates gunas, but the spirit, the Lord, superin- 
tends, as the gunas modify themselves ; all that is part of its 
own nature, that intellect creates gunas ; as a spider does his 
thread, so that creates gunas.” 

In A, 88 ff. : “See the difference between intellect and 
spirit, ksetrajna ; one creates gunas, one does not create gunas ; 
as the fly and udumbara so are they joined ; both being differ- 
ent, but joined by the Source ; as a fish and water are joined 
so are they; the gunas know not the soul, atman, but the 
soul, sa, knows the gunas always. But being an observer of 
the gunas (the spirit) imagines them created (by himself). 
The soul, atman, with the senses and intellect as the seventh, 
which are moveless and ignorant, illuminates the object, pada, 
like a lamp. Intellect truly creates the gunas, the spirit, 
ksetrajna, looks on; this is their connection. There is no 
support for the intellect and spirit. Mind creates intellect but 
never creates the gunas ... A Yogin in his proper nature 
creates (srjate) gunas, as a spider his web.” 1 

C 33 begins as in B, “ Know that this is the difference,” 
down to the image of the fish; then, omitting the fly, etc., 
goes on as in A : “ The gunas know not the soul, atman, but 
the soul knows gunas always, but, being an observer of the 
gunas, it imagines itself the creator. There is no support 
for the intellect . . . 2 the intellect, buddliir antara, with the 

1 Unique. Mind here is for atman in B. 

2 A senseless addition is found here, followed by srjate hi gunan sattram 
kjetrajnah paripafyati (as in A). Sattva, itself a guna, rests^ on rajas, xii, 



senses, which have no eyes and are ignorant, makes the senses 
luminous like a lamp (the intellect alone sees, the senses are 
like lamps) . . . this is even the fulfilment of its nature that 
(intellect creates) gunas as a spider his thread; the gunas 
should be recognized as a web.” 1 

A Stimkhya text is here changed into a later philosophy, 
with soul substituted for spirit, and the Yogin making gunas. 
Hence also the intellect is grouped with senses as ignorant in- 
struments of the soul, while Mind is creative soul. Even apart 
from the philosophical modifications here 'visible, it is difficult 
to see how the synthetic method can account for these three 

213, 12, sattvam ca rajasi stliitam, jnanadhisthanam avyaktam buddhy- 
ahamkaralaksanaih tad bijarii dehinara ahull. Compare 215, 25, jnanadhi- 
sthanam ajnanam vijnananugatarii jfianam ajniineua ’pakrsyate. But we 
have in ayrayo na ’sti sattvasya a phrase in which sattva is equivalent to 
conscious buddhi. The varied readings show clearly that the text has been 
tampered with. In ai;rayo na ’sti sattvasya gunah fabdo na cetana in 240, 
14, followed by sattvarii hi tejah srjati na gunan viii kathamcana there is 
still another parallel to our text. So in 241, 3 £f., sattva is buddhi, higher 
than citta, as it is said “merge citta in sattva” (247, 5 and 9, the Yogin’s 
suksma buddhih). Elsewhere citta, by the way, is an organ "lower than 
mind,” 276, 16. The version in 194, 44, is aijrayo na 'sti sattvasya kse- 
trajnasya ca kajeana, sattvam manah samsrjate na gunan vai kadacana 
(after the words srjate hi gunan sattvam), where manas must represent atman 
in the version above. The form gunah fabdo na cetana appears, a scribe’s 
error apparently, in 286, 36, as gunasargena cetana, before the meaningless 
words : sattvam asya srjanty anve gunan veda kadacana. The epic sattva 
is well known : “ One is fitted for Braiiman existence as sattva gradually 
departs,” i. e., as circumscribed jiva becomes pure. Compare also 217, 21-25 
(210-217 are a professed adhyatma of NarSyana), where it is said that jiva 
quits rajas and goes about like sound but in a body, and then gets established 
in Source, and finally leaves even that body and enters “end of body which 
rests on nothing,” niragraya. 

1 Other common metaphors and similes are that of the cocoon (pp. 36, 
151), the “bonds of hope,” acapacn, Gita, 10, 12; the net, xii, 242, 7 ff. ; but 
unique is the weaver of xii, 217, 30 : “ As a weaver passes the thread through 
cloth with a needle, so the thread of transmigration is fastened with the 
needle of desire, samsarayati (sariisarasutra) trsnasucya. Compare foam-like 
body and bird-like soul, xii, 322, 7 ; as well as the elaborate river-metaphors 
(taken from the battle-epic), where the bank is truth, waves are untruth, 
desire is a crocodile, and the river of the unmanifest goes into the sea of 
transmigration, iii, 207, 72; xii, 251, 12 ff. (Dh. Pad., 251, n’ atthi moliasamam 
jalam n’ atthi tanhasama nadi). 




conversations. From an historical point of view the problem 
is of course simple. 

The question asked above, “What would become of the 
gunas in the absence of intellect?” is taken up and continued 
at the end of the discussion : “ When the gunas, the strands 
spun by intellect, are dispersed, pradhvastah, they do not cease 
to be, na nivartante; a cessation, nivrtti, is not perceived. 
This is beyond the sphere of what is immediately perceptible 
(but) it is ascertainable through reasoning, anumana. So some 
decide, while others say they cease to be, nivrtti. Let one 
consider both views and decide as one thinks best, loosening 
the firm knot of the heart (an Upanishad phrase) caused by a 
difference of judgment,” 194, 50-52. If and C have “their 
activity, pravrtti, is not perceived,” for “a cessation is not 

The Yogin, who according to the teaching of this lecture 
can overpass the gunas, is said in the last section, in a supple- 
ment, xii, 252, ff., to surpass even the destruction of gunas, 
atikrantagunaksaya, and reach the highest goal. 

Details of Philosophical Speculation. 

It has been shown thus far that there are not only three 
religious philosophies in the epic, but also three formal sys- 
tems, one inculcating the twenty-five, one the twenty-six, and 
one the thirty-two categories. 

These broad differences are sufficient to show how entirely 
lacking in any uniform plan or scope is epic philosophy as 
a wdrole, and also to prove that the epic does not represent a 
preliminary chaos of opinions, but reflects at last three per- 
fected and systematized schemes of philosophy. I turn now 
to some details of speculation, incongruous for the most part, 
reflecting different interpretations and different views ; but in 
some cases noteworthy not so much for their lack of harmony 
with other epic schemes as for the irniqueness of views found 
only in one or two passages of the pseudo-epic, amid a mass 
of theories covering the same general subject. 



The Sixty Constituents of Intellect. 

This group, one of the most elaborate in the epic, is obtained 
by an “ enumeration,” parisariikhyana, which analyzes the ele- 
ments, xii, 256, 1 if. They are thus distributed : “ Earth 
has ten, firmness, weight (gurutva), hardness (kathinya), the 
function of productivity, scent, density (also gurutva, but 
explained as prathamanata, pindapustih), ability (to hold 
scents), compactness, support, endurance. Water has ten, cool- 
ness, taste, moistness, fluidity (dravatva), adhesiveness and 
softness ( ? snehasaumyatii), tongue, dispersion, also, and 
softening (grapana) of earthy things (these make nine, but 
the commentator supplies ‘ freezing ’ from ca, ‘ and,’ which I 
render ‘also’! Probably bhaumanam contains an old error). 
Fire, ten, dangerousness, light, heat, cooking, brightness, pain, 
passion (and is) swift; (it has) sharpness and ever upward 
flaring. Wind (air), ten, tempered touch, (it is) the organ 
of speech, vadasthana ; (it has) independence, power, speed, 
emission (of secretions), activity, movement (of breath), life 
(atmata, of the vital airs), and birth. The characteristic con- 
stituent of air (space) is sound ; (it has also) comprehensive- 
ness, openness, non-support, non-suspension, unmanifestness, 
steadfastness (avikarita), non-resistance (apratlghatita), ele- 
mentality, and changes (bhutatvaiii vikrtani ca, ‘that is, it 
causes hearing and apertures in the body,' X.). Thus related 
are the fifty constituents (gunah paiicagatam), which are the 
essentials of the five elements." To these are added nine 
constituents of mind and five of intellect, as follows : “ Cour- 
age, reasoning, memory (so the commentator renders upapatti 
and vyakti, perhaps individuality), creation (visarga, rendered 
‘ loss of memory ’ by the commentator), imagination, patience, 
good, evil, and swiftness, are the nine characteristics of mind. 
The destruction of the pleasant and the unpleasant (in deep 
sleep), judgment (vyavasaya), concentration, doubt, and insight 
are recognized as the five characteristics of intellect.” The 
two last, samgaya and pratipatti, are rendered by the commen- 
tator in just the opposite meanings, namely knowledge in 



doubtful matters and the application of other proofs as well 
as direct perception. In the light of explanations current 
elsewhere in the epic, where “ doubt-making ” is an attribute 
of mind, and judgment that of intellect, “doubt,” which is 
here clearly attributed to intellect, must indeed, from a syn- 
thetical point of view, Ik; interpreted by its opposite, or one 
may fall back on the remark cited below, that this is all 
nonsense. From an historical point of view, however, the 
statement may stand beside the many other inconsistencies of 
the epic. 

The section closes with a query on the part of the listener as 
to how intellect has five constituents and how the five senses 
are reckoned as attributes, katham pancendriya gunah ; to which 
the answer is the stanza : ahuh sastim buddhigunan vai bliuta- 
vigista nityavisaktah, bhutavibhutig c-a ‘ksarasrstah putra na 
nityam tad iha vadanti, “ They say that the constituents of 
intellect are sixty. These are distinguished by the elements ; 1 
(but) are always attached (to the intellect). The manifesta- 
tions of the elements are created by that which is indestruc- 
tible. They say that that is non-eternal.” “ That,” it is 
added, “ which has been declared to you here is foolishness, 
cintakalilam, and unorthodox, anagatam. Learning the whole 
truth in regard to the meaning of elements, gain peace of intel- 
lect by acquiring power over the elements ” (bhutaprabhavlit, 

The sixty may be got by adding the five gunas of intellect 
to the five elements plus their fifty characteristic constituents ; 
but the commentator says the true count is seventy-one, five 
elements with their fifty constituents added to mind and intel- 
lect with their nine and five constituents respectively. 

Two views are given. One is that there are fifty and nine 
and five constituents of five (elements), one (mind), and one 
(intellect) = 71. The other is that intellect has sixty con- 
stituents, five of its own, fifty of the elements (as parts of 
intellect), and the elements themselves (which are different 

1 The commentator paraphrases bhutaviyistah with panea bhutany api 
buddher eva gunah, “ the five elements are constituents of intellect.” 



from the constituents). The latter view is repudiated as 
imorthodox, and the final injunction is given to turn from 
this calculation to Yogi-discipline. 

This unorthodox enumeration is represented elsewhere by 
the title of Civa, who is called sastibhaga, xiii, 17, 72, and per- 
haps also by the mysterious manoviruddhani in the enumera- 
tion of the psychic colors explained below. Seven hundred 
vyuhas, or forms of activity, are traversed by the soul on its 
way through red and yellow, to white, when it courses above 
the eight worlds. Then follows, xii, 281, 46: 

astau ca sastim ca Qatani cai ’va 

manoviruddhani mahadyutmam 

“The eight (worlds) and the sixty and the hundreds (of 
vyuhas) are impediments to the mind of the illuminate.” 
The sixty are here explained as constituents of existence still 
adhering to the white soul. The commentator, however, gives 
an entirely different explanation from the one above, and 
though much the same in regaixl to the last two cases, his 
interpretation is not quite uniform. In the former case, the 
god enjoys tattvas or topics, experienced as stated at the 
beginning of the Mantlukya, in unconscious slumber, wake- 
fulness, and ordinary sleep, each of the latter being the real 
or illusionary fine and gross elements added to the nineteen 
“doors of enjoyment,” soul, five breaths, and the usual thir- 
teen (ten organs, mind, intellect, and egoism) ; while two of 
the sixty are attributed to dreamless slumber, cetas, soul, and 
subtilest capacity. In the latter case, the three states are sur- 
passed by a fourth state, to which the impeded white soul can- 
not attain. The impediments are much the same as those 
above, but include ignorance, desire and acts (the triad men- 
tioned above), and the states themselves. 

The Seventeen. 

In the exposition given in xii, 276, 6 ff., above, p. 156, there 
is a group of seventeen with an added spirit, making eighteen 
in all. Further there are “eight beings of beings," which re- 



mincl one of the “ eight sources,” but instead of the usual group 
we find here the gross elements, Time, being, and not-being 
(egoism is not a factor here at all). 1 

The group of seventeen plays an important part in epic 
categories, but it is clear from a comparison of the cases that 
there is no symmetry of system in the explanation. It is in 
short, as is the case in other instances, a Saihkhyan term used 
because it is an old term, but explained differently in different 
cases. One form we have just examined ; another I gave in 
the first chapter, above, p. 33, where was shown a late group 
of seventeen, containing most of the elements of the same 
group in the Vedantasara, five elements, mind, intellect, ego- 
ism, five organs of sense, spirit, atman, and the three gunas or 
constituents of all that is not pure spirit. 

On the other hand the Sariikhyan group, as in Aphorisms 
iii, 9, may be understood of the bodily constituents (ten organs, 
mind, intellect, and five elements) in a praise of ('iva who cre- 
ated the “seven guardians and ten others who guard this 
city,” 2 vii, 201, 76. The city here is the body, as in the Upa- 
nishads and Gita (bVet., 3, 18 ; G. 5, 13), elsewhere called 
“house,” as in v, 33, 100, “this house of nine doors, three 
pillars, five witnesses, under control of the spirit.” 3 

1 This exposition is called “ silly talk,” dustapralapah, xii, 280, 23, because 
it does not recognize that the course of transmigration may be brought to an 
end. For it is taught in the following chapter that not knowledge, penance, 
and sacrifice, but only self-restraint, can result in the attainment of Vishnu, 
the supreme God. For as a goldsmith purifies gold in fire so the soul is puri- 
fied by many rebirths or by one alone. Ilari creates, whose self consists of the 
eleven modifications, ekada 9 avikaratma, the sun is his eye, his mind is in 
the moon, his intellect is in knowledge, etc., and the gunas are essentially of 
God, 281, 9, 11-12, 19-21, 24. Here, as I have elsewhere pointed out, eleven 
modifications take the place of the regular sixteen, evidently the organs and 
mind without the elements. 

2 In conjunction with the two birds (spirits) and pippal trees (vikaras), 
manasau dvau suparnau vacayakhiih pippalah sapta gopah dafa ’py anye ye 
puram dharayanti. Compare for the birds and pippal tree JIund. Up. iii, 1 ; 
l, v et. iv, 6. 

3 The five senses, mind, intellect, egoism, and the gross body, make the 
ntne ; the pillars are restraints, ignorance, desire, action ; the house is the 
body ; the witnesses are the senses, says the commentator, who at Gita, 5, 13, 
gives a different explanation of the nine. The witness (as in popular style, i. 



Another passing allusion is found in sii, 280, 4, “ freed from 
the seventeen,” where (since the context excludes objects of 
sense, gunas, and the “eight”) the seventeen are explained 
by the commentator as five breaths, mind, intellect, and ten 
organs (the eight being objects of sense and gunas). Another 
passage alluding to the seventeen is taken in the same way : 
“ Who are free of the seventeen, the gunas, and acts, the fifteen 
kalas, particles, being abandoned, 1 they are released,” xii, 335, 
40. So again in xii, 352, 15-16 : “ The highest spirit is not 
affected by fruits, as the lotus leaf is not affected by water ; 
but the other, the active spirit, karmatman, is bound by the 
bonds of salvation 2 and it is bound also by the group of seven- 
teen,” where nu), group, is used as in the first example above, 
though the group is a different one. 

It follows that the epic is not consistent with itself but 
interprets the “ group of seventeen ” in different ways. 3 

74, 31, hrdi sthitah) is sometimes made sixfold, as the spirit and fixe senses, 
xiii, 7, 5. Various poetical modifications occur : “ A house, agarakam, of one 
pillar, nine doors,” xii, 174,59; a city, xii, 210, 37; nine doors again (still 
differently explained by the commentator) in xii, 240, 32, where the spirit is 
hansa (compare 240,29-31). A very elaborate working-up of the body-city, 
with senses as citizens, buddhi as Lord, etc., will be found in xii, 255, 9 ff. 
The hansa passage reflects the Hpanishads : 240, 29 = Qvet. iii, 16 ; 30 = v. 1. 
of tjvet. ib. 20; 31 has the unique dvaidhibhavn (atmanah) of Maitri, xii, 11 : 
32 = later form of Qvet. iii, 18. On p. 45, 1 gave kalah pacati in Strip, as acci- 
dental or universal. Not so here, however, where Maitri vi, 15, kalah pacati 
. . . yasmins tu pacyate kalo yas tam veda sa vedavit, appears complete (with 
the v. 1. tam vede 'ha na kagcana) in 240, 25. So too 5 I. 17 = Katha iii, 15 ; 
and 26 = <Jvet. iv, 19; while in 15, manisa manasa viprah pagyaty atmannm 
atmani (evarii saptadajam dehe vrtarii sodagabhir gunaih) there is a direct 
copy of the older form, £vet. ; v> 17 j e tc. £1. 19, 20, 21 copy the Gita. 

1 ye hlnali saptadagabhir gunaih karmabhir eva ca, kalah pancadaga 
tyaktas te mukta iti nigcayah. Here the commentator takes gunas as sattva, 
rajas, and tamas. On the fifteen kalas. see below. 

2 Moksabandhah, perhaps moha should be read, unless moksa implies 

3 There are of course other groups of seventeen. Thus in xii, 269, 25-26, 
Agni is seventeenth in the sacrificial group, plants, cattle, trees, withes, butter, 
milk, sour milk, ghee, land, points of compass, faith, time (are twelve), the 
three Vedas, the sacrificer (are sixteen), and seventeenth is Fire, the house- 



The Sixteen (A) Particles. 

What has happened in the mixture just described is 
obvious enough. The fifteen kalas, mentioned above as 
something to be abandoned, imply a sixteenth kala, the 
not-to-be-abandoned psychic entity itself. The impediments 
are called indifferently kalas and gunas, the former being the 
old designation, as in Mund. Up. iii, 2, 7, “ the fifteen kalas 
disappear.” Here as in Brh. Aran., i, 5, 15, the sixteenth is 
the soul ; but in Prag. Up. vi, 2-5, the soul is the source of 
the sixteen, sa puruso yasminn etah sodaga kalah prabhavanti, 
Purusa makes them, each from the preceding : “ breath, fai th, 
five elements, sense, mind, food, energy, austerity, hymns, 
sacrifice, the world, and the name (individuality),” and they 
all flow back into Purusa in reverse order. In xii, 47, 58 ff., 
(where the samkhyatman is yogatman, mayatman, vlgvatman, 
goptratman) God is “ the Samkhyas’ Seventeenth, having three- 
fold soul (tridhatman, awake, dreaming, in dreamless sleep), 
standing in soul, enveloped in the sixteen gunas.” The six- 
teen in xii, 210, 33 are the eleven organs and five objects of 
sense, which come from (1) the Unmanifest, producing (2) act- 
born intellect, which produces (3) egoism, whence come, one 
out of the other, (4) air, (5) wind, (6) light, (7) water, (8) 
earth, the eight fundamental sources on which the universe is 
established (vs. 29, the sixteen modifications, ten organs, five 
objects of sense, and mind). Compare also above the “ freed 
from six and sixteen.” So in xii, 242, 8 = xiv, 51, 31, where 
every creature has a body, murti, and “ consists of sixteen,” 
murtiman sodacatmakah. The Upanisliadic kalas and the 
Samkhya groups have united, and in turn are affected by 
other later groups. In xii, 240, 13, there is a group of sixteen 
“always in the bodies of incorporate creatures,” the five 
senses and the five objects of sense, the svabhava or individual 
nature, intellect, cetana, and mind added to two vital breaths 
and to spirit itself ; while in 302, 24, svabhava and cetana are 
apparently not included in the “ sixteen gunas ” which encom- 
pass the body; or, if the sixteen be interpreted as inclu ding 



them, then in both cases we have a group of sixteen quite 
distinct from that in the previous section, where organs and 
objects of sense make the number. Further, in the former of 
the two last sections, cetana is distinct from manas, with which 
it is elsewhere identified (see the section cited on p. 34 from 
the third book). Compare also the account of creation in 
xii, 233, 10 ff., already referred to, where the seven mahat- 
mans, intellect, mind, and the elements, unite to make body 
as a base for spirit, garlraiii grayanad bhavati, murtimat soda- 
gatmakam, 233, 12, into which enter mahanti bhutani. The 
elements are the gross, as they are described in gl. 8 (gunah 
sarvasya purvasya prapnuvanty uttarottaram), and there seems 
no reason for differentiating them from the Great Beings, 
though the commentator takes them as intellect and tanma- 
tras, and the sixteen as gross elements and eleven organs, 
explaining the whole process as the creation of the lihga in 
the sthula body. 

The group of sixteen plus a seventeenth, as given in the 
scheme above, is a combination of two schedules, one the 
regular seventeen of the Aphorisms, the other an earlier group 
of sixteen only, in which the sixteenth is the permanent spir- 
itual part as contrasted with the fifteen impermanent parts, 
like those of the moon, xii, 305, 4. 

The Sixteen (B) or Eleven Modifications. 

The epic (as already cited) gives the modifications as eleven 
in number. Apart from the usual explanations of these 
eleven, there is a passage, xii, 253, 11: “Three higher gunas 
are in all creatures, besides the five gross elements, with mind, 
which is essentially analytic, vyakaranatmakam, as the ninth, 
intellect the tenth, and the inner soul, antaratman, as the 
eleventh.” Here the commentator explains the three as igno- 
rance, desire, and action (avidya, karna, karma, gl. 9), though 
in the text bhava, abliiiva, and kala, are given as three addi- 
tions (gl. 2), with other departures from the scheme already 
recognized in what precedes. But apart from this special 
case, the fact remains that hi some parts of the epic, as in iii, 



213, 18 (p. 37), xii, 281, 20, only eleven modifications are 

On the other hand, sixteen modifications, eleven organs and 
five elements, as in the regular Sarhkhyan system, are fully 
recognized, as in xii, 311, 8 if., and elsewhere. 

There is, therefore, no uniform epic interpretation of the 

The Eight Sources. 

As given above from xii, 210, 28 and 311, 10, the mula- 
prakrtayah or eight fundamental procreative powers are the 
Unmanifest, intellect (“born of activity,” the result of the 
equilibrium being disturbed by tejas, energy), egoism, air, 
wind, light, water, and earth ; or in other words (the fine ele- 
ments being ignored, as usual), the five elements and self- 
conscious intellect as the first manifest production of the un- 
manifest produce everything. But in Gita, 7, 4, the “ eight 
sources ” are these elements plus mind, self-consciousness, and 
intellect. The terminology, it may be observed, is already 
broken up in the Gita. In this passage “ another source,” 
prakrti, is the jivabhuta, which is the same with one of the 
“ two spirits,” purusas, in 15, 16, one of which is ‘ all beings,” 
with a “ third spirit,” the Lord, Igvara, paramatman, added in 
17, who is not identified with the aksara but is “higher.” 
When, however, egoism is rejected in favor of spirit, as in the 
“Secret of the Vedanta, ’ then the group of eight appears as 
the six senses “ (the five senses which are perceptive, vijiianani, 
with mind as the sixth), intellect and spirit. Other groups 
of eight, like the last, seem to be based on this early grouping 
of productive elements. They are assumed in xiii, 16, 54, 
where Civa is “ the eight sources (above ‘ eight forms ’), and lie 
who is above the sources,” and they are personified in the per- 
sonal creation of xii, 341, 30 if., as “eight sages,” who are 
sources, though created from the elements : 

Marleir Angirac ca ’trih Pulastyah Pulahah Kratuh 

Vasisthaqca mahatma vai Manuh Svayambhuvas 

jfieyah prakrtayo'stau ta yasu lokah pratistliitah 



Compare 210, 28, mulaprakrtayo hy astau jagad etasv avastlii- 
tam. As already noticed, the system requires that the ele- 
ments here should be “ fine,” and this is occasionally expressed 
(see p. 129), but elsewhere the fine elements are ignored in 
this group of sources. Then the five (gross) elements are 
productive, which leaves only eleven modifications. 

The Vital Airs and Senses. 

In xii, 302, 27, there are seven breaths, the usual five and 
in addition an adliah anilah and a pravahah. Instances where 
ten and five vital breaths are mentioned have already been 
given. So with two, which are often the only airs recognized, 
as in xii, 240, 13. These are all old groups, 1 and represent 
as varied opinions in the epic as in earlier literature. 

Generally speaking, plants are ignored in the elaborate an- 
alysis of categories, but they are specifically mentioned at 
times. Thus in xii, 183 ft’., there is an account of creation. 
Water was the first creation after space. Water pressing made 
wind. The friction of wind and water made fire which became 
solid and thus formed earth. There are five sense-making ele- 
ments in all created things. Trees do not appear to possess 
them, but they really do. They have space or how could 
leaves comes out ? They have heat as is shown by withering. 
They have ears, for at the sound of thunder they lose leaves, 
and sound is heard only with ears. They have eyes for a 
■withe can wind its way, and there is no path without sight. 
They can smell, for good and bad smells, of incense, etc., make 
them flourish or decline. They taste, for they drink water. 
So all creatures have the five elements. The earth-element 
is seen in skin, flesh, boue, marrow, sinew; the fire-element, 
in energy, wrath, sight, heat, and digestive fire ; the air (or 
space) element in ear, nose, mouth, heart, and stomach (usu- 
ally not as here, 184, 22, but in all the apertures) ; the water- 

1 Even the ten are recognized in Chit. Br. xi, 6, 3, 5, dace ’me puruse prana 
atmai ’kadagah (called rudrah). These can scarcely he the organs, for as 
such they would iuclude the karmendriyas, which do not “ depart ” at death. 
The names are given above, p. 36. Compare the rudras of xii, 317, 5. 



element in slime, bile, sweat, fat, blood. There are five vital 
airs (winds) which cause a person to move, 184, 24-25 : 

pranat praniyate pram vyanad vyayacchate tatha 
gacchaty apano 'dhaq cai ’va 1 samano hrdy avasthitah 
udanad ucchvasiti ca pratibhedac ca bhasate 
ity eva vayavah panea cestayanti ’ha dehinam 

The five senses belong to the five elements ; one smells by 
reason of the earth-element; tastes because one has the ele- 
ment of water ; knows color through the eye as the fire- 
element ; knows touch through the wind. Smell is of nine 
sorts; taste is of six sorts; color (and form), of sixteen sorts 
(color as distinguished from form is of six sorts, white, black, 
bright-red, yellow, blue, yellow-red); wind has a double char- 
acteristic, sound and touch ; touch is the characteristic of wind 
and is of many sorts, viz., twelve; air (space) has but one 
characteristic, sound. But there are seven sorts of sound (the 
gamut) called sadja, rsabha, gandhara, madhyama, dliaivata, 
paneama, nisada. Whatsoever sound of drum, thunder, etc., 
is heard is contained in this group of seven sounds (notes). 2 

The more extended account of airs in the next chapter gives 
ten vital breaths or airs, though it describes but five, nadyo 
dacapranapracoditah, xii, 185, 15 (as noticed above, p. 86, 
with the correspondence in the third book). In xiv, 50, 42 
ff., the same (duplicated) account says smell is of ten sorts ; 
color (form), of twelve sorts; sound of ten sorts (the gamut 
and also “ sounds which are agreeable, disagreeable, and com- 

1 This is the later view that apana is the anus wind, payiipasthe 'panam, 
Pragna Up. iii, 5. 

2 On the six colors mentioned together in the Rig Veda, and the light of 
thirty-four kinds, see my article on Color Words in the Rig Veda, Am. Journal 
of Phil, iv, p. 190. Seven recitations or notes are recognized in the Chand. 
Up. ii, 22, 1 ; the roaring note is the Agni note ; the unclear is Prajapati’s ; 
the clear or definite is Soma’s ; the soft smooth, is Vayu's ; the smooth strong, 
is Indra’s ; the heron-note is Brhaspati’s ; the inharmonious, is Varuna’s. 
The names here are indefinite and apply vaguely to seven divinities. They 
are found also in other early literature. The epic names have no analogy in 
the Upanishads till the Garbha. On the other hand the epic grama, gamut, is 
late. Compare above, p. 13, van! ; also saptatantri vina, iii, 131, 11, “ the 
seven-stringed lyre,” called sadgrainaragadisamadhiyukta, in H. ii, 89, 68. 



pact”), although the two descriptions are almost identical. 
Each, however, has added new factors. The Anugita list 
betters the careless text above, whereby the sound called 
“Fifth,” pancama, stands in the sixth place (xii, 184, 39). 

The Five Subtile Elements. Gross and Subtile Bodies. 

The word for subtile element, tanmatra, is late and, as I 
think, its equivalent is not often to be understood. The ear- 
lier schemes were content with “ elements ” ; the later, or a 
divergent interpretation, introduced fine elements, suksmani, 
the latest have the classical term tanmatrani. Of course the 
commentator often interprets fine elements where none is 
mentioned. Thus, in xii, 205, 15, “ as the elements disappear 
on the destruction of the gunas, so intellect taking the senses 
exists in mind,” where subtile forms may be inferred, as 
they may be in xiv, 51, 13, where vigvasrj is doubtful (v. 1.). 
In xii, 252, 21, avigesani bhutani, and in xii, 311, 8 if., where 
the modifications of the five elements are again elements 
(above, p. 129), fine elements are recognized. In xiii, 14, 
423, viditva sapta suksmani sadangam tvarh ca murtitah, 
“knowing thee as having in bodily form the subtile seven, 
and having six limbs,” the commentator may be right in 
analyzing the seven as intellect, egoism, and five tanmatrani, 
as he does in the case of the Y ogin’s liiiga, soul, also said to 
have “ seven suksmas,” xii, 254, 7. 1 Elsewhere there are eight 
(powers?) characteristics of the subtile body of the Yogin, 
xii, 317, 6. 

But it must have caused surprise in the many schemes 
given above, that a clear indication of this theory is so often 
lacking where it would be most in place. The elements are 
simply mahabhutas (sic, or bhutani). Only the latest part 
of the epic has the technical word, i, 90, 13-14, where the 

1 Perhaps, however, the sevenfold knowledge of the Yogin is meant as in 
Sutra, ii, 27. The passage above, xiii, 14, 423, is a copy of xii, 254, 15, where 
the seven are explained as senses, objects, mind, intellect, mahat, the unmani- 
fest, spirit (the six are here explained as all-knowing, content, knowledge 
without beginning, independence, ever-clear sight, endless power). 



spirit, ksetrajna, is connected with the tanmatras before birth 
in the body; and xiii, 14, 202, where the order of Cpiva’s 
creation is “ mind, intellect, 1 egoism, the tanmatras, and the 
organs.” 2 

In xii, 202, 18 ff., when the soul leaves the body and takes 
another, it is said : “ A man leaving his body enters another 
unseen body. Abandoning his body to the five great (gross) 
elements, bhutesu mahatsu, he takes up a form also dependent 
on these, tadagrayam 3 eai ’va bibharti rupam. The five 
(senses) exist in the five great elements and the five objects 
of sense, in the senses.” Here there is another body, but it is 
composed of the same great elements and no other elements 
are recognized. The new body is called a linga, 4 but so is the 
old, grotradiyuktah samanah sabuddhir lifigat tatha gacchati 
liiigam anyat, “possessed of hearing and other senses and 
having mind and intellect he passes out of one body to 
another,” gl. 14. 

Elsewhere it is said that the beings that pass out of the 
gross body pass into a subtile, suksma, body, and are called 
suksmabliutani sattvani, “ fine beings,” which “ wander about 
like sunbeams,” superhmnan, atimanusani, xii, 254, 1-8 (sattva 
is bkutatman). The passage in xii, 345, 14 ff. has already 
been referred to. Here the sun is the door (as in the Igii) and 
the dead become paramanubhutiih, then manobhutah, and then 

1 Here mati stands for buddhi, as it does in xii, 202, 21, sarvlini cai ’tani 
manonugani, buddliim mano 'nveti niatib svabhavam, “ the senses follow 
mind, mind follows intellect, intellect follows the pure entity (here equiva- 
lent to paramah svabhavah of 203, 1). 

2 The word tanmatra occurs only in late Upanishads, according to Col. 
Jacob’s Concordance (his reference s. panca° includes Maitri, iii, 2). To the 
last, Garbe, in his Samkhva-Philosophie adds (p. 239) Katha, iv, 8, referring 
to Regnaud, Materiaux pour servir a. l’lnstoire de la philosophic de l’lnde, ii, 
31, 32. This is an error. The Katha knows nothing of tanmatras. Prajna 
must be meant, where matras are mentioned, iv, 8. 

3 Compare tan-matram, but in the passage cited, tad must refer grammati- 
cally to the great elements. 

4 So in xii, 307, 18, the Yogin, still in his gross body, becomes quiet as a 
lamp in a windless place, shines like a lamp (or is like a stone or piece of 
wood). When he shines forth and is nirlingah and moveless, he would not be 
reborn. Here linga seems to be merely a distinguishing mark. 



traigunyahinah, and enter Vasudeva (nirgunatmaka), the sarv- 
avasa (compare Igavasya), the home of all (or dwelling in 
all). We may compare Vasudeva derived from sarvabhuta- 
krtavasa, xii, 348, 94. The Yogin soul, “clothed in seven 
subtile things,” has also been referred to above, p. 39. 

In these cases there is evidence of a general belief in a 
subtile body, but evidence against a general belief in subtile 
elements, negative, of course, but rather strong when the 
elements called great beings (not necessarily gross, implying 
antithesis of subtile) 1 are said to be the constituents of the 
second body. I add another similar case where no mention is 
made of subtile elements, though the elements and the subtile 
post-mortem body are discussed, since it is an interesting pas- 
sage in itself and also offers a particularly convenient oppor- 
tunity for the introduction of the idea of subtile elements, 
but no such idea is suggested. 

The discussion begins with an account of creation, explains 
the five elements, and proceeds with an argument in regard to 
the psychic agent. Life, it is said, is invisible and the ques- 
tion comes whether there is any vital, jlva, spirit, and how it 
survives apart from the body, when the latter “ passes into the 
five elements ” (i. e., into the gross elements, tasmin pancatvam 
apanne jivah kim anudhavati, xii, 186, 10). “When a man’s 
body has been eaten by birds, or has fallen from a cliff, or has 
been burned, how can life come to him again, kutah sariijiva- 
nam punah, 13. If the root of a cut-down tree does not grow 
again, but only the seeds of the tree grow, how can the man 
(cut-down) reappear ? The seed alone, which has been started 
previously, that remains in existence ; the seed comes from a 
seed, but dead men perish when they die,” 15. 2 “ No,” says 
the teacher, “ there is no destruction of the vital spirit, jlva. 
The vital part of a man, pram, enters another body ; the body 

1 The application of great in mahabhuta is expressly said to he (not in 
antithesis to subtile, but) on account of their unlimited character, amitanam 
mahagabdo yanti bhutani sambhavam, tatas tesam mahabhutafabdo 'yam 
upapadyate, xii, 184, 3. 

2 Compare BAU. iii, 9, 28, retasa iti ma rocata . . martyah svit mrtyuna 
vrknah kasman mulat prarohati. With the fire-simile, cf. Qvet. i, 13. 



alone is destroyed. The vital spirit supported by the body, 
gariragrlto jivah, is not destroyed when the body is destroyed ; 
for it is like the flame when the wood is burned ” (implying 
that though invisible it exists). “ Just so,” says the objector, 
“ it is like the flame, but no flame is apprehended when the 
wood is used up, and I regard such a fire, when the wood is 
used up, as destroyed, since it has no visible course, nor proof 
(pramana), nor thing to hold to,” samsthana. To this the 
answer is : “ The fire is not apprehended, because it has dis- 
appeared into air without a support. So the vital spirit, on 
abandoning the body, exists like air, 1 but like fire it is not 
apprehended, because of its subtilty, suksmatvat; the vital 
breaths are upheld by fire and this fire must be regarded as the 
vital spirit. When breathing is restrained, the breath-uphold- 
ing fire is destroyed. When the bodily fire is destroyed, then 
the body (deham, n.) becomes senseless and falls and becomes 
earth, yati bhumitvam ; for earth is the place it goes to, ayana. 
Breath and fire go to air, for these three are one; the pair (of 
other elements) is fixed on earth. These (elements) assmne 
form only in connection with bodies (either mobile or im- 
mobile, 187, 9-10). . . . The five senses are not universally 
found 2 (and the body’s resolution into elements does not 
affect the soul) ; the inner soul alone carries the body, it alone 
smells, tastes, hears, etc. The inner soul is (not local but) 
found in all the parts of the body, presiding over that (mind) 
which has five (characteristics), in that (body) which consists 
of five (elements) . . . The soul does not die when the body 
perishes.” 3 

This is Paramatman doctrine, ib. 23, and since from the 

1 xii, 187, 6, jivo hy akayavat sthitah (sarvagato nityay ca, comm.), reminds 
one of BATJ. ill, 2, 13, akayam atma, only the strange Buddhistic assumption 
(of Karma alone remaining) is here carefully guarded against, though the 
preceding simile suggests the soul’s fate to be that in the Upanishad. 

2 Literally: “ In respect to what you are saying (whether the operation of 
mind and senses indicates an agent) there is no general application of the 
five,” 187, 19. 

3 mithyai ’tad ahur mrta ity abuddah : dayardhatai ’va ’sya yarirabhedah, 
187, 27. 



beginning of the discussion where the elements are introduced, 
184, 1 ff., to the close as given above, there is every opportu- 
nity to introduce the fine elements, it is evident they have no 
place in this system. We must either assume, therefore, that 
they are known in some parts of the epic and are not known 
in others, owing to a difference historically, or that they are 
taught and not taught in different passages, owing to a funda- 
mental doctrinal difference. The synthetic interpreter is wel- 
come to either horn of this dilemma. 

The orthodox popular belief, which of course is also taught 
in the epic, is that one can go to heaven with a “ divine form,” 
as in xviii, 3, 42. In xvii, 3, 22-28, one goes to heaven 
“ with Ins (human) body.” The reason may be that explained 
in the words 1 “ because of God’s residence in them, the gross 
elements are eternal.” These life-breaths and so forth exist 
eternally even in the other world, for a (fruti says so, in the 
words : “ Even when gone to the other world the life-breaths 
of incorporate beings always (exist),” xv, 34, 10 (text, above, 
p. 25). 

The body comes, according to the epic, from earth alone or 
from various elements. According to the scheme given above 
from xii, 184, 4, the body is made of earth. So the ear comes 
from air; the eye from the sun, etc., xii, 276, 11, tasya bhu- 
mimayo dehah. Compare xii, 240, 7, “ from earth the body, 
from water the fat, from light the eyes.” Here wind is the 
support of the two vital breaths, pranapiina^rayo viiyuh, and 
air (or space) is in the holes, khesv akagam, of corporate 
beings, a scheme of creation which attributes the “great 
beings ” (elements) to the “ first creation ” of a personal 

In xii, 306, 5, the characteristics of male and female parents 
are traditionally 2 three each, as inherited by the offspring: 

1 mahabhutani nityani bhutadhipatisariijrayat, xv, 34, 5. 

" fufruma . . . vede fastre ca pathyate. It is added : “ Authoritative is 
what is delared in one’s own Veda, svavedoktam, and what is read in the 
fastras,” a restriction as to the Veda not elsewhere admitted. 




bone, sinew, marrow from the father; skin, flesh, and blood 
from the mother. But in §1. 24 it is said that skin, flesh, 
blood, fat, bile, marrow, bone, and sinew are all eight pro- 
duced by the male, 1 gukrena prakrtani. Here tradition is set 
aside for the sake of the new philosophy. 

The growth of the body is described in xii, 321, 114 ff., the 
seed and blood, male and female, uniting produce a flake, 
kalala, which becomes a bubble, budbuda, which develops into 
a lump, pegl. From this lump come the limbs; from the 
limbs, nails and hair. At the end of the ninth month, “ name 
and form (individuality) ” are born. 2 

Besides one subtile body, the epic may recognize two, as do 
the Vedantins and later Samkhya pliilosophers (Garbe, Sam- 
kliya Phil., p. 267). But the following text, I think, scarcely 
supports this interpretation of the commentator : “ When the 
spirit in a body is out with rajas, it would wander about, like 
sound, with a body ; having a mind unaffected by the result of 
action (the spirit) is established in Prakrti because of its free- 
dom from affection.” 3 The commentator thinks that when the 
spirit is in Prakrti it has a very minute body, different from the 
span-long or thumbkin body. 4 This is his explanation also of 
the unfinished sentence in xii, 254, 13. In 12 one sentence 
ends with the statement that unclarified spirits “ do not see the 
bhutatman in bodies.” Then in 13, “ those who are devoted 

1 Apparently a clear contradiction of the preceding, but excused by the 
author on the plea of understanding the inner meaning, and not the words 
alone, of Veda and Qastra, grantharthatattva ! 

2 The same process is described in late Samkhya texts (Garbe, p. 273). 
Compare the Garbha Upanishad. “ Name and form ” is a phrase sometimes 
amplified : “ The Lord creates name and form and acts,” xii, 233, 25-26 (as in 
Brh. Up., i, 6, 1, nama ruparn karma, which may be referred to here, yaduktam 
vedavadesu . . . tadantesu). 

8 rajovarjyo 'py ayam dehi dehavaii chabdavac caret, karyair avyahata- 
matir vairagyat prakrtau sthitah, xii, 217, 21. The next half-stanza, adehad 
apramadac ca dehantad vipramucyate, is interpreted by the commentator to 
mean “the three bodies (sthula-suksma-karana) being abandoned, the soul 
(without body), because of its mental freedom, is released definitively.” 

4 The subtile body is “ span-long ” in xii, 200, 22; “the size of a thumb” 
it wanders by reason of its connection with the lifiga, v, 46, 15, and 27 ; xii, 
235, 175, angustliamatrah purusii dehasthah. See above, p. 32. 



to Y oga-Castra, desirous of seeing that soul, — (things) with- 
out breath, (things) without form, and what (things) are 
like thunderbolts.” Here the commentator takes the three, 
anucchvasani, amurtani, yani vajropamany 'pi, as bodies devoid 
of intelligence, suksrna or subtile boches, and, thirdly, bodies 
indestructible even in the atonic destruction, or karana- 
garirarii, rvith atikramanti, overpass, to be supplied in the 
text. If anything is supplied it is “ they see,” but the pas- 
sage is clearly without sense as it stands and probably repre- 
sents a later and awkward interpolation of the three bodies. 

The Colors of the Soul. 

The color of the soul is assumed through its union with 
the body, in the same w r ay as wiien one near a fire gets a red 
color, xii, 202, 17. The incorporate spirit, dehin, is said to be 
without color, but it is tinged with the fruit of acts, and so is 
said to attain to color, varna, which is of course specifically 
“darkness.” “ But when the creature by means of knowledge 
puts off darkness, born of ignorance, then appears eternal 
Brahman ” (pure, without color, 201, 26). “ As wind,” it is 

said, “ becomes colored with dust and so itself colors all the 
air (space), thus the spirit, jlva, without color, because of 
acts’ fruits becomes color-tinged,” xii, 280, 9 ff. 

This simple idea of pure white soul (as in Cvet. Up. iv, 1) 
being darkened by contact with impure darkness-born not-soul, 
and eventually becoming clear and colorless again, is worked 
up into a confused theory of spirit-color in the next chapter, 
where jlva, spirit, has six colors, sadjlvavarnah, xii, 281, 33, as 
follow's : “ Spirit has six colors, black, yellow-green (or grey), 
and blue, the middle color ; red, more helpful and good, bright 
yellow, and, best of all, white. White is best, spotless, without 
sorrow, leading to success. . . . The course creatures take is 
made by their (spiritual) color. Color is caused by one’s 
former acts (Time, as often, represents the Karma). The 
dark color leads to a low course and hell. After hell the 
spirit attains yellow-green (liarit = dliumra). When jlva is 
endowed with sattva it casts off tamas (darkness) by means 



of intelligence, and after blue attains to red and lives as a 
human creature.” Then the spirit attains to yellow as a god, 
returns to hell, and goes on in the same way to white, finally 
surpassing the three states (gunasj). 1 The inner meaning of 
this passage, according to the commentator, is that when the 
spirit has the three gunas, tamas, rajas, sattva, in quantitative 
proportion to this sequence, the result is that the spirit is 
black ; but in the order tamas, sattva, rajas, yellow-green (or 
grey); rajas, tamas, sattva, blue; rajas, sattva, tamas, red; 
sattva, tamas, rajas, yellow ; sattva, rajas, tamas, white. The 
whole theory, which is alluded to again in 292, 4 ff., seems 
to be an elaboration of the simple thesis of the preceding 
section given above. In the passage following, the “higher 
color ” is gained by “ pure acts,” varnotkarsam avapnoti narah 
punyena karmana. The identification of light with heaven 
(“bright-yellow gods,” above) is as natural as that of dark- 
ness with hell. Thus xii, 190, 1 ff., after it is said that “truth 
is light and darkness is lies,” we read : “ Light is heaven and 
darkness is hell ; man gets a mixture of both in this life, truth 
and lies.” Compare Patau jali’s Aphorisms, iv, 7: “Yogin’s 
work is neither white nor black.” I see no support in the text 
for the elaborate explanation of the commentator, as recorded 

In xii, 303, 46, there are “ three colors, white, red, and black, 
with which are affected all things in Prakrti.” Here these 
are set parallel to the gunas (red apparently corresponding to 
energy, rajas), as signs of the soul, which goes to hell if it is 
tamasa, humanity if rajasa, heaven if sattvika ; apparently an 
intermediate view between the six colors and the simple an- 
tithesis of pure and impure, white and dark. The tricolored 
being is known in a phrase common to epic, v, 44, 25, and 
Upanishad, <yvet., iv, 5. 2 

1 The commentator, instead of taking the states to be gunas, takes them as 
waking, sleeping, and deep slumber, ending in tunja, the fourth state. 

2 Epic text, xii, 303, 46 : yuklalohitakrsnani rupany etani trini tu sarvany 
etani rupani yanl ’ha prakrtani vai. £vet. Up. iv, 5: ajam ekam lohitafukla- 
krsnam bahvlh prajah srjamanam sarupah (Muller gives the varied readings 
in his note, SBE., vol. ii, p. 250). For v, 44, 25, compare above, p. 28. 



The Five Faults of a Yogin. 

In xii, 241, 3 if., the faults of Yoga as known to the seers, 
Kavis, are desire, wrath, greed, fear, and sleep, kama, krodha, 
lobha, bhaya, svapna, two added to an ancient trio. In xii, 
301, 11, the five Yoga faults to be “cut off” are registered 
as raga, moha, sneha, kama, krodha. In xii, 302, 55, the 
“path-knowing Kapila Samkhyas” give as the five faults, 
kama, krodha, bhaya, nidra, gvasa. In xii, 317, 13, the five 
faults are simply the actions of the five senses. See also the 
list above, p. 119. 

Patanjali, ii, 3, recognizes five klegas “to be abandoned” 
(heyah), avidya ’smitii ragadvesa ’bhinivegah. Five to be “cut 
off ” and “ to be abandoned ” are also recognized in the Dham- 
mapada, 370, panca chinde, panca jahe. In the epic the “ five ” 
are known as such, but different expositions explain them 

Discipline of the Yogin. 

The perfected Yogin, who, by means of the sevenfold dha- 
ranas, methods of fixing the mind, has overcome seven, the 
elements, egoism, and intellect, attains to “complete and 
faultless illumination,” pratibha, in which state he surpasses 
the gunas and performs miracles. These technical terms of 
the Yoga are only two of many found in the later epic. 
Pratibha, upasargas, the eightfold power, the various com- 
fortable “sittings,” calculated to induce concentration of 
thought, e. g., vlrasana, the eodanas, “ urgings ” (by which 
one controls the breaths), the “ pressing of breaths ” into the 
heart-canal, or into the space between the brows, the fixed 
hours of exercise in mental discipline — all this Yoga-machin- 
ery is as well known to the epic re writers as to Patanjali. 
That the epic here precedes the Sutra-maker may be inferred 
from the fact that in the matter of “ faults ” (above) and in 
other technical terms it does not always follow the latter, 
though it has the Sutra terminology to a certain extent. 
But, on the other hand, there can be little doubt that the 
epic-writers were steeped in Yoga-terms and used to Yoga- 



practices of extreme refinement, for they reveal a very inti- 
mate acquaintance with Yoga-technique. Over against these 
adepts, or scientific Yogins, stand the vulgar ascetics, whose 
practices consist simply in the austerity of painful posturing. 
The latter forms are antique, and continue, of course, through 
the whole epic, as indeed they continue till now in India ; but 
in contrast with those who practise the scientific rules of the 
skilled Yogin, the “ one-legged, up-arm ” ascetic belongs to 
the vulgar cult, inherited as “ Veda-enjoined penance," where 
the wretch is not so much engaged in control and samadhi, 
graduated concentration, as in mortifying himself to get power 
or win God’s grace. Even Vishnu thus stands by his “ eight- 
finger-high-altar,” and performs austerities, “ standing on one 
leg, with upturned arm and face ; ” and it is the worshippers 
of such gods who retain as their sole means of winning divine 
grace the same sort of practices. Ko sharper contrast can be 
imagined than the two disciplines, that of tire votary and that 
of the scientific student of psychology (whose theology rests 
in Brahmaism), as presented in the epic. 1 

The Destructible and Indestructible. 

Both spirit and the Source according to the Saiiikhya system 
are eternal and indestructible, xii, 217, 8; Gita, 13, 19. They 
are therefore not created things. But spirit in other passages 
is a “ created thing ” and so is the source, xii, 205, 24. For 
according to the Brahmaistic interpretation, both of these are 
destructible so far as their entity goes. The twenty-fifth is 
reabsorbed and the twenty-fourth is also absorbed into Brah- 
man, xii, 808, 7 ff. See above, pp. 134, 137. “ Lord Time's 
Retaking” pratyahara, is the name given to the cosmic re- 
absorption as explained in xii, 234, 1 ff. The universe becomes 
subtile and metaphysical, adhyatma. All things are first 
burned and enter the condition of earth, till earth looks bare 

1 The chief chapters to be compared will be found in Qanti (237, 241, 317 ; 
also pp. 44, 107, above), but for details I must refer to a paper read at the 
Meeting of the Oriental Society in April, 1900 (to be published in the Jour- 
nal, vol. xsii). 



as a tortoise shell. Then water takes up earth ; fire, water ; 
■wind, fire ; air, wind ; mind, air (-with sound, etc., i. e., mani- 
fest mind passes into unmodified mind) ; the moon, as sarii- 
kalpa or fancy, swallows mind, citta ; then Time swallows this 
as knowledge. 

Up to this point the retroaction is at least intelligible but it 
is interrupted here by a revealed text: kalo girati vijilanarn 
kalam balam iti grutih, balaih kalo grasati tu, tain vidya 
kurute vaee, “Time swallows knowledge, poiver sw alio tvs 
Time, and Time swallows power; then Wisdom overpowers 
Time.” Finally: “The Wise One puts into himself the 
sound, ghosa, of air or space.” That is unmanifest, highest, 
eternal Brahman, “and so Brahman alone is the recipient of 
all creatures.” 1 

The Gods and the Religious Life. 

The orthodox Brahman’s insistence on the four stadia of 
life is found in the normal attitude of the poets. Opposed to 
this is the direct teaching that these stadia are quite unneces- 
sary, xii, 32T, 26-27 : “ In the first stadium one can be per- 
fected, what use is there of the other three ? ” Compare iii, 
297, 25, mil dvitlyam, etc. 

In some passages the god Brahm&n is indestructible and 
self-created; in others he is a creation; in some he is below 
Vishnu, in others above him ; in some, he is below Civa ; in 
others above him. 2 Brahman, again, appears as the equal of 

1 fl. 17 : evarii sarvani bhutani brahmai ’va pratisarncarah. This absorp- 
tion is the counterpart to the personal creation of Brahman (see p. 112), from 
the “ Seed made of Brahman-glory, whence all the world,” 213, 1. I do not 
pretend to understand the final process of reabsorption described above : 
tkajasya tada ghosam tarii vidvan kurute ‘tmani, tad avyaktam param brahma 
tac chafvatam anuttamam. The eternal sound here implicated in Brahman 
may be that “Word without beginning or end, Wisdom, uttered by the Self- 
existent, from which, as Veda-sounds, the Lord (as cited in the note, p. 178) 
in the beginning creates names, forms, and acts,” xii, 233, 24-26. 

2 In xii, 340, 116, Brahman knows that Vishnu is greatest ; but in xii, 285, 
165, Vishnu is unable to comprehend the greatness of (,-iva. Compare on the 
mixed ideas concerning Brahman, Holtzmann’s essay, ZDJIG. xxxviii, p. 167 S. 
I cannot agree with the author in the opinion that Brahman is the chief God 
of the “ older epic,” but only of the older tales incorporated into the epic. 



the other two gods in the trinitarian theosophy, which is rep- 
resented in the epic, but only sporadically and in its latest addi- 
tions . 1 He is sometimes looked upon as the chief of all gods, 
but his supreme attributes are in other passages taken by his 
later rivals. Three stages are clear, with a top story added 
last of all. The earliest tales received into the epic know 
no god higher than Brahmdn, the later pseudo-epic knows 
no god equal to (a Paqupata) Qiva. Between the two lies the 
mass of the epic teaching, where supremacy is given to a sec- 
tarian Vishnu. The very latest additions to the epic adopt a 
synthetic view and make of this religious olla podrida one har- 
monious whole, where all three great gods are one. 

Arjuna is a form of Vishnu. He is taught this with won- 
der and great amaze in the sixth book. But our amazement 
at his amazement is still greater, for this doctrine, apparently 
so new to him, was revealed to him long before, in the third 
book, and on that earlier occasion he appeared fully to appre- 
ciate the fact that he was divine and identical with Krishna, 
facts which in the sixth book he has totally forgotten . 2 

Heaven and Hell. Death. 

Inconsistent as is the Karma doctrine with the notion of 
heaven and hell, the Hindu, like Pindar, successfully combines 
the two beliefs by imagining that metempsychosis follows the 

1 For the usual caturmurti, compare iii, 203, 15; vii, 29, 26; xii, 335, 8. 
In iii, 272, 47, is found the only definite expression of the late trinitarian 
belief in a trimurti, an interpolated section (compare my Religions of India, 
p. 412); though it may be implied in i, 1,32 and xiii, 16, 15, but only here 
till we reach the Harivarija, 2, 125, 31. It appears first in the later Upani- 
shads, or in late additions, as in Maitri v (as distinguished from the close of 
iv), above, p. 46. Among other religious novelties the pseudo-epic introduces 
Citragupta, Death’s secretary, xiii, 125, C ; 130, 14 ff. In several points, such 
as in this and in grammatical peculiarities, the Anujasana shows itself later 
in some parts even than £anti, all ignored, of course, by the synthesist. 

2 Compare iii, 12, 10. In this passage, Arjuna exalts Krishna as the su- 
preme Lord of the universe, and Krishna in turn identifies the two : yas tvarii 
dvesti sa marii dvesti, etc., ib. 45 (Vishnu says the same thing almost to Rudra 
in xii, 343, 133; yas tvarii vetti sa marii vetti, yas tvam anu sa mam anu). 
Arjuna’s godhead is proclaimed to him in iii, 41, 35, 43 ; 47, 7. On the hymn, 
iii, 12, compare Lassen, Ind. Alt., i, p. 489. 



penalty of hell, or reward of heaven. The two views stand 
sometimes separate, however, and the hero is promised an 
abode in Indra’s heaven without any allusion to metempsy- 
chosis ; or one is promised a high or low birth hereafter with- 
out allusion to the older teleological fancy. Ordinarily in the 
former case, the rule is that a good man goes to heaven and a 
bad man goes to hell, as in the Upanisliads, e. g., Mund. i, 
2, 10, and in the epic generally. But in one exegesis quite a 
different view is taken. The idea here is that a fairly good 
man goes first of all to hell ; while a man who on the whole is 
rather sinful than good goes first of all to heaven. Afterwards 
the good man goes to heaven and the bad man goes to hell. 1 

The popular notion of the Yogin is not at all that of absorp- 
tion into Brahman. “ Grieve for the living, not for the dead ; 
this pious hero after his death, like a Y ogin, has become a be- 
ing with a human body and shines glorious like a king.” 2 In 
heaven there are cool breezes and perfume, no hunger, thirst, 
toil, old age, nor sin, but “ eternal happiness,” in heaven, which 
is here, in contrast to hell, the “highest place,” xii, 190, 13- 
14. So in the Sabhas. The Yogin “revels in joy, knows no 
sorrow, and rides around on high in a heavenly car, attended 
by self-luminous women,” xiii, 107, 130 (compare the ramali 
sarathah of Katha Up. i, 25). This is the happiness of a Yogin 
after death, a view of course diametrically opposed to that of 
the philosophy taught elsewhere, for it is taught as final, not 
as preliminary. 

In various passages it is taught that a good man should aim 
at attaining to heaven. This too is not put forth as a half-view 
with a reservation, as in the case of the Upanisliads. But in 
other cases it is expressly just such a half-view. 3 Heaven is 

1 bhuyistham papakarma yah sa purvam svargam acnute, etc., xviii, 3, 14. 

2 tam aindavim atmatanum . . . gatah, yii, 71, 17. Compare xii, 332, 53, 
vayubhutah prareksyami tejorafim diyakaram (not here to the moon, which 
changes): “In the form of wind I shall enter the sun” (to live with the 
seers) ; yatra na ’vartate punah (50), “whence there is no return.” 

3 Here it may be objected : But this is for warriors, and even in the tlpan- 
ishads those that worship Prajapati as matter instead of spirit are materially 
blessed. This raises the question again which I touched upon at the outset. 



here a good place for good but unintelligent people, but it is 
scorned by the philosopher. “ I have done with heaven, away 
with thee, heaven, whither thou hast come,” says an enlight- 
ened king; “let the priest receive my merit if he wishes,” 
xii, 199, 77-78. The priest, orthodox, is recognized as still 
striving for heaven and likely to go to hell, in the old way : 
“ Hell is where priests go,” it is said rather bluntly, ib. 14-15, 
nirayarh nai ’va yata tvam yatra yata dvijarsabhah, yasyasi 
Brahmanah sthanam. For of all the heavens of all the gods 
it is said, “these are but hells to the place of the Highest 
Soul,” xii, 198, 6. 

All kings but one go to Yama’s heaven in the Sabha 
account ; 1 in the battle-scenes most of them go to India’s 
heaven. But in vi, 16, 20, they go to the Brahma-world. 
Again, the heaven one goes to depends either on one’s gunas 
(as explained above), or, according to where one dies (Tlrtha), 
or, as a third explanation, according to the place in the body 
through which the soul escapes at death. If it goes through 
the feet, one goes to Vishnu’s place ; if through the arms, to 
India’s place ; if through the crown, to Brahmdn, etc., xii, 318, 
1 ff. (with vigvedevan in 5, common in the pseudo-epic). 

Death, it may be observed, is usually a male ; but in vii, 53, 
17 and xii, 258, 16-21, a female. There are here two accounts 
which, though together opposed to the view held everywhere 
else, are of critical value, not on this account (for a poet may 
perhaps be allowed to unsex death), but on account of their 
being almost identical, two versions of one tale, one bearing 
traces of greater antiquity than the other. 2 

In one part the warrior auditors are taught the deepest mysteries, in another 
they are taught what is not taught in the Upanishads except as introduc- 
tion to true teaching. Synthetically considered, the epic teaches nothing 
systematic in these varying expositions. 

1 Yama’s home is here a heaven of delight, elsewhere in the epic it is a hell 
of horrors. 

2 The account in Drona is here the later of these two similar scenes, as has 
been shown by Holtzmann, ZDMG. xxxviii, p. 218. In philosophy, death is 
the dissyllabic Ego as opposed to the eternal, immortal, three-syllable non- 
ego, or mama versus namama (" this is mine ” is a thought deadly to truth, 
and untruth is death), xii, 13, 4 and xiv, 13, 3 (identical passages). 



The Cosmic Egg and Creations. 

According to the old belief, the universe comes from a 
cosmic egg. The philosophical schemes, of course, discard this 
egg, but we hear of it in the popular accounts often enough 
and meet it in the first verses of the epic. Occasionally, how- 
ever, in the personal creation, which stands in so sharp con- 
trast with the more philosophical schemes, this becomes a sub- 
ject of controversy. Thus in xii, 312, the “ Unmanifest ” is a 
person, who first creates plants as the food of all incorporate 
things. “ Then he produced Brahman, born in a golden egg. 
Brahman lived in the egg a year. Then he came out and put 
together the four forms of all beings, and earth and heaven 
above — as it is said in the Vedas, dyavaprthivyoh 1 2 — and 
then the middle space. After this he created egoism, a being, 
bhuta, and four sons besides, who are the fathers’ fathers. 
The gods are the sons of the fathers ; by the gods the worlds 
were filled. Egoism, he that stands in the highest, created 
fivefold beings, earth and the other elements.” Several verses 
follow on the impossibility of the senses acting alone (“ the 
organs do not perceive, etc. Mind alone sees. Mind is the lord 
of the senses,” etc.). 3 Here the egg-born creator is acknowl- 
edged in a scheme which is a mixture of mythology and philos- 
ophy. But in xiii, 154, 16 ff. : “ Some fools say that Brahman 
was born of an egg . . . but that is not to be regarded. How 
could the unborn be born? Air-space is the egg, according 
to tradition, and out of that was born Brahman, the forefather. 
(He required no support, for he is) personified consciousness, 
the Lord. There is no egg ; there is Brahman . . . the unman- 
ifest eternal Creator Lord” (15). This passage is not merely 
an allegorical interpretation of the egg-myth ; for in the former, 
Brahman creates space after he is born of the egg from which 
he is born, while here the egg is space. The number of crea- 

1 That is, the Vedic form implies the truth of heaven and earth as here 

2 In this passage, ete vigesa mahabhutesu, 312, 12, repeats the first half- 

stanza of 311, 14, cited above, p. 129. 



tions in philosophy I have already discussed. They are given 
as nine, or again as five. 1 

The Grace of God. 

The belief in the saving grace of God is found only in the 
later Upanishads. It asserts that one sees the Self (or Lord) 
by the grace of the Creator, Katha Up., i, 2, 20 ff. ; Cvet., iii, 
20 ; vi, 21 ; Murid., iii. 2, 8. One is chosen, and cannot get 
salvation by knowledge alone. This general view is that 
maintained by the epic poet, who says : “ The Vedas and 
Orders, though established on various opinions, nanamatasa- 
masthitah, unite in worshipping Spirit as the personal God 
by whose grace one is saved.” So again : “ That man can see 
Him, to whom He gives His grace,” yasya prasadam kurute 
sa vai tam drastum arhati, xii, 337, 20, (a verse found also in 
the pseudo-Ramayana). The grace of God is here the chief 
element of salvation, opposed to what is recognized as the 
severer school of those who attain salvation scientifically 
either by knowledge of soul or of God. This older system in 
the Upanishads is represented by those who are saved by 
knowledge alone ; in the epic, by like-minded men, who have 
worked out a system or science of salvation, and depend wholly 
on this science, jnana, or on ascetic practices, tapas, yoga, 
super-added to this science. Both of these are recognized as 
older systems in the epic, compared with the grace-of-God 
theory, and practically they are thrown over by the adherents 
of the latter school, who, however, differ from their ancestors 
in the Upanishads by a clear mark of lateness, in that they 
specify that the God whose grace saves is Krishna alone. 
Salvation not through knowledge, even of God, not through 
the grace of God, but through the grace of the man-god is the 
saving way, the easier way, or as it is called in the Gita, the 
“less troublesome way,” 12, 5. 

Side by side stand in the epic these two great modern modi- 

1 These are the modifications of God, avidyasargas and Tidyasargas, fiye 
in number in xii, 303, but when the account is repeated in 311, nine in aiL 



fications of the older Upanishads: there, knowledge, wisdom, . 
jnana, vidya, contrasted with the later grace of the “ Creator- 
Spirit,” at most recognized as (,'iva. Here, the Samkhya- j 
Yoga system, contrasted with the later Krishna cult. “I ( 
will release thee from all thy sins, grieve not,” says the man- 
god, Gita, 18, 66. But the Yogin replies: “Sink or swim, 
let one put his trust in science alone,” xii, 237, 1 and 
238, 1, and claims that he is purified not by Krishna but 
by Yoga knowledge, rejecting even the purity induced by 
bathing in the sacred pools (for his purity is “obtained by 
knowledge”), which elsewhere in the epic are said to purify 
from all sin. 1 But inasmuch as the Yogin's science postulated 
what the Samkhya denied, a personal God, the former became 
a bridge between the atheist and the devotee, a bridge, how- 
ever, occasionally repudiated by the latter, who does not always, 
as usually, claim that he is thus philosophic, but exclaims: 
“By Samkhya and by Yoga rule I meditate the way of God 
and find it not,” xii, 352, 7-8. 

The irreconcilable difference between the Samkhya and the 
faith of the Krislmaite could be removed only by mollifying 
one of these extreme views. Either the atheistic (or even 
Brahman) philosopher had to win over the adherents of the 
man-god to renounce him and return to the “ ship of salvation 
of knowledge,” or the devotee, haring admitted that the 
Yogin’s Spirit was God, had to identify his Krishna with that 
Purusha I g vara. Late as are all the purely philosophical 
chapters of the epic, they still show which power prevailed. 

1 There is of course, further, the ijivaite, who worshipped not Krishna hut 
another as the highest God, not to speak of those that remained true to 
Vedic tradition and went for salvation no further than sacrifices and gifts. 
There are also, within the group of philosophers, those who recognized only 
the earlier twenty-five principles, and those who recognized twenty-six, as ex- 
plained above. There is also the fractional sectary, who regarded Krishna 
as the “ half of the fourth” of the “ root-abiding Mahadeva” (as tatstha , p. 
44, he creates existences, xii, 281, 61-62). All these divergent beliefs are 
represented in startling and irreconcilable antagonism in an epic concerning 
■which the unhistorical view is dass es aclite zu einer einheitlichen Auffassung 
abgerundete Elemente sind, welche das Epos bietet, Nirvana, p. 84 ! 



Faith absorbed unfaith. The religions philosophy of the epic 
is a successful attempt to uphold Krishnaism not only against 
the science of atheism, but against a deistic science that postu- 
lated God but saw no godship in Krishna ; a science which in 
its turn is technically elaborated, a long advance on the vague 
speculations of the Upanishads, but not yet as uniform as in 
the completed system. Krishnaism stands to Samkhya-Yoga 
chronologically as stands the later grace-of-the-Creator theory 
to the earlier knowledge of the Upanishads. But both epic 
Samkhya-Yoga and Krishnaism are later even than this modi- 
fication of Upanishad teaching. Latest of all is trinitarianism. 
Side by side stand all these creeds, each pretending to be a 
definitive answer, each forming part of the contents of a poetic 
vessel, into which have been poured the vinegar and oil of 
doubt and faith ; but : 

o£os t’ a\urf>d t’ ravrcu kvtcl 

Si\<xnaTovvT’ dv ov ir pocrtvvtTrois. 



alamkrtam fubhaih fabdaih 
samayair divyamanusaih 
cbandovrttaif ca vividhair 
anvitam vidusam priyam 

A Tale adorned with polished phrase 
And the wise lore of gods and men, 

With verses turned in various ways 
Replete, a joy to scholars' ken. 

Epic Versification . 1 

The poetry of the epic is composed in metres, chandas, of 
three sorts. The first is measured by syllables, the second by 
morse, the third by groups of morse. These rhythms ran the 
one into the other in the following course. The early free 
syllabic rhythm tended to assume a form where the syllables 
were differentiated as light or heavy at fixed places in the verse. 
Then the fixed syllabic rhythm was lightened by the resolution 
of specific heavy syllables, the beginning of mora-measurement. 
The resolution then became general and the number of morse, 
not the number of syllables, was reckoned. Finally, the morse 
tended to arrange themselves in groups and eventually became 
fixed in a wellnigh unchangeable form. Part of this develop- 
ment was reached before the epic began, but there were other 
parts, as will appear, still in process of completion. Neither 

1 I wish to acknowledge in beginning this chapter on epic metres the great 
help afforded me by Professor Cappeller of Jena, who put at my disposal a 
manuscript on the metrical forms in the epic, in which all the metres were 
located and the tristubhs of the first three hooks were analyzed seriatim. I 
need hardly say that this loan has materially lightened the labor of preparing 
the following sketch, a loan the kindness of which was the more appreciated 
as it was entirely unsolicited, though most gratefully received. 



of tlie chief metres in the early epic was quite reduced to the 
later stereotyped norm. The stanza-form, too, of certain 
metres was still inchoate. 

The mass of the great epic (about ninety-five per cent) is 
written in one of the two current forms of free syllabic 
rhythm; about five per cent in another form of the same 
class; and only two-tenths of a percent in any other metre. 
The two predominant rhythms, §loka and tristubh, are in 
origin the oldest Indie or pre-Indic rhythms, while of the 
others some are in turn early developments from the first epic 
rhythms. For convenience of reference, before discussing 
these rhythms in detail, I give a list of all those used in one 
or both of the two epics according as they are free syllabic 
(gloka, tristubh), fixed syllabic (aksaraechandas ), 1 mora-metre 
(matrachandas), and group-rhythms (ganacchandas). 

§loka: a stanza of two verses (hemisticlis) of sixteen 
syllables each, restricted to a certain extent as to the place 
where heavy and light syllables (or long and short vowels) 
are permitted. Originally the stanza consisted of four 
verses of eight syllables each and many traces of this di- 
vision, by independent “quarters,” padas, survive in the 

tristubh: a stanza of four verses of eleven syllables each, 
arranged with very little restriction (and consequently of 
various types) in the Mahabharata; reduced to one prevail- 
ing type in the Ramayana. Increased by one heavy- sylla- 
ble in each pada, this metre is called jagatl, but the two 
types are interchangeable in the same stanza. Fixed types 
of this metre are common in verse form, but rare in stanza 
form 2 except as given in the next group (of four-verse 

1 The fixed syllabic is called also varna vrtta, “syllabic verse” (vrtta = 

3 That is, pure in the form (a) and (b), xj xj xj xj xj ( ) ; 

(e) , \j xj ; (f) xj xj xj These 

are called (a) upendravajra ; (b) vangastha(bila) ; (e) palini; (f) vatormi; 

or (a) and (b) with the opening xj , called (c) indravajra and (d) in- 

dravahga, as they have eleven or twelve syllables, respectively. When (a) 
and (c) or (b) and (d) are mingled, the stanza is called upajati. 






with the verse filed as 

Tathoddhata, a tristubh w , \j v 

bhujamgaprayata, a jagati w w , w ,w 

drutavilambita, a jagati w w wva w u, w 

vaifvadevi, a jagatl , w 

rucira, an atijagati 1 w w , w w w w w v_/ 

praharsini, an atijagati , wwww, w vj 

mrgendramukha, an atijagati uvwj, \j\jj w <_/ 

asambadha, a ? akvari 2 

vasantatilaka, a {akvari w , w w w, w w vz 

malini, an atigakvari , w w 

fardulavikridita, an atidhrti , w w , w w, 

ww ; w, w 

matraehandas J 
(ardhasamavrtta) ] 

/puspitagra and aupacehandasika, stanzas of two verses, 
each verse having sixteen and eighteen morse in prior 
and posterior pada, respectively, the morse being ar- 
ranged in syllables more (puspitagra) or less (aupac- 
chandasika) fixed. 

aparavaktra and vaitaliya, the same in catalectic form, 
each pada being shortened by two mor®. 
matrasamaka, a stanza of four verses, each verse having 
v sixteen mor®. 


,(arya, aryagiti, upagiti), stanzas of two verses, each verse 
containing eight groups of mor®, the group of four 
mor® each, but with the restriction that amphibrachs 
are prohibited in the odd groups, but may make any 
even group and must make the sixth group, unless in- 
- deed this sixth group be represented (in the second 
hemistich) by only one mora or four breves; and that 
the eighth group may be represented by only two mor®. 
The metre is called aryagiti when the eighth foot has 
four mor® ; upagiti, when the sixth foot irregularly has 
' but one mora in each hemistich. 3 

1 That is, a jagatl with one syllable over, ati, or with thirteen syllables in 
the pada. The second atijagati above is sometimes called praliarsanl. 

2 That is, having fourteen syllables in the pada, fifty-six in the stanza. 
The atijakvari and atidhrti have fifteen and nineteen syllables in the pada, 

3 Brown, Prosody, p. 17, points out that this metre is almost that of Horace, 
Odes.iii, 12 : miserar | est neq a- I mori | dare lu- 1 dum neque I dul- 1 ci mala I 
vino, etc. ; and sic te | diva po- ( tens Cypri | sic fra- I tres Hele- I nae I lu- 
cida | sidera, etc., save that the sixth group is here of two mor®. 




Cloka and Tristnbh. 

J • • 


The number of verses in a 9 loka or tristubh stanza may be 
decreased or increased by one or two, respectively; but in 
the great majority of cases, two in a gloka and four in a 
tristubh constitute a stanza. Sometimes, however, where one 
or three hemistichs make a stanza, it is merely a matter of 
editing. Compare, for instance, i, 90, 22; i, 93, 19-21 with 
3,682-83; iii, 4, 17 with 234; iii, 111, 14 If., with 10,040, ff. 
But, on the other hand, no arrangement can always group the 
hemistichs into uniform stanzas. Thus in xii, 350, 49 if., five 
tristubh hemistichs follow three gloka hemistichs. A stanza 
of three hemistichs is apt to close a section, as in vii, 54 and 
187. In G. vi, 49, 55, there is one hemistich in excess because 
53 a-b Avere added to the original, and this is doubtless the 
cause of many such cases ; though it is also true that a half 
stanza is often found where there is no reason to suspect a 
later addition. Six padas in a tristubh occur occasionally. 

But in the case of the (jloka, the padas are metrically linked 
in pairs, while tristubh padas are metrically independent. 
The tjloka, therefore, is a couplet. Its two halves are metri- 
cally disjunct and may be treated as independent wholes. 
Each hemistich is a complete Averse. The two halves of this 
verse, the quarters, padas, of the whole stanza, are sometimes 
knit together into euphonic combination and a syntactical 
whole. But, relatively speaking, this is seldom the case. 
The unity consists rather in the fact that one half of the verse 
is metrically different from the other and cannot be substi- 
tuted for it, whereas in the tristubh any pacla can be substi- 
tuted, if the sense permits, for any other . 1 The different fall 
of the cloka padas may be seen A'ery Avell when the words are 
almost identical : 

1 In some forms of the tristubh, however, there is a restriction in the final 
syllaba anceps of the first and third padas, not found in the second and 
fourth padas. In such cases (discussed hereafter) the tristubh, like the ploka, 
consists of two parts (hemistichs) and the perfect independence of the pada is 
modified. This does not affect the free epic tristubh. 



amitranam bhayakaro mitranam abhayamkarah 
galabha iva kedaram maqaka iva pavakam 
na ’tantrl vidyate vlna na ’cakro vidyate rathah 
rukmapunkhair ajihmagrai rukmapunkhair ajihma- 
gaih (G. vi, 20, 26 and 19, 68) 1 
kim nu me syad idam krtva kim nu me syad akur- 

yato dharmas tatah Krsno yatah Krsnas tato jayah 
pagyan grnvan sprcan jighrann acnan gaccban svapan 

japate japyate cai Va tapate tapj’ate punah 

The final syllaba anceps of all padas indicates, however, 
that the gloka, like the tristubh, originally permitted the 
same metrical fall in both padas, and such we know to have 
been the case in the older metre from which the §loka derives. 
The Mahabharata retains this identical measure here and there, 
as in 

tad vai deva upasate tasraat suryo virajate, 

but such cases, usually reflecting or imitating the older verse 
of the Upanishads, as in this example, v, 46, 1, are regularly 
avoided, even by the substitution of irregular or dialectic 
forms. Thus in viii, 84, 12, where the same verb is employed, 

Duryodhanam upasante parivarya samantatah 

The gloka verse (hemistich) does not often indicate its 
unity by its form. Generally its prior half, or the pada (to re- 
tain this word for the division of eight syllables), is not united 
with the posterior pada. Verses that do unite the two usu- 
ally give lists of objects, which is the ordinary case in the 
early epic, though the later epic does not hesitate to make 
freer use of this unit-verse. But on the whole, though com- 
mon enough in post-epical writing, this is by no means typical 
of the epic itself. The great bulk of the poem does indeed 
furnish a goodly number of examples, but relatively speaking 
cases like the following are rare : 

1 The other verses are found in R. vii, 36, 22 ; 7, 3 ; ii, 39, 29 ; M. iii, 62, 10 ; 
vi, 23, 28 ; 29, 8 ; xiii, 14, 159. 



mahamaniqilapattabaddhaparyantavedikam, ii, 3, 32 
aikyasamyogananatvasamavayavigaradah, ii, 5, 3 
vayarii hi devagandharvamanusyoragaraksasan, iii, 53, 29 
jambvamralodhrakhadirasalavetrasamakulam, ib. 64, 4 
gihhagardulamatamgavaraharksamrgayutam, ib. 39 
badarengudakaqmaryaplaksagvatthabibhltakaih, ix, 37, 61 
gadamusalanaracaqaktitomarahastaya, ix, 46, 66 
drcyate hi dharmarupena ’dharmam prakrtaq caran, 
xii, 261, 6 

ajayata maharajavange sa ca mahadyutih, xiii, 10, 35 
sa bhavan dandasamyogena ’nena hrtakilbisah, G. iv, 17, 58 
bhavadbhir niqcayas tattvavijnanakugalair mama, G. iv, 
32, 5. 1 

The hemistich of the gloka is also generally independent of 
the rest of the stanza in sense as well as in metre, but it is not 
infrequently united with it syntactically, as in vi, 19, 12, 

na hi so ‘sti pumahl loke yah samkruddham 

drastum atyugrakarmanam visaheta nararsabham . 

Not a mortal on earth exists, who deep-incensed 

Mighty, a chief of awful strength, could a mo- 
ment behold in war. 

So samalarhkrtam : gatam, in the first chapter of Nala, 11; 
krodhasya ca vinigrahah : karyah, xii, 330, 10 ; asambhavyam 
vadham tasya Yrtrasya vibudhadhipah : cintayano jagama ’gu, 
R. vii, 85, 15, etc. Inside the hemistich, the padas are fre- 
quently euphonically independent (hiatus) ; 

Prajangho Valiputraya abhidudrava, R. vi, 76, 22. 
na kimcid abhidhatavya aham, R. vi, 118, 10 
ma vinagaih gamisyama 2 aprasadya ’diteh sutam, 

R. vii, 35, 63 

1 E. (Bombay) has cssura between padas and avoids both these forms 
(sariiyogat in 18, 64, for samyogena, etc.). 

2 G. here, 38, 113, has the future imperative, gamisyadhvam. Other ex- 
amples of hiatus may be seen in R. v, 60, 8; vi, 60, 8; vii, 11, 42 etc. 
besides the ample collection of Bohtlingk for the first four books. 



Saumitram samparisvajya idam vacanam abravlt, 

R. vi, 23, 1 

nihanyad antaram labdhva uluko vayasan iva, R. vi, 

17, 19 

^aranany a 9 aranyani aqramani krtani nah, R. vii, 6, 5 

In G. the hiatus is usually avoided, but it is sometimes kept 
here, as where R. vii, 21, 19 has gorasam gopradataro annam 
cai ’va (adrakslt) and G. rectifies the grammar but keeps the 
hiatus, gopradatrnq ca annam. 1 In the last book of the poem, 
hiatus in G. is more common than in the earlier epic; for 
example, G. has the hiatus of R. vii, 6, 40, svadhltaih dattam 
istarii ca aigvaryam paripalitam. On the other hand, within 
the pada attempts are sometimes made to avoid hiatus at the 
expense of form, as in R. vii, 109, 4, brahmam (cf. 88, 20) 
avartayan param. Contrast is often the cause of hiatus, both 
in the pada, as in apayaih va upayaih va, R. iii, 40, 8, and in the 
hemistich, as in hlnam mam manyase kena ahlnarh sarvavikra- 
maih, R. vi, 36, 5. 2 So in the Mahabharata, satyanama bhava 
’§oka, agokah gokana<janah, iii, 64, 107. The latter epic 
otherwise presents the same phenomena: 

yesam mutram upaghraya api bandhya prasuyate, 
iv, 10, 14 

upavartasva tad brahma antaratmani viqrutam, v, 

43, 69 

viveqa Gangarn Kauravya Ulupl, xvii, 1, 27 

deva ’pi marge muhyanti apadasya padaisinah, xii, 


anahutah praviqati aprsto bahu bhasate, v, 33, 36, etc. 

There is nothing peculiarly epic in hiatus. It is found in 
precedent and subsequent poetry. Its occurrence in the 

1 R. in the second hemistich has grhahs ca grhadatarah (acc.) svakarma- 
phalam afnatah, ag for bhuj, as in M. iii, 32, 6. 

2 Emphasis also may cause hiatus, as in dharmatma iti, R. i, 21, 7 ; na tu 
vaktum samartho 'ham tvayi atmagatan gunan, R. ir, 8, 5 ; or it may be em- 
ployed to save the life of a word, as in daksinarthe 'tha rtvigbhyah, xiii, 93, 25 
(the commonest hiatus is this before r, as in sarve ca rtavah ; karayasva rse; 
anye rksavatah, etc.). 



Mahabhasya, as in gayana vardhate durva aslnam vardhate 
visam, IS., xiii, p. 461, may be epic. 

The cadence of the gloka, like that of all other poetry, de- 
pends on the sense, and the caesura cannot be determined by 
rule. In most cases there is a caesura at the end of the pada, 
but it is frequently shifted, as in kva ’rjunah nrpatih ? gighram 
samyag akhyatum arhatha, R. vii, 31, 11. A complete sen- 
tence seldom exceeds the limit of a stanza, and when it goes fur- 
ther it may be set down as a mark of lateness. Quite anomalous 
in epic style are those long sentences, usually relative, which, 
as in Gita 2, 42-44 and 6, 20—23 run through twelve or four- 
teen padas. Still more awkward are the sentences found in the 
later epic. Thus in xii, 302, occurs a sentence, not of four- 
teen padas as in the Gita, but of fourteen glokas (5-17) : yet 
this is surpassed in the same section by a sentence of thirty 
glokas, which even then has no finite verb and in reality never 
comes to an end at all (24-52). Such monstrosities, however, 
belong only to the pseudo-epic. 

Like the gloka, the tristubh, in euphony and sense, may be 
a couplet, the first two and last two padas making a unit, as 
in iii, 118, 20 c — d, anyahg ca Vrsnln upagamya pujam : cakre ; 
vii, 2, 33 a — b, na tv eva ’ham na gamisyami tesam : madhye 
guranam. Euphonic unity is illustrated by the elision in vii, 
163, 14 of a in adrgyanta at the beginning of the pada after o • 
by tang capy : upopavistan between c — d in i, 191, 19 and 
by the complete hemistichs: 

yada ’crausam Bhimasena ’nuyatena ’gvattliamna para- 
inastram prayuktam, i, 1, 213 
Sunltha-Vakrah, i, 187, 15 (compare in gloka; 
danan, viii, 20, 3; bahugo Vidura-Drona-Krpa- 
Gaiigeya-Smjayaih, ix, 61, 20) 
antam, iv, 64, 27. 

Ordinarily, however, disjunction and not conjunction of 
padas is the rule. Thus between b — c, iii, 132, 5, a + a, and 



even between a — b and c — d. Here also hiatus appears even 
in the pada, as in i, 1, 214 b, svasti ’ty uktva astram astrena 
gantam (so must be read) ; or in i, 74, 30 e, ahag ea ratrig ca 
ubhe ca samdhye. It may then be expected between padas, 
as in 

yada ’vamansthah 1 sadigah greyasag ca, alplyasag 
ca, i, 88, 3 a — b 

vanaspatln osadhlg ca ’viganti, apo (= apo) vayum, 
i, 90, 11 a — b 

santi loka bahavas te narendra, apy ekai ’kah, i, 92, 

13 a — b 

So in Yajnaseni : ekambara, ii, 67, 34 a — b ; utsaliami : ayus- 
man, iii, 192, 67 c — d; putri : Iksvaku, ib. 70 c — d ; tapag 
ca : amatsuryam, v, 43, 20 a — b ; acaryena : atmakrtam (text 
-nat), v, 44, 14 a ; apo 'tha adbhyah salilasya inadhye, v, 46, 
3 a. B. occasionally rejects (betters) the text of C., as in vi, 
129 c — d, stands na ca ’pi te madvagaga maharse, 'nugraham 
kartum arha lii me matih, where B. 3, 61, has na ca ’dharmam, 
etc. So in viii, 4,340, pagcad vadliisye tvam api, sampramudha, 
aham, etc., where B., 85, 33, lias inudham. Both, however, 
continue with aham hanisye 'rjuna ajhnadhye, and in the next 
verse both have prasahya asyai ’va in c — d. 2 Other cases 
are: gatruhanta: uvaca, viii, 85, 30 c — d ; mudarn ca lebhe 
rsabliah Kurunam, ix, 17, 18 d; uttana-asye na havir juhoti, 
xii, 246, 27 a ; bibheti : agraddheyam, xiv, 9, 27 c — d ; Madam 
nama asuram vigvarupam, xiv, 9, 33, c (from the text in B., 
namasuram, and in C. 251, Madam namanam) ; Tilottama ca 
7 py atha Menaka ca: etas, H. 2, 89, 71 a — b. Examples from 
the Riimayana are given by Bohtlingk, or may be seen in the 
conjunction of maharathasya : Iksvaku, R. vi, 14, 12 a — b; 
abhyupetya : uvaca, R. vi, 59, 45 c — d. In both metres, to 

1 The first foot consists of five syllables. 

2 B.’s reading in iii, 112, lo d, caliteva ca ’sit for caliteva asit, 10,065, 
may be to avoid hiatus. In ii, 63, 6 d = 2,110, both texts have acintito 
'bhimatah svabandhuna, where hiatus may be assumed, though not neces- 
sarily, as also in iii, 197, 13 b, na (vai) vasan'i pitaro (a)sya kurvata. Ib. 15 
a — b, both texts have hiatus, uksanara paktva saha odanena asmat kapotat 
prati te nayantu (give you for). 



avoid hiatus, irrational particles are often inserted. A good 
example is: pura krtayuge tata hy asld raja hy Akampanah, 
vii, 2,029, where B., 52, 26, omits the first hi. 


Connection of padas by rhyme is not uncommon. It is less 
noticeable in glokas than in tristubhs on account of the alter- 
nate trochaic and iambic cadence employed in the former, and 
some, for example, may think that in iii, 65, 65-66, 

vasasva mayi kalyant 
prltir me parama tvayi . . . 
ihai ’va vasatl bhadre 
bhartaram upalapsyase 

the rhymes of the nameless queen are practically unfelt, 1 but 
this is scarcely possible when alternate rhymes occur, as in 
R. ii, 88, 7 : 

gifcavatsu sugandAisw 

In gl. 13 of the same section, three successive padas end in 
-am ; in 14, two end in -a ; and in 23-25 seven end in -am, 
or -am, with some inserted besides : 

ayantritahayad vi/>a w 
anavrtapurad varum 
rajadhanlm arakshuut 
aprahrstabahtju nyu nark 
visamasfAtm anavrftcm 

So in tristubhs, rhymes are both irregular and reg ular , as in 
R. iv, 24, 13, 

1 Compare, however, the affected initial assonance (with the same differ- 
ence) in R. iv, 33, 62 : 

Taraya ca ’py anujnatas 
tvaraya va ’pi codituh 



acintaruyawi. parivarjanJya»i ampsawtyam svana- 
veksa inyam 

and in R. vi, 73, 55, where three padas end in -dhani, -bhani, 
-kani, respectively; the same (in -tdni, -jani, -riant) appearing 
also in a puspitagra stanza, R. v, 20, 36. In R. iv, 28, 41, we 

caranti ruparjunavaskani 

gajah smsanydni vananta rani 
dhruvam parisvajya saroru hani 
kadambapuspawi sakeqartmi 
na vani hrsta bhramarah pibanti 

In the following passage the effect of rhyme is given by simple 
repetition of the whole word, R. iv, 28, 25 (not in G.) : 

nidra canaih keqavam abhyupaiti 
drutam nadl sagaram abhyupaiti 
hrsta balaka ghanam abhyupaiti 
kanta sakama priyam abhyupaiti 

words put into the mouth of love-sick Rama (kamapradhanah, 
as he is called) by some late poetaster, who, not content noth 
the last stanza, adds to it (27) : 

vahanti varsanti nadanti bhanti 
dhyayanti nrtyanti samacvasanti 

Compare also in the same section, weak rhymes in -tanam, 
-vanam, -kanam, -ranam (at the end of the pada in 31). This 
reaches its height in the ridiculous (late) section R. v, 5, 
where the same word is repeated at the end of each pada 
till even 6 is a relief, where occurs the alternation : -panko, 
-paiikah, -lanko, -^aukah. But elsewhere in R., e. g., ii, 16, 
47, three padas of a tristubh end in -am, the other in -ani(d) ; 
and in the preceding stanza three padas end in -aih-, though 
jagatl padas are here interchanged with tristubh. 

Foot may rhyme with foot or with alternate foot in the 



gloka, just as pada rhymes with pada, that is, either with a 
modification of the precedent syllable, thus, x, 15, 84, 

evaih kuru 

na ca ’nya tu 

or even with alternate rhyme, as in R. v, 59, 24, 


ca suqronl 
ca Janaki 

but the same sound may also be repeated without any such 
precedent difference, as in x, 15, 14, 

adkarmac ca 
krto 'nena 

Such light fundamental rhymes cannot be said to be pro- 
duced without design. They are, in fact, the vulgar rhyme 
of the common proverb, such as is conspicuous in all popular 
sayings. Compare for instance the following Marathi 
proverbs : 

(a) icchi para 

yei ghara 

(b) jyatse kude 

tyatse pudhe 

(c) svarga lokl 


(d) zase zhada 

tase phala 1 


Alliteration, according to the native rhetorician Danclin, is 
affected rather by the Gaudas than by the Vidarblias, the 

1 (a) what is wished for another will come to one’s own house; (b) evil is 
in front of an evil man (honi soit qui mal y pense) ; (c) in heaven the river 
Vaitaranl (the river of death precedes the joy of heaven) ; (d) as is the tree, 
so the fruit. Manwaring, Marathi Proverbs. The earlier anustubh shows the 
rhyme better on account of the iambus in the prior pada, e. g., RV. v, 86, 5 : 
arhanta cit puro dadhe 
ahfeva devav arvate. 



latter preferring cognate sounds to mere repetition. The ref- 
erence is rather to classical affectations than to epic style, 
where alliteration is a common trick, but is not so overdone 
as it is in the works of later poets. A great deal of it is 
probably unconscious, or at least required and almost unavoid- 
able. Still, the later epic writers certainly affect the anuprasa 
which Dandin says is not liked by the Yidarbhas. Thus in 
vii, 118, 16, 

muda saruetah paraya mahatma 

raraja raj an surarajakalpah 

and in viii, 94, 54, 

nihatya Karnaiii ripum ahave Vjunah 
raraja rajan paramena varcasa 
yatha pura vrtravadlie qatakratuh 
So in ix, 35, 24, 

dcqc, tu rfeyani rfanani vividkani ca 
and in iii, 63 21, 

y'ayraha ’y'ayaro y rah ah 

or iii, 64, 118, 

/fca ’si fcasya ’si fcalyani, /cim va, etc. 

Cf. iv, 14, 12, 

ka tvaih kasya ’si kalyani, kuto va, etc. 

or iii, 64, 99, 


The taste for jingling is clearly seen hi such examples from 
both epics as the following : 

Taro 'bravlt tatas tatra, G. v, 1, 49 
cayanam qayane eubhe, R. v, 10, 50 
pralmamlnamakaram, vii. 146, 3 
Kurwprestha Kuruksetre kurusva mahatlrii kriyam, 
ix, 37, 57. 

Alliteration is sometimes built on a foundation of older 
phrase, such as bhlmo bhlmaparakramah, Ramo ramayatam 
varah. Thus in R. vii, 42, 22-23, 



mano 'bhirama ramas ta 
Kamo rarnayatam varak 
ramayamasa dharmatma 

A good deal of this is due to the later revisors. Thus R. 
v, 56, 51 (also a pun in sa lilam), not in G., 

sa lilanghayisur bhlmam salilam lavanarnavam 
kallolasphalavelantam utpapata nabho harih 

As it is quite impossible to tell what proportion of such 
verses reverts to the original epic, it must suffice to show that 
epic poetry as we have it, while not attaining to the perfected 
abominations of classical works, nevertheless employs alliter- 
ation to portray situations. Thus the raudrarasa in R. vi, 
65, 41, 

raudrah cakatacakrakso mahaparvatasarimibhah 

where the “ harsh thunder-sound ” is well given by §akataca- 
krakso. Admirable, too, is the phonetic imitation of motion, 
stumbling, falling, and dying in Mbh. vii, 146, 86 : 

babhramuq caskhaluh petuh, sedur mamliif ca, Bharata 

The rhapsode’s clay is moulded variously, but it is the same 
stuff, the last example being a studied improvement, to suit 
the situation, of viii, 19, 2: 

vicelur babhramur necpih petur marnluc ca, Bharata, 

repeated in 21, 16, with varied reading, but leaving (tresuh) 
petur mamlug ca (sainikah), and varied in 19, 15 with the 
fatal marisa of the later poets (here in place of Bharata). 1 
The examples given above show both the Northern and the 
Southern style used in both epics. 

That Valmiki was copied by his successors goes without 
saying. The pseudo-Ramayana shows, e. g., vii, 32, 64: 

1 One of the signs that the completed Maliabharata is posterior to the 
Ramayana. Compare A. J. Phil., to!, six, p. 142. It is a Buddhistic term, 
mariso, foreign to the Ramayana but current in the Mahabharata and later 
Sanskrit works. The word, be it noted, is as old as one pleases, but its stereo- 
typed employment in the Bharata puts that whole work from a synthetic 
point of view on a par with other non-Buddhistic literature using it. 



sa tu iahusahasrena ialad grhya da^ananam 
iaJandha Salavan raja Z>alim Narayano yatha, 

and this atrocity in G. v, 32, 45 (not in B.) : 

suvarnasya suvarnasya suvarnasya ca bhavini 
Ramena prahitam devi suvarnasya ’nguriyakam, 1 

where the poetaster alliterates the whole word in an attempt 
at pathetic repetition. Though this is not in B., yet the 
latter countenances iii, 39, 18, where “ words beginning with 
R ” frighten Rama’s victim : 

ra-karadlni namani Ramatrastasya Ravana 
ratnani ca rathac cai ’va vitrasam janayanti me. 

Similes and Metaphors. Pathetic Repetition. 

On epic similes and metaphors an interesting essay remains 
to be written. As these subjects lie quite apart from a study 
of the verse itself, I shall at present make only one or two 
observations touching on the significance of these figures. 
First of all, the presence in the epic of rupakas, metaphors, of 
this or that form, no more implies acquaintance with a studied 
ars poetica than do such phenomena in other early epic 
poetry. The pseudo-epic has a disquisition on rhetoric, as 
it has on every other subject, but rhetoric is older than Rhet- 
oric, and I cannot see that illustrations of later norms found 
in the epic prove acquaintance with those norms. 

In the rewritten Gita, unquestionably one of the older poems 
in the epic, though not necessarily an old part of the epic, we 
find that the current dipo nivatasthah simile is introduced as 
a “traditional simile,” upama smrta, 6, 19. Such stock sim- 
iles belong to neither epic, but to the epic store in general, as 
may be seen by consulting the long list of identical similes 
in identical phraseology common to both epics. But the 
epics lack the more complicated figures of classic form, just 
as they lack the later complicated yamakas. What they have 

1 Compare G. iv, 42, 12 = 44, 12 (angullyam, sic, in the latter), where the 
ring is “ engraved with the mark of Kama’s name ” (as arrows are marked in 
M.). So K. v, 36, 2, (anguliyakam) Ramanamankitam. 



in abundance is (a) the simile ; (b) the simple metaphor; 
(c) the double metaphor. They have also a most atrocious 
mixture of metaphor and simile, as in R. vi, 41, 45, te tu 
vanara§ardulah §ardula iva dahstrinah, “ those ape-tigers like 
fanged tigers.” The simile is sufficiently illustrated in Ap- 
pendix A. I note only that it may be doubled, Rahur yatha 
c andram iva, “ he, like Rahu, him, as if the moon ” (overcame). 
Illustrations of the double metaphor are found, for example, in 
xiii, 107, 33, sarasvatim gopayanah, keeping silence (“herding 
fluency ”) ; xiv, 90, 95, svargargalam lobhabljam, “ heaven’s 
bar has greed as its seed ! ” 

For my present purpose it is necessary only to point out 
that the later part of the epic exceeds the earlier epic in 
involved metaphor. Nothing, for example, in the early epic 
is quite equal to xiii, 107, 26, where after mentioning bil- 
lions, sagara, in 21, the poet adds: 

avartanani catvari tada padmanl dvadacja 
Qaragniparimanam ca tatra ’sau vasate sukham, 

which means that one remains in bliss fifty-one padmas of 
years, sixteen plus the aggregate of the (five) arrows (of 
Love) into the (seven) 1 flames = 35 (+16). 2 But parallels 
almost as extravagant (including the go-pay simile above) have 
been noticed by Professor Lanman in the interesting essay 
referred to in the last note. Not so striking, though in style 
more rhetorical than is found in the love-passages of the early 
epic, is the metaphor of iv, 14, 25 : 

atmapradanavarsena samgamambhodharena ca 
(jamayasva vararohe jvalantam manmathanalam, 

“0 graceful maid, quench the mind-shaker’s (Love’s) glowing 
fire with the rain of self-surrender and the water of union.” 

1 PW., s. paragni, says three fires. But compare yad agne te fivam rupam 
ye ca te sapta betayah, i, 232, 10, and saptarcis, passim : and Mund. Up. ii, 1, 8. 
Besides, the result is 35 and one multiple is 5, so the other must be 7 (flames). 

2 These high numbers, while not confined to the pseudo-epic (Ind. Streifen, 
i, p. 97 8 .), receive fresh additions there in names of numbers before un- 
known. Compare xiii, 107, 63, for example, where occur the fanku and 
pataka : tatha cankupatake dve yugantam kalpam eva ca, ayutayutam tatha 
padmam samudram ca tatha vaset. On similes, cf. Lanman, JAOS. xx, p. 16. 



Another rhetorical trick, which appears not to have been 
noticed in the epic, is the occurrence of distinct attempts at 
“pathetic repetition.” A comical example is given above. 
I have noted cases but rarely, and only from late parts of 
the great epic, but I cannot say they are not found else- 
where. 1 The first is from viii, 75, 6-7 a: 

ratharii sasutam sahayaiii ca kameit 
kagcid rathl rnrtyu vacant ninaya 
ninaya ca 'py ekagajena kagcid 

rathan bahun mrtyuvage tatha ’gvan 
rathan sasutan sahayan gajang ca 
sarvan arln mrtyuvagam garaughaih 

Another is found, H. 3, 118, 9 = 15,776 : 

adraksam adraksam 2 ahaiii sunirvrtah 
piban pibahs tasya vapuh punah punah, 

(B. has puratanam) 

and in the next stanza : 

samsmrtya samsmrtya tam eva nirvrtah. 

Tins differs from simple repetition, such as that of janami 
in R. iv, 33, 53 ff., but only in the effect aimed at. Per- 
haps the yada ’grausam passage may be included. 

Cadence in pioka and Tristubh. 

The gibberish of xii, 10,399 (v. 1. in 285, 125), 

hayi hayi huva hoyi huva hoyi tatha ’sakrt 

is interesting as showing the epic’s recognition of this form 
of interjectional piety (gayanti tvaih suragrestha samaga 
brahmavadinah) ; 3 but I introduce it here as illustrating the 

1 Without the attempted pathos, mere repetition is an ancient trait ex- 
hibited as early as the Rig Veda, as pointed out, e. g., by Weber, Vedische 
Beitrage, 1900, p. 7, on BV, ii, 11. Repetition of the same words in succeed- 
ing stanzas is perhaps best illustrated by B. ii. 28, where duhkham ato yanam 
is the pathetic refrain. 

2 Compare RV. i, 25, 18, daryam . . . daryam. 

8 Compare the stobha ib. 105: hun huh hunkaraparaya, etc. 



common occurrence of the repetition of the final foot of the 
prior at the beginning of the posterior pada. This extreme 
example duplicates even the syllables, but in the pathya form 
of the §loka the duplication of the whole foot, while not re- 
producing the syllables, may extend backward as well as for- 
ward, thus giving three identical feet, as in R. vii, 28, 6, 

na bhetavyam na gantavyam nivartadhvam rape 

Such a verse, however, is often modified as in iii, 168, 80 : 
nibodhata mahabhagah (Jivam ea ’qasta me 'naghah, 

or, if the first two are maintained, by making the third foot 

_ w—w or . The different possibilities concern us 

here only as they affect the cadence, for the monotony of the 
pada is varied quite as much by the rhetorical cadence as by 
the foot. Even the stereotyped diiambic close of the posterior 
pada is constantly broken by a choice of words which, far from 
lending themselves to iambic rhythm, impede it. So instead 

of the posterior m w, w _ ^ _ the pada must often be 

read as ^ 9 9 W 9 while in the prior pada ^ v _ w , 

w m is frequently to be read as m w, _ w w _ v. Pro- 

nounced cretics and dactyls often claim recognition, as at R. 
vi, 17, 12, 

Ravano, nama, durvrtto, raksaso, raksaseqvarah, 1 

or ib. 17, 67, vidyate tasya samgrahah ; ib. 18, 7, iti ho ’vaca 
Kakutstho vakyam, satyaparakramah. Hence even in the 
more rigid posterior pada the cloka presents great variety. 
The effect, for example, of the diiambic ending is quite lost 
in the following typical examples: 

balad adaya, viryavan 
nava, panca ca, sapta ca 
sandhim Ramena, Havana 

To read such padas mechanically, as if they had a pause 
before the diiamb (as Occidental scholars almost always read 

1 A stock phrase, the parallel to Ravano lokaravanah, R. vi, 20, 21, etc. 



them), is vicious. The gloka, more than any other metre, 
must be read by sense rather than by scheme. The latter 
method is bad enough in all metres, but peculiarly so in the 
short §loka, where, unless the stress jibes with the words, 
the result is a peculiarly painful tum-tum, which in no way 
gives the rhythm,- for in reality the ^loka is a metre of 
great subtlety and force, in which neither iambic nor tro- 
chaic cadence has ever held sway, but both interchange 
with pleasing variety even in pathyas, 1 often uniting in a 
dactylic or choriambic measure, as in iii, 56, 24, 

kim abravie ca nab sarvan, 
vada, bhumipate, 'nagha 

or R. vi, 65, 11, 

gaccha qatruvadh&ya tvam, 

Kumbhakarnajayaya ca 

or ib. 59, 47, 

tam abravln mahateja 

Ramah, satyaparakramah, 
gaccha, yatnaparac ca ’pi 

bkava, Laksmana, samyuge 

With the same freedom at the outset, the tristubh, instead 
of embracing all forms, as it might have done, continued on 
a more and more restricted path. It kept the iambic cadence 
much more closely than did the 9loka and contracted its 
middle to an almost unvarying shape. It thus grew more 
and more monotonous, and not having even the advantage 
of hemistich-unity it became a mere collocation of hen- 
dekasyllabic verses, each pada having the same unvarying 
quantity : 

W \J \J \J \J 

1 Still greater variety is given by the melodious vipulas, of which I shall 
speak below. But seven-eighths of epic verse are in pathya form, that is, 

half the syllables in the verse are unalterably fixed as w and y_, yo, 

eo that it is of interest to see how with this self-imposed restriction the Hindu 
poet still manages to make verses so melodious, energetic, and varied, when 
read properly. 




(called upajati), as in Horace’s 

trahuntque siccas machinae carinas. 1 

The only way to save from dead uniformity a rhythm so 
stereotyped was to shift the cmsura frequently. 2 In the 
Ramayana, where upajatis are the rule (the Mahabharata 
tristubh did not reach the same level of monotony), there is 
often a constant play from fourth to fifth or a remoter syl- 
lable, as the place of rest. With the usual pause at the 
fifth, the dactylic middle foot is converted into an ana- 
paestic iambic slide, as in the following examples from R. 
iv, 43, 62; 44, 16; v, 32, 10, the last two examples showing 
also the lighter caesura not of sense-pause but of breathing : 

(a) tatah krtarthah 

sahitah sabandhava 
maya ’rcitah 

sarvagunair manoramaih 
carisyatho ’rvlm 



bhutadharah plavaiiigamah 

(b) sa tat prakarsan 

harinam mahad balam 
babhuva vlrah 

pavanatmajah kapih 

vyomni viquddhamandalah 
qaql ’va naksatraganopaqobhitah 

(c) svapno hi na ’yam 

na hi me 'sti nidra 

1 Brown’s Sanskrit Prosody, p. 9. On the other hand the jagatl corre- 
sponds in outer form to the iambic trimeter with twelve syllables. I treat 
the jagati throughout as a tristubh with one syllable added (the final syllaba 

anceps of the former becoming fixed as brevis), \j kj ; not assuming this 

as a genetic fact but as a convenience, the same body appearing in both and 
the padas being interchangeable except in the aksaracehandas. 

2 On the derivation of types fixed in respect of the initial syllable {the 
upendra and indravajra being derived from the upajati and not vice versa), 
see below, the section on the Stanza. 



qokena duhkhena ca 
sukhaiii hi me 

na ’sti yato vihxna 
tene ’ndupurnapratimananena 

But this tendency ran to extremes also, and as the syllabic 
arrangement became fixed, so the caesura became stereotyped, 
till stanzas showed an almost unvarying caesura of the painful 
type of R. v, 47, 30, 

iti pravegaiii tu 

parasya tarkayan 
svakarmayogam ca 
vidhaya vlryavan 
cakara vegarii tu 
mahabalas tada 
matiiii ca cakre 'sya 
vadhe tadamm 

or of R. vi, 126, 55, 

tatah sa vakyair 

madhurair Ilanumato 
nigamya hrsto 

Bharatah krtafijalih 
uvaca vanlm 

manasah praharsinim 
cirasya purnah 

khalu me manorathah 

Even if Vahnlki did not write these stanzas, which may be 
doubted, a greater poet than he is guilty of the same sleepy 
iteration of cadences, as may be seen in Raghuvai^a iii, 30 ; 
v, 18; vii, 19 (caesura after the fifth in all padas) ; vii, 16 
(after the fourth in all padas). 


Alternation of tristubh and jagati padas in the same stanza 
helped somewhat to mitigate the weary effect of this metre ; 
but it gradually yielded before the yloka or passed into other 



forms. One of its decadent uses was to furnish new tags for 
the end of chapters of glokas. This was an old use, but it 
is extended in the later epic. The different texts show no 
uniformity in the insertion of these tag-tristubhs, one text 
having several, where another has one or none, just as in 
the case of other tag-metres, for example, a puspitagra, G. iii, 
39, 42 ; two ruciras between G. iii, 56 and 57, but none in R. 
P lainl y a late insertion, for instance, is the imitation-stanza 
which serves as a tag to G. iii, 43, 42 (not in R.), 

kalasya kalaq ca bhavet sa Rainah 
saiiiksipya lokahq ca srjed atlia ’nyan, 

Mann, ix, 315 ; Mbh. ix, 36, 40, 

sa hi kruddhah srjed anyan devan api mahatapah 
xiii, 152, 16, 

adaivaih daivatam kuryur, daivataih ca ’py adaivatam 
lokan anyan srjeyus te 

Such tags may, in fact, be made of adjacent glokas. An 
instance is given below where a rucira has thus been created. 
As regards tristubhs, G. iii, 62 ends with a tag made out of 
a gloka omitted in this text but kept in the other, na garma 
labhate bhiruh and na vindate tatra tu garma Maithili. A 
good example is found in R. vii, 75, 18 ff., where a tristubh 
tag is added in almost the same words with those wherewith 
the following chapter begins, showing that with the division 
into two chapters a tag was simply manufactured out of the 
next stanza ; as is still more clearly indicated by the fact that 
76, 2 answers the question of 75, 18, vaiovas trtiyo varno va 
gudro va (’si)? gudrayonyam prajato 'smi. Evidently only 
one verse intervened, the gloka: tasya tad vacanam grutva 
avakgirah . . . uvaca ha. 1 

1 The same thing occurs in R. iv, 50, where the chapter closes with the 
gloka : papraccha Hanumaris tatra ka ’si tram kasya va bilam. Then fol- 
lows the tag : tato Hanuman girisamnikagah krtanjalis tam abliivadya vr- 
ddham, papraccha ka tvam bhavanam bilarii ca ratnani ce ’mani vadasva 
kasya, simply repeating the last gloka in tristubh form. G. very properly 
drops the gloka ; but it is clear that originally the gloka closed the question. 



The tag-function of the tristubh is also known in the Maha- 
bharata, notably in the one tristubh found in the Nala, iii, 76, 
53, which has been regarded as spurious on account of its 
isolation. But the following sections, after the Nala episode, 
show just the same conditions, the end of chapters 83 and 
100. So, too, at the end of ix, 24 and 28. Hariv. 2, 66, and 
69 end with one jagati each ; 2, 68, with three. 

The present text of the Ramayana shows many cases of 
tristubhs and jagatls interpolated into the middle of a gloka 
section. Some of these at least are clearly the finale of former 
chapters. Thus R. vi, 69, 15 looks like an inserted jagati, but 
its function is to close the chapter in G. 48, 13. So R. vi, 69, 
88-96 appear as a group of interpolated tristubhs ; but in G. 
the same group is a tag to chapter 49. Probably the break 
in R. vi, 69, 44, G. 49, 31, is the original finis of a chapter. 
Occasionally, when one edition breaks a chapter, only the new 
division is found to have tristubh or jagati, as an accepted 
sign of conclusion, as in R. iii, 11, after 70; G. 16, 41. 

A special function of the later tristubh is to produce pathetic 
effect. 1 In this guise it wins new life and makes whole chap- 
ters, as in R. v, 28, where the burden of the chapter is ex- 
pressed by ha Rama ha Laksmana ha Sumitre, etc. ; or in R. 
iv, 24 (not in G.), a lament, the dolorous style of which may 
be illustrated by the reminiscent verses, 13-14: 

prapto 'smi papmanam idaiii vayasya 

bhratur vadhat Tvastravadhad ive ’ndrah 
papmanam Indrasya mail! jalaiii ca 

vrksacj ca kamaiii jagrhuh striyac; ca, etc. 

Closely allied is the employment of the tristubh to describe 
not mental conditions but operations of nature. The Yedic 
pra vata vanti patayanti vidyutah, RV., v, 83, 4, appears in 

1 This begins in the Mahabharata as an extension of the tag-fuDCtion. 
Compare the illustrations given in A. J. Phil., vol. xix, p. 18 ff. A good ex- 
ample of the sentimental effect, intensification of horrors, etc., deputed to the 
tristubh by predilection, is found in R., v, 54, 30 ff. The action is in ylokas. 
The moral effect is given by the following tristubhs. 



R. iv, 28, 45 as varsapravega vipulah patanti pra vanti vatah 
samudirnavegah, in a long section -wholly descriptive. Another 
example is found in R. iv, 30, 28-57. 

yiokas and tristubhs are not often commingled, save in a 
few late passages of the great epic, i, 232, 10 if. ; Hariv. 3, 
82, 3 ff. ; and in R. v, 41 ; G. 37 (chiefly upendras), through- 
out a section. In R. a few long passages occur in the sixth 
book, 59-61, 67, but apart from these books the exchange 
of the two metres is avoided. 1 In the Sanatsujatlya, v, 46, 
there is, indeed, a regular gloka refrain besides other glokas 
intermingled with tristubhs, but this is because the author 
is reducing ITpanishad stanzas, and at the same time adding 
some of his own. The practice belongs to those scriptures, 
and is not generally kept up in the epic, though occasion- 
ally a §loka or two appears among tristubhs, as in ii, 64, 9-10. 
In xii, 350, 49 ff., two tristubhs (the second having three 
hemistichs) are inserted between Qlokas (after a ijloka of 
three verses). 

Common Forms of Cloka and Tristubh. 

From a mechanical point of view', the prior pada of the §loka 
and the tristubh are identical, except for the fact that to the 
eight syllables of the 9loka pada the tristubh appends a scolius 
or amphibrach. The natural division of the eight syllables in 
each case is into groups of four or five, followed respectively 
by four or three. For convenience the group of four, which is 
found oftenest, is usually called a foot, and to have a name I 
shall so designate it. Now' in epic (Mahabharata) poetry, every 
foot of the cjloka pada is found in the tristubh, and, vice versa 
(as will be seen from the follow-ing table), every prior foot and 
every last foot of the tristubh’ s eight syllables is also a corre- 
sponding Qloka foot : 

1 G. ii, 110, 3 fl. is not in R., and appears to be an interpolation. The par- 
allels to G. v, 89 are also lacking in R. Verse 7 in G. vi, 34, is praksipta (the 
passage is not in R., but compare R. iv, S3, 53). 



Priob foot of £loka 


1, w 

2, w 

3, w w 

4, w w 

5» v w w 

6, w w w 

7, w w 

8 , KJ \J 

Last foot of (Jloka 
and Tristubh. 

1, w 

2, w w w 

3, — w w — 

4 , 

5, _w 

6, w w 

7, w_ 

8, w w 

But, curious as is this purely mechanical identity, it is subject 
to three limitations, which prevent the effect one might think 
would be caused by it. First, the tristubh ’s eighth syllable is 
long, while in the <jloka, since the pada ends here, the same 
syllable is anceps. Second, the scolius of the tristubh is usu- 
ally closely united with the second foot, while in the case of 
upajatis and some other tristubhs the ctesura occurs in a 
majority of cases after the fifth syllable, so that the feet are 
not in reality what they are in the measured division given 

above ; but the pada appears, for example, as w _ w , 

w w _ w _ w, whereas in the Qloka the usual caesura is after 
the fourth, and only in certain cases falls after the fifth sylla- 
ble. But the third difference, that of the general effect given 
by the 9loka cadence and that of the corresponding syllables 
in the tristubh, is produced by the interrelation of the first 
and second foot. Here there is a wide divergence, and it is 
the preference for one combination over another that makes 
the greatest difference between the form of the gloka as a 
whole and the tristubh as a whole. Although it is true, as 
has been remarked by Professor Jacobi, that the essential 
difference in metres lies not in the opening but in the close of 
the pada, yet in this case the interrelation just referred to is 
almost as important. Thus, to take a striking example, while 

w is a second foot both in §loka and tristubh, in the 

former it is pathya, “ regular,” in all combinations, the com- 
monest of all, while in tristubh it is a rarity in any combina- 
tion. So w w occurs after four or five forms of the first 

foot in gloka, yet is never a favorite, in tristubhs after six 
forms, and is here everywhere common. 



It is, however, interesting to see in how many cases a per- 
missible form of both metres is used, so that one cannot tell 
which metre one is reading till the pada is nearly complete. 
Ordinarily the general rhythm determines the anticipation 
and the expected metre is duly met ; but not infrequently is 
the justified anticipation deceived, and the metre, still keeping 
on the lines of the preceding form, suddenly changes. A 
penultimate verse, for example, in R. ii, 38, 14, begins maya 
vihlnam varada prapannam, but we no sooner learn that this 
is a gloka verse, not a tristubh tag (as we might expect from 
its form and position), than in 15 we read imam mahendropa- 
majatagardhinim, the real tag of the section. 

The form just cited is the usual one in which the gloka 
coincides with the body of the tristubh. Sometimes, as in set 
phrases, the same words are used ; thus in G. ii, 18, 33, and 55, 

prasadaye tvam girasa karisye vacanam pituh 
prasadaye tvam qirasa yatavrate (tristubh) 

or in R. vi, 106, 4 and 59, 36, 

tam apatautam sahasa svanavantam mahadhvajam 
tam apatantam sahasa samiksya (tristubh) 

With the prevalent upajati caesura and almost after a 
system of upajatis (one gloka intervening), appears in R. vi, 
69, 130, sa vayusunuh kupitag eiksepa gikhararii gireh, a per- 
fect upendravajra, pada in a gloka verse. Such alien padas 
are not very common in the midst of a gloka system, 1 but 
are common in close conjunction with tristubhs, as if the 
poet either wished to trick or could not himself get the last 
metre out of his ear. Another instance like the one above is 
found in R. v, 54, 48 ff., where only a gloka hemistich inter- 
venes between a tristubh system and the tristubh-like cadence 
of the gloka : vyarajata ’ditya iva ’rcimall ; Lankam samastam 
sampidya langulagnim mahakapih, nirvapayam asa tada samu- 
dre (haripungavah). Cases where a whole gloka is interposed 

1 But compare R. v, 2, 31, anena rupena maya na yakya raksasam purl ; 
R. vi, 43, 17, yarirasamghatavahah prasusruh yonitapagah ; Nala, 3, 1, tebhyah 
pratijriaya Nalah karisya iti, Bharata ; and ib. 12, but no more cases till 6, 8. 



are not at all rare. In R. vi, 67, 99-101, 99 ends in a tristubh, 
100 is a §loka pathya, 101 begins sa Kumbhakarnasya §aran 
§arire (sapta, viryavan). Less striking is the case where only 
one pada of a gloka of choriambic form (second vipula) corre- 
sponds to the tristubh it follows, for here the former’s cadence 
is not kept up. Such a pada needs no intervening pathya, but 
may follow directly on the tristubh, as in R. vi, 67, 21-22, 

pradudruvuh samyati Kumbhakarnat 

tatas tu Nllo balavan (paryavasthapayan balam) 

When an unimportant word or a superfluous adornment, 
an unnecessary adverb or epithet, is added, it arouses a suspi- 
cion that some of the tjlokas may be reduced from an older 
form. Thus vidyunmali appears to stop a jagatl in R. vi, 
43, 41 a, 

cilaprahara ’bhihato (vidyunmali) nigacarah 
So in R. vi, 69, 138 a, 

khadgaprahara ’bhihato Hanuman (roarutatmajah) 

So, too in the verse cited above, haripuhgavah fills out the 
verse where mahakapih precedes, a sufficient subject. In 6. 
iv, 60, 2, nivedayamasa tada maharsiiii (samhatanjalih) ; in 
the other example above, sapta, viryavan ; and in the following 
example both terminals (even the accusative) are unnecessary, 
R. vi, 71, 37, 

tato ‘tikayo balavan praviqya (harivahinlm) 

vispharayamasa dhanur nanada ca (punah punah) 

And very likely, since an inspection of epic phraseology 
shows that there were many stereotyped turns of expression, 
there were phrases used first in the tristubh which were pre- 
served in a crystallized form in the general 9loka solution in 
which the epic was immersed. But to say, except in the case 
of such stereotyped phrases, whether this happened in any one 
instance, would be at best rather an idle expression of opinion. 1 

1 In sadhu sadhv iti (te) nedm; (ca) drstva fat rum (or raksah) parajitam, 
R. vi, 44, 31, G. 19, 37, a stock phrase in either form, an old tristubh, 
v , might be preserved, but a varied reading is more likely. 



Certain verbal forms lend themselves best to one cadence and 
it is not surprising, for instance, that one turn should go to 
make both gloka and tristubh (R. v, 47, 10 ; vi, 106, 14), or 
that the exact form here is elsewhere, G. vi, 89, 25 (R. has 
liayan), used as part of another tristubh, so that we find: 

pracodayamasa rathaih sa sarathih 
pracodayamasa ratham surasarathir uttamah 

pracodayamasa gitaih garais tribhih 
pracodayamasa gitaih garair hemavibkusitaih 

On this point I have only to add that a complete jagatl 
pada, as well as a tristubh pada, may thus appear in a gloka, 
as in the example above and in R. v, 57, 15 b, 

sa purayamasa kapir dico daga (samantatab) 

and that, next to the cboriambic form, the old tristubhs in 

^ w v'-/'-/ v/ — and — w — w>_/w w are most 

often incorporate in glokas, as in Nala, 4, 28, varnyamanesu ca 
maya bhavatsu; 9, 4, vyadiryate ’va lirdayaih na cai ’nam; 
and 12, 39, patatribhir bahuvidhaih samantad, etc., etc. Pro- 
fessor Jacobi has suggested that the gloka has borrowed such 
forms from the tristubh. This seems to be a reasonable sug- 
gestion, yet it should be said that the argument advanced in 
favor of it is scarcely valid. Professor Jacobi bases the deri- 
vation of the second vipula from the tristubh on the assumed 

fact that in this form of the pada “ \J almost never takes 

the place of _ w IS. vol xvii, p. 450. This statement, 
however, is based on a rather restricted area of examples. 
In the Bharata glokas, _ v w ^ is not uncommon except in 
late passages, and even there two or three cases out of 
twenty-five to thirty are not very unusual. All that we can 
say is that final brevis is much less frequent than in the 
first vipula. 



The Epic Cloka. 

The Prior Pada of the 6)loka. 

The Pathya. 

The pathya, or ordinary form of the first pada, should 
exclude sporadic cases, but including them for convenience 

we may say that the pathya foot w w is preceded by five 

kinds of feet, sporadic choriambus or proceleusmaticus ; iambic, 

w ; pyrrhic, ^ _ w w ; trochaic, w and w ^ w ; 

spondaic, ^ and ^ w The frequency of these feet 

advances in the order here given. With the exception of a 
sporadic choriambus or other wild irregularity, all these forms 
occur passim, even that with precedent iambus. This last is 
sure to be found so many times in a given number of Qlokas 
and it must therefore be marked as occurring passim rather 
than as common ; but it is far less frequent than the other 
forms, often less than half as frequent as the pyrrhic, as this 
is often only half as common as the precedent trochee. The 
relation between the trochee and spondee is from one-lialf to 
two-thirds. A curious fact in regard to the avoided iambus 
(before the iambus of the pathya, as hi the posterior pada) is 
that when used it is sometimes preferred in its double form. 
Thus in xii, B12 ff., for about two hundred hemistichs, the 
precedent spondees, trochees, pyrrhics, and iambs are (respec- 
tively) 82, 54, 29, 11; but of the 11 iambs, 10 are double 

w _ kj _ (against w _). On the other hand, in xiv, 59 ff., 

these precedents are 73, 38, 31, 20; and of the 20 iambs, only 
8 are double ; while the opening stanzas of the Gita (intro- 
duction, eh. 13) show 96, 62, 27, 14 ; but only 6 double iambs 
out of the 14. The precedent double iambus is characteristic 
also of Pali verse. It does not seem to me that any great 
weight is to be laid on this or that ratio in the use of these 
feet, since all are used by epic writers everywhere, and the 
only striking distinction as regards their employment is that 
spondees naturally (it is a matter of nice ear to a great extent) 
occur oftenest before an iamb, and iambs least often ; while 



trochees and pyrrhics lie between. But very often a double 

trochee (_ w __ w) is preferred to a spondee (_ w ). 1 As 

regards minor differences, as for example whether uu_ u or 

v w is used more frequently, I have not thought it worth 

while to gather the statistics. Only the curious preference 
in later writers for three successive iambs seemed worth notic- 
ing, as it leads to the hemistich of eight iambs sometimes 
affected by doggerel epic poets. 2 Such a combination regu- 
larly occurs only at the beginning of a prior pada, being 
tabooed in the posterior pada, though occasionally found there. 
The general (not inviolate) rule for the pathya is that any foot 

may stand before \j xt which does not make tribrach or 

anapaest after the initial syllaba anceps of the pada. The final 
syllable of the pathya is long in about two-thirds of the eases. 

More important are the facts in regard to the preference for 
certain forms combined with the vipulas, although these make 
but a small proportion of prior padas. 

The Vipulas. 

The vipulas (syllables five to eight) are four in number: 
(l)uwy,(2)_ U uu, (3) (4) _ w _ a. Only 

the third (as indicated) has an almost invariable caesura. In 
respect of the general rules for these vipulas, from an exam- 
ination of a considerable mass of material, I would state first 
that the epic (jloka generally conforms, as far as I can formu- 
late them, to the following conditions : 3 

1 The preference for — \j — yy instead of yy is illustrated below. 

Cases of double iambus before the pathya seem to me rather characteristic of 
the popular and late scholastic style than an archaic survival (the late scho- 
lastic often coincides with the popular through a common carelessness or 
ignorance). To be compared are Simons, Per Qloka im Pali, ZDMG., vol. xliv, 
p. 84 fi., and Oldenberg, ib. Iiv, p. 194. The latter seems inclined to see (with 
due caution) evidence of antiquity in the precedent iambus. I regard this 
combination rather as a sign that the writer is more careless. 

s See below for an example. 

3 Besides the articles above, see Colebrooke ; Gildermeister, ZK1I. v 260 • 
Weber, IS., vol. viii; Oldenberg, Bemerkungen zur Theorie des £loka, ZI)MG.' 
xxxv, p. 187 ; and Jacobi, IS., vol. xvii, p. 443; Pas Ramayana; and Gurupu- 
jakaumudl. Professor Jacobi’s rules given first as “ valid for the older epics ” 



1. The first vipula, w o w usually follows ^ 

id , or id w , though it is sufficient to have the pre- 

ceding syllable long (even this restriction is not always 
observed). The later style has fewer cases of the first of 
these combinations. The caesura is after the (pada’s) fourth 
or fifth syllable, sometimes after the sixth. The last syllable 
of the vipula is prevailingly long but not infrequently short, 
especially apt to be short after the diiambic opening. When 
the caesura is after the fifth syllable of the pada the last 
syllable of the vipula as a rule is long (winch would indicate 
that this caesura is later than the one after the fourth). 

2. The second vipula, usually follows m _ w _, 

though a preceding m or even m w is not a great 

rarity. Any other precedent foot is sporadic only. The 
caesura is after the fourth or fifth syllable of the pada, 
inclining to the latter place (at times twice as frequent). 
The last syllable of the vipula is sometimes short, most often 
when the caesura is after the fourth syllable of the pada, but 
is prevailingly long, especially in the later epic, where a short 
final is often rather rare (rarer than in the first vipula). 1 

3. The third vipula, id usually follows m _ w 

The caesura is very rarely after any other syllable than the 
fifth, and is seldom neglected. The last syllable is indiffer- 
ently short or long. This is the most rigid form, both in 

were modified in the later articles cited (1884, 1893, 1896). Professor Olden- 
berg’s observations give an excellent comparison of Manu’s practice with 
that of an epic passage. The statements in Colebrooke’s and Weber’s works 
mentioned above, based on the rules of native metrieists, often conform, 
through no fault save that of the metrieists, neither to epic nor to classical 
usage and historically considered are useless as regards the extant epic ploka. 
Professor Jacobi’s rules, as modified by him, though not exhaustive, are gen- 
erally quite unimpeachable and give the best (as did Gildermeister’s in his 
day) presentation of epic conditions. I follow his order in numbering the 
four vipulas, and his rules, with some revision. 

1 The age of the piece affects the quantity of the final syllable. For ex- 
ample, of the two lotus-theft versions, the prior (as is often the case) is the 

more modem (xiii, 93). Here there is no case of w w w, but fourteen cases 

of y \j (one hundred forty-nine clokas) . But in 94, in the compass of 

forty flokas, y \y occurs six times (against , four times). 



respect of caesura and of precedent foot, so that the pada is 
almost always w , 

4. The fourth vipula, _ w _ ± l , usually follows m ^ 

but in some sections is found quite as often after m 

and iiu The caesura rarely changes from the fourth 

syllable. The last syllable of the vipula is generally long. 

5. The Mahabharata has what may be called a fifth vipula, 

vj w It occurs sporadically in all parts of the epic and 

is not very uncommon, though not so current as in the 
Upanishads. This form crops up occasionally in the Pura- 
nas, but is ignored by ValmTki and later Kavis. 

These epic conditions may be condensed into one short rule 
of general usage: All vipulas are found after h _ w but 

with occasional exceptions 1 only the first vipula after s' 

and idu , and no other precedent feet are admitted be- 

fore vipulas. The caesura is free (usually after the fourth or 
fifth syllable) in the first and second vipula ; after the fifth 
in the third; after the fourth in the fourth vipula. 

The chief difference between the normal type of the epic 
pada and that of classical writers lies in the circumstance that, 
as contrasted with the facts stated above, in classical works 
there is 

1 ) almost complete absence of the fourth vipula, 

2) greater rarity of the first vipula after diiambus, 

3) greater strictness in the caesura of the third vipula, 

4) very rare exceptions in the employment of other prece- 
dent feet (e. g., the third vipula after w , Ragli. xii, 


5) almost exclusive use of long finals in first and second 
vipulas. 2 

Thus it will be seen that there is still an appreciable advance 

1 The commonest exception is found in the case of the fourth vipula. On 
an average half-a-dozen exceptions occur in the course of a thousand hemi- 
stichs, but excluding the fourth vipula only one or two exceptions, generally 
in the form , \j \j 

3 On the rarity of the fourth vipula in classical writers, see Jacobi, IS., vol. 
xvii, pp. 443. The rule for the long finals is cited by Weber, IS., vol. viii, 
p. 345 : sarvasam vipulanam caturtho varnah prayena gurur bhavati. 



to be noticed in the classical style as compared not only with 
the style of older parts of the epic but also with the normal 
epic. Fewer vipulas (especially fewer second vipulas) in 
general, avoidance of the fourth vipula, and greater strictness 
in the use of vipulas mark in some passages an advance even 
on the normal epic. 

There is no “ epic usage ” in respect of the proportion of 
vipulas to pathyas. The fact that there is considerable variety 
proves little in regard to difference of authors, since many 
conditions affect the ratio. Not only is there apt to be a 
larger number of vipulas in scenes of excitement, as Pro- 
fessor Jacobi, I think, has somewhere observed, but also a 
monotonous list develops vipulas, partly because it is apt to 
be composed of names which, as they are harder to manage, 
always receive a certain latitude of treatment, partly because 
the dulness of the subject requires the livelier effect of the 
skipping vipula. The vipula (in excess of the normal) may 
then be due to a) personal style ; b) intensity ; c) formality ; 
d) avoidance of dulness; to which must be added imitation 
or actual citation of older material. For this reason there is, 
in mere ratio of vipulas to pathyas, no especial significance, 
as may be further shown by the fact that on an average this 
ratio is about the same in the Ramayana and Bharata, though 
each poem shows great variations within itself. Thus in the 
first thousand verses (hemistichs) of the Ramayana’s third 
and fourth books respectively the vipulas are 125 and 118, 
or one-eighth. But twenty thousand hemistichs, which I 
have examined from all parts of the Bharata, give twenty- 
six hundred vipulas, or a trifle over the same ratio. I do 
not then lay much stress on the presence or absence of vip- 
ulas in an epic section unless it shows remarkable extremes. 
Thus if we compare the 1098 9loka verses of the Raghu- 
vahqa and the 1070 which make the first half of Nala, we 
find that in Nala the ratio of vipulas is one-sixth, while in 
the Raghuvanga it is one-fourteenth (184 in Nala, 76 in 
Raghuvanga). But this paucity of vipulas, though common 
to most classical writers, is not found in Magha (according 



to Professor Jacobi because be was a Westerner, loc. cit. 
p. 444), so that in itself it is no criterion of lateness. 

The number of vipulas gives the general average (of 12| 
per cent) already noticed. 1 But this ratio is sometimes almost 
halved and sometimes nearly doubled, small sections of two 
hundred verses (hemistichs) not infrequently showing from 
fourteen to forty-six non-pathya forms ; while in special cases 
even greater disproportion may be observed, some of which 
when taken into consideration along with other elements may 
still be worth noting. Thus as between the old tale, Upa- 
khyana, of Namuci, as told in ix, 43, 33 ff., and the following 
account, lianta te kathayisyami, of Skanda, in 44, 5 ff., the 
weight of probable seniority lies with the Vedic tale. Here 
there are vipulas enough to make the ratio 33^ per cent, 
instead of the average 121 p er cent; whereas in the Skanda 
tale there are only half as many. But again, the list of 
Skanda’s followers, ib. 45, 86 ff., shows fourteen vipulas in 
fifteen qlokas, as the list of Mothers in 46 shows forty-six in 
one hundred <jlokas, and the list of nations in xii, 101, 3 ff., 
has thirteen vipulas in twenty qlokas, all of these, however, 
being names and therefore exceptional. There are, on the 
other hand, good reasons, apart from vipulas, for considering 
that the conversation of Sulabha and Janaka is not an ancient 
part of the epic (bad grammar is one item), and here in nearly 
four hundred cases there are but eight vipulas, or less than 
3 per cent ; instead of the average 12| per cent. 

Not the number of vipulas per se, but the use of vipulas 
may be a determining factor. The refined classical style 
differs, however, not from the epic alone but from the 
Puranas, where obtains even greater freedom than in the 
epic, especially in the nice test of the fourth vipula. Thus, 
fifteen fourth vipulas is not a high number in a thousand 
Puranie verses, e. g., exactly this number is found in Vayu 

1 In simple narrative, with no disturbing factors, the compass ranges from 
fourteen to thirty vipulas in one hundred flokas (two hundred cases), three 
times more often above twenty than below it, and seldom exceeding thirty, 
for instance, only once in the first 4,000 cases of the ninth book. 



Parana, ch. 4-9, five hundred glokas ; and in the epic section 
of Canti from the end of the prose in 243 to the end of 351 
(13,224-13,740). The Agni Purana has as many as fifty-seven 
fourth vipulas in the same number of verses, the first twenty 
chapters, five hundred and five glokas. But if we compare 
the use of the vipulas we see at once a striking difference in 
these passages. The epic selection has fifty second vipulas 
and thirty-two third vipulas ; the Vayu selection, thirty-three 
second and fifty -one third ; the Agni selection, twenty-six 
second and fifty third; withal, despite the carelessness in the 

last, which gives four cases of the second after ^ w and 

three of the third after m That is to say, even the 

late and careless Puranic style still inclines to the third instead 
of second vipula, which is the classical preference. If, how- 
ever, we revert to an older selection of the epic, we find, for 
instance, in the heart of the Bhagavad Gita (830-1,382), that 
the second vipula (in the same number of verses, hemistichs, 
namely one thousand, which in all the examples now to be 
given is the number to be assumed) has twenty-nine cases and 
the third but eleven; that is, the proportion is not only 
reversed but is in very striking contrast both to the norm of 
the Ramayana and Raghuvanqa on the one hand and the 
Puranas on the other. Coincident with this is the further 
fact that, whereas Valmlki and Kalidasa have proportionally 
few first vipulas after diiambus, both epic selections above 
have more first vipulas after diiambus than after any other 
combination; while the Puranic specimens are quite classical 
in this regard, the Vayu having only one-fourth, the Agni 
only one-third of all the first vipulas after diiambus. An ex- 
tract from the Anu§asana Parvan of the epic, §1. 3,732-4,240, 
shows also an approach to the classical model (ten first 

vipulas after diiambus, twenty-three after h and 

w w each). The last case has thirty-six second vipulas 

against fifty-four third vipulas and only seven fourth vipulas 
(whereas the Gita extract has twenty-two fourth vipulas). 1 

1 The five texts, Gita, Kala, Anuf. P. ; Ramayana iv, 1-11, and Raghu- 
vanja show as fourth vipulas (in 1000 verses) 22, 10, 7, 2, 0, respectively 




A curious fact is, further, that, -while this extract of the 
Anugasana, which is a medley on the gifts of cows, origin of 
gold, and other late stuff, has but seven fourth vipulas in five 
hundred glokas, the following chapter on (kaddlias, the basis 
of which is old (rules expanded from Manu’s list of guests), 
has four in sixty glokas. Another interesting fact is that the 
thousand verses winch lead up to and follow after the extract 
from the Gita given above, 495—830, 1,382-1,532 do not keep 
the ratio between the second and third vipulas, but approach 
the later norm, having an equal number of each vipula. The 
Anugita itself contains only one-half as many “ irregular ” 
forms as does the Gita in the same amount of matter ; 1 but 
following this the epic narrative is expanded in modern form, 
and here, where the subjects are the mountain festival, recapit- 
ulation of the Bharata war (xv, 61, 1), digging for buried 
treasure, Pariksit’s birth, demise, and restoration to life, loos- 
ing the white horse, and Arjuna's renewed battles, the metre 
becomes almost classical, with scarcely a single violation of 
vipula rules and with only five cases of the fourth vipula 
to the thousand verses. Compare for instance the vipulas in 
Raghuvanga, the Ramayana (iv, 1-11), and Agvamedhika 2 
Parv. 59-77, according to vipulas: 




















The vipulas of the first thousand verses (hemistichs) of Nala are, in their 
order, 91, 33, 50, 10. Though modernized, the irregularities in Nala are 
antique: 3, 13, iva prabhain ; 12, 105, Nalam nama Timardanam (changed to 
damanam); 16, 37, katham ca bhrasta ( ? ) jnatibhyah; 20, 18, tvam iva 
yanta (now eva) ; in 12, 55, and 91, vilapatlm must be read (grammar is 
of no importance here, as will be shown below). 

i They are three cases of the second vipula after M and w w 

respectively ; five and one each of the fourth vipula after the same feet 

3 The strictness here may be measured by the fact that there is only one 
case of final hrevis in the second vipula and only three in the first ; no case 

of second or third vipula after any precedent foot save w w (and only 

one of the fourth vipula). Further, only one-third of the first vipulas follow 
a diiambus. 



Also in tlie first thousand hemistichs of A§rama there are 
only four cases of the fourth vipula. Like Magha of the 
West, the Mausala, on the other hand, which treats of 
Dvaraka and was probably a elan-tale of the West, comes 
much nearer to the antique standard, having ten fourth 
vipulas in five hundred hemistichs, three of them irregular, 
besides one further vipula irregularity (stz. 47, 132, 211, 253). 1 
It should be added too that, though (as just stated) there are 
four fourth vipulas in the first thousand hemistichs of the 
fifteenth book, yet they are all found in the fust seventy- 
seven verses, and from this point on there is not another case 
of fourth vipula for one thousand hemistichs, which is as 
classical as Vahnlki. This last selection is, hr fact, almost 
precisely on the classical model, and differs from it anyway 

only in having two second vipulas after This 

would imply an acquaintance with the classical norm, which 
can perhaps scarcely be doubted in the case of the writers 
who finally completed the poem. 

A very interesting example of how the antique null make 
the poet hark back to an older norm is given by the Sauptika. 
It will be remembered that this is almost pure narrative, but 
that at one point C'iva is addressed with a hymn and his 
demons are briefly described. This occurs just at the middle 
of a selection like those above of one thousand hemistichs. 
Now up to this point there is no fourth vipula at all, but 
with the hymn and names come five fourth vipulas within 
thirty-five (jlokas. Then the narrative is resumed, and till 
the end of the thousand hemistichs appear only three more. 
Some smaller points here also deserve attention. The num- 

1 In the next Parvan, there are four fourth vipulas in two hundred verses, 
but three are at the beginning and in three successive hemistichs, and of 
these, two are forced by proper names. That proper names are quite impor- 
tant may be shown by the catalogue at the beginning of the Harivanpa, 
where the names force up the fourth vipula to twelve (seven of these being 

in nom. prop,), and a third vipula occurs after w (in a name) ; as 

contrasted with the next thousand verses, where there are only four fourth 
vipulas. Bhavisva, partly owing to imitation of Gita and Smrti, partly to 
names, has nine in its first thousand verses. 



ber of first and second vipulas with ctesura after the fifth is 
double that of those with csesura after the fourth, and there 
is only one first vipula, and no second vipula, with final brevis. 
Finally, there are only fourteen cases of first vipula after 

m w out of fifty-four in all. Thus from every point of 

view T the same result is obtained. The little Parvan is com- 
paratively refined in style (number of vipulas, 54, 30, 35, 8). 

No doubt this parisamkhya philosophy is tiresome reading, 
but as it is even more tiresome to obtain the facts than to 
glance at them, I shall beg the reader to have patience while 
I give the results of a few more reckonings, since I believe 
they are not without a certain value. What I want to show 
is that the treatment of the fourth vipula goes hand-in-hand 
with that of other factors involving a more or less refined 
style, but not necessarily with all of them. I will take as 
my first illustration the tent-scene from Drona 72-84, and 
ib. 51-71, a group of apparently old stories on the “ sixteen 
kings that died” and allied tales. In the former there are 
four, in the latter twenty-one fourth vipulas to the thousand 
hemistichs ; in the former there is but one slight irreg- 
ularity (hu , _ w w _) ; in the latter there are six. 

But in the former there is one more second vipula than 
there is third ; in the latter these stand thirty to forty- 
seven; while after diiambus in the former there are nine 
out of forty in all, and in the latter sixteen out of fifty-five 
in all. In other words, in the last test there is scarcely 
any difference, but in that of second and third vipulas 
such evidence of antiquity as is furnished at all by this 
test is in favor of the former, whereas in the other tests it 
is in favor of the latter specimen. I have not selected these 
specimens, however, to show that all these tests are use- 
less. On the contrary', I believe they may be applied, but 
all together and with constant reference to all other factors. 
The modifying factor here, for example, is that though the 
tales of the “ kings that died ” are undoubtedly old, yet they 
are told (or retold) in such modern careless Sanskrit that 
final i is here kept short not only before br but even before 



vy. It is not enough then to say that a story in Drona or 
A nugasana is “undoubtedly old,” because perhaps it smacks 
of antiquity or even is found in a Buddhist record. It is not 
the age of the story but the age of the form in which it is 
couched that marks the age of the literature. This specimen, 
for example, enumerates earth’s islands as eighteen in num- 
ber, a sure mark of lateness, but here supported by other 
data. Another extract from Drona, an ordinary battle-scene, 
adhy. 92-100, has, to be sure, thirteen fourth vipulas, but the 
vipulas, in their order, run 44, 14, 87, 13, with not a single 
irregularity of any sort, while only ten of the forty-four are 
after diiambus ; in other words, as clean a scheme as might be 
met in Valmlki, except for the fourth vipula, and even here 
eight of the thirteen are in proper names. Less classic in 
appearance, but still far removed from the free epic type, 
is the passage dealing with the deaths of Bhurigravas and 
Jayadratha (vii, 141-146, not quite a thousand verses), im- 
portant because of its mention of Valmlki, 143, 67. Here 
the vipulas run 43, 33, 18, 11 (four of these in nom. prop.), 
with three irregular forms of the second vipula. 1 A fourth 
of the first vipulas follow iambus. On the other hand, in the 
death of Drona and the following scene, vii, 190-198 = 8,695- 
9,195, only one-sixth of the first vipulas follow' iambus and 
there are no certain exceptions. The scheme of vipulas is 
here 30, 28, 43, 9 (two in nom. prop.), that is, a more modern 
preponderance of third vipulas. Several other features show 
modern touches. Thus in 192, 7, Rudrasye ’va hi kruddhasya 
is either a very careless vipula or contains an example of the 
Puranic licence (taken from Prakrit) of short vowel before 
kr; while in the same passage, gl. 13, eso or esa hi parsato 
vlrah, we have to choose between careless sandhi or careless 

metre. In 190, 33, the antiquity of uv is in an inherited 

name, Jamadagnih, where, as in similar cases, the old licence 
persists even into Puranic wTitings . 2 In 195, 44, kadarthx- 

1 In 146, 7, occurs the rare combination \j, \j \j w • The read- 

ing of C. 6,245 = 146, 92 is vicious, and is corrected in B. 

2 Names, formula;, and numerals often retain this licence, e. g., rsayaf ca, 



krtya is a late phrase, and in 191, 37, the stereotyped man- 
oeuvres are twenty-one in number (the earlier epic having 
fourteen). Here, then, the vipulas (110 in number, slightly 
below the average) do not badly represent the period of the 
selection, which is a worked-over piece, intended to save the 
heroes from blame, and is often incongruous with the rest of 
the epic ; as in the humbug of the war-car “ not touching the 
ground hitherto.” When Y udhistliira tells a lie his car drops 
to the earth for the first time! But “hitherto” there has 
been no mention of this conscientious chariot, which here is 
represented as having floated just above the earth. 

In Kama we may compare the thousand verses of 18-29, 
where there is late battle-action (guna for jya for example), 
with the five hundred fifty vei’ses of old tales in 33-34. Each 
has seven fourth vipulas, though one is only half the length 
of the other. In Sabha the interest centres on the gambling- 
scene, certainly the kernel of the old tale. Here, ii, 50 ff., for 
a thousand verses, there is the greatest number of fourth 
vipulas (thirty-six, nine of which are in proper names) and 
the most irregular forms ; three cases of a third vipula after a 
spondee, one case of a prior pada ending in iambus, two cases 
of the “ fifth ” vipula, w w _ one case of first vipula after a 
brevis, besides six cases of ordinary exceptions (second 
vipula not after _ w _), all of which remove the piece far 
from the almost classical norm found in some of the cases 
given above. It is in fact Puranic. 1 Of course the scene is 
intense and exciting; but I opine that no poet who had once 
learned to walk the straight and narrow way of the later 
stylists would ever get so excited as to use thirty-six fourth 

xii, 349, 78 ; da<;a devah, Ag. P. xvii, 6. The same cause induces the fourth 
vipula in many cases of the Ramayana. For example, the only fourth vipula 
in the first thousand verses of R. iii, vaikhanasah valakhilyah, 6. 2. 

1 Compare for instance the 505 9 lokas or 1010 verses in the first twenty 
chapters of the Agni Purana, where the vipulas in their order are 41, 26, 50, 
67, with six irregular second vipulas (not after iambus) ; five third vipulas 
not after iambus ; and only nineteen of the fifty-seven fourth vipulas after 
iambus. The first vipula in the gambling-scene is run up by the repetition 
of one phrase. They are in order, 60 (odd), 34, 51, 36. 



vipulas in a thousand verses ! Besides, there are other pas- 
sages almost as dramatic. If we compare the Jatugrha and 
four hundred verses of the Hidimba stories, which together 
make about a thousand verses, we find eleven fourth vipulas, 
half of which are in proper names, only one case of a third 
vipula not after m _ w and three ordinary exceptions in the 
case of the second vipula. The Klcaka in Virata is also a 
lively scene, which with a slight addition of circumjacent 
verses contains a thousand verses (325-825), and here the 
vipulas are in order, 42, 24, 52, 6 , with no unusual exceptions 
and only three ordinary exceptions in the second vipula ; 1 
while five of the six fourth vipulas are in proper names and 
in the title rajaputrl. 

But since it may be objected that the subject matter is after 
all the essential factor, I will compare a philosophical section 
where the matter is that of the Bhagavad Gita, for example 
(lanti, adhy. 311 and following for one thousand verses. Here 
the vipulas in their order are : 









Compare R. iii, 1- 

-16, 60 



It will be seen that the extract from Canti is almost on a 
metrical par with the ordinary narrative of the Ramayana 
(1010 verses). But further, of the three cases of fourth 
vipula in Canti, one is in a proper name and there are no 
anomalous forms of unusual character, and only two ordinary 
exceptions (second vipula), while the Gita has a dozen irreg- 
ularities of all kinds (including “ fifth vipulas ”). I may add 
to these specimens the instructive opening of Udyoga, where 
for nearly two hundred 9 lokas there is epic narrative followed 
by the old tale of Nahusa and Indra. The vipulas, for one 
thousand hemistichs, are here 55, 25, 46, 10, respectively, but 
nine of the ten are in the old tale, adhy. 9 ff., 9 I. 227, the 
other one being in a proper name. In the old-style didactic 

1 By ordinary exceptions I mean cases where the second vipula does not 
follow an iambus. 



verses, v, 35, 60 ff., on the other hand, there are six fourth 
vipulas in only five §lokas. 

Whether we are entitled to draw from these data conclu- 
sions in regard to the time when the several selections were 
written may be doubted in all cases when the percentage of 
fourth vipulas is not sustained by other factors. But it seems 
to me, as I have said, that it is not unreasonable to assume a 
more modern authorship in the case of a sustained refinement 
of style. Even in cases where the data are not of an extreme 
character I think it is legitimate to question whether a com- 
parative refinement is not of significance. Take for example 
the thousand verses of Udyoga, 119-133 (4,000-4,500). Here 
the subject-matter of the selection is the Bliagavadyana. 
Nothing in the account seems antique; on the contrary, the 
whole story appears on the surface to be a late addition. Now, 
going beneath the surface, we find that the vipulas are in order 
48, 23, 39, 13, but that eight of the last are in proper names. 
The collateral evidence agrees with tire two factors here 
shown (preponderance of third vipula over second, compara- 
tive scarcity of fourth vipula); for of the forty-eight only 
twelve are after iambus; of the twenty-three, nineteen are 
after iambus ; wlrile of the four ordinary exceptions (after 

— ) f' vo are in the same phrase, yatha Bhlsmah (lanta- 

navah ; the third vipula is perfectly regular or has at most 
one exception, manena bhrastah svargas te (though, as a 
matter of fact, there cannot be much doubt that we have here 
the late light syllable before bhr) ; the five fourth vipulas not 
in proper names are all after iambus except one, contained in 
an hereditary phrase, esa dharmah ksatriyanam. Here then, 
though there is not the striking classical smoothness found in 
parts of the pseudo-epic, the few fourth vipulas agree with the 
other data in marking the piece as rather Tefined, perhaps 
modern, when compared with the oldest epic style. 

When, however, the data are contradictory, as often 
happens, we may imagine a rehandling, as in the suspected 1 
Narayana exploitation in 9anti, from the end of the prose in 
1 Compare Buhler in Indian Studies, No. ii, p. 52. 



343 to the end of 351, about a thousand verses, 13,224-13,7 40, 
where the scheme of vipulas is in order 80, 50, 32, 15 ; thirty- 
one of the eighty being after iambus ; with five cases of irreg- 
ular second vipula and perfectly regular third vipula (save 
for a slightly neglected ceesura, dharinapratisthuhetug ca). 
The fourth vipula here owes its large number solely to names, 
numbers, and an old phrase. Thus we find, not after iambus, 
tasmin yajiie vartamane (like the regular phrase tasmin 
yuddhe vartamane); Vasudevam (second foot); Sarhkhyam 
Yogam Paficaratram; Sankhyayogam (second foot); Paiica- 
ratram (second foot); Vaiklianasah phenapebliyah ; Sarva- 
krcchram (name of vrata) ; astadanstrau ; leaving two cases, 
durvijheyo duskarag ca and jayamanam (as second foot) 
after m ; with five more after iambus. 

Rather a striking example of the mixture of styles is given 
by ix, 48, where Indra and the jujube-girl are concerned. 
This is plainly interpolated with a (yiva parody. Compare, 
for instance, piito 'smi te gubhe bhaktya tapasa niyamena ca, 
in the Indra dialogue, with §1. 45 (in the interpolation), prito 
'smi tava dharmajne tapasa niyamena ca. Now the original 
Indra tale has fifteen vipulas in the first thirty odd verses; 
but the same number of glokas in the following (fiva parody 
shows only five vipulas. 

Again it must be remembered that some rather modem 
selections are interspersed with old material. In the six 
hundred odd verses of the Qakuntala episode, for instance, 
the style is modern to a certain extent, the first vipula being 
less common after iambus than after spondee, and only one 
ordinary exception occurring in the second vipula, wliile 
there are no unusual anomalies. But the passage has thirteen 
fourth vipulas, which is not a refined ratio and may be ex- 
plained only partly by the presence of Dharmagastra material, 
brdi sthitah karmasaksi, bharyam patih sampravigya (Manu, 
ix, 8). In my opinion the episode is old, but, like many 
ancient tales in the epic, it has been rewritten and in its 
present shape is not so old as the vanga and Yayati episodes 
following, where there are as many fourth vipulas and more 



anomalies. This episode has recently been made the subject 
of an interesting study by Dr. Winternitz , 1 who believes that 
it is of very doubtful antiquity, because it is lacking in the 
Southern manuscript examined by him and because the knot 
is untied by a “divine voice,” instead of by a ring. One 
point not noticed by Dr. Winternitz must be remembered, 
however, namely that the Harivah?a recognizes the episode 
and cites from it, apropos of the “ divine voice,” 2 so that it 
existed in the present version, if not in its exact form, before 
the Harivan§a was added to the poem ; though I should not 
deny on that account that it was of doubtful antiquity. 

I think I have now shown sufficiently that the different 
parts of the epic cannot revert to one period, still less to one 
poet, and will leave this minute analysis with a repetition of 
the statement that, whereas the parts already cited clearly 
reveal more styles than we may attribute to one age or man, 
occasional freedom of style in respect of vipulas does not in 
itself indicate antiquity ; but when all the elements agree in 
refinement, this sustained refinement certainly points to a dif- 
ferent environment and may imply that some parts of the epic 
are later than others. There is a refined style and there is a 
careless style, but the latter is late Puranic as well as antique, 
and mere carelessness proves nothing beyond the fact that the 
poet either did not know or did not regard classical rules. 
On the other hand, even the careless Puranic writers gener- 
ally show a greater number of first vipulas after spondee than 
after iambus and more third than second vipulas. When, 
therefore, even these rules are not upheld and we find besides 
other irregularities, such as the three cases of the fifth vipula 
in the Gita, we may rest assured that the writer was rehand- 
ling material more antique than that of other passages. I say 
rehandling, because the Gita has clearly been rewritten by a 
modernizing hand, as is shown not only by the circumstance 
already noticed that the heart of the poem differs in style 
from its beginning and ending, but also, for example, by the 

1 Indian Antiquary, 1898, pp. 67 and 136 ff. 

2 i, 74, 111 = H. i, 32, 12. 



fact that in Gita, 12, 15 we read yasman no ’dvijate lokah, a 
metrically bettered form of yada ca ’yam na bibheti, a phrase 
found intact in other parts of the epic. 1 

The usual epic §loka, apart from occasional variations, 
differs, as I have said, from the classical model most conspicu- 
ously in vipula licence ; as will clearly be seen at a glance 
on comparing the normal epic forms with the classical in the 
following tables, where is given first the average epic usage : 

First Foot 

Second Foot 


w \j 













^ kj 





and then the forms permitted and almost never exceeded in 
Kalidasa (“ common ” here means not unusual yet not passim) : 

Firat Foot 

Second Foot 

v-/ w 

__ _ 

i ^ 

w ^ 







^ w 


1 Per contra, in the Sanatsujata Parvan, v, 46, 26, yatlio ’dapane mahati is 
a metrical improvement on Gita, 2, 46, yavan artha udapane. Other later 
features in the Gita are the long sentences already referred to ; the sporadic 
intrusion of the Maya doctrine (discussed above in Chapter Three), and per- 
haps also the recognition of the Vedanta Sutra. 



The usual Ramayana <jloka agrees with this later scheme, 
except in admitting sporadic cases of the fourth vipula after 
an iambus . 1 

But, to get a comprehensive notion of the epic gloka, in its 
rarer forms as well as in its normal or average appearance, 
one must contrast these tables with the next, which gives, I 
believe, about all the Bharata combinations for the prior 
pada : 

First Foot 

Second Foot of Prior Pada of £loka in the MahabhSrata 

w w. 

— oo id 



w_wi d 

^ — 



p 0 

P 14 














P to 














P 3 

p H 












id \j 

P 4 














Hu vu 

P 5 











hH uv 


































Abbreviations : p, passim ; c, quite common ; r, rare ; s, very rare, sporadic. 
The interrogation marks indicate doubtful cases, for which the illustrations 
(as numbered in the table) must be consulted in Appendix B. For the corre- 
sponding table of tristubh forms, see below. 

1 For the few exceptions to these much more restricted forms of the 
Ramayana, see Jacobi’s I’amayana. There is to this uniformity not a single 
exception, for example, in the two thousand hemistichs found in R. iii 1-16; 
iv, 1-11. Final brevis is rare in the second, but not in the first, R. vipula. 



Midway between the classical and the normal Bharata gloka 
stands that of the Ramayana. The latter does not admit 
many forms found in the Mahabharata. Some of these are 
older, some are later. But in its aberrations from the subse- 
quent type of the classical writers the Mahabharata is much 
freer than the Ramayana ; freer not only in admitting other 
types of gloka than those found in the Ramayana, but also in 
the way of handling glokas common to both epics. The gloka 
of the Upanishads (Ivatha, Kena, I§a) admits as prior padas, 

^ \J W 

w \j w 

^ KJ w W 

w w KJ 



W W W w 

Quite so free the Mahabharata gloka is not, but it admits 
here and there as second foot ^ _ and and as 

first foot, __ w w which is also found as first foot of the 
second pada. So free as this the Ramayana is not. From the 
occurrence of these freer forms we are entitled, however, to 
say only that the Mahabharata is occasionally freer in its 
gloka-foot than is the Ramayana. But it is generally freer, 
and much freer, in the non-observance of vipula rules. Tins 
“ characteristic stamp ” of the Mahabharata, as Professor 
Jacobi calls it, 1 in distinction from the Ramayana, is one that 
it shares to a great degree, as I have said above, p. 79, with 
the early Buddhistic and Upanishad gloka, which is so wide 
a province that the explanation given by Professor Jacobi 
seems to me to be inadequate. 

Yet if, as I think, the qloka of the Ramayana shows that it 
is in its present form not only more refined (which is con- 
ceded) but also later than parts of the Mahabharata, the latter 
no less is later than the Ramayana in other parts. There are 
five sorts (perhaps stages) of gloka reflected in epic and pre- 

1 Gurupujakaumudi, p, 53. 



epic literature (besides its parent Vedic anustubh). The first 
is the free gloka of the Upanishads. The second is the less 
free, but still unrefined, gloka of certain parts of the Mahabha- 
rata. The third is the current Bharata gloka. The fourth is 
the gloka found in parts of the pseudo-epic, a gloka which 
stands on a par with the gloka of the Rdmayana. The fifth is 
the continuous iambic gloka, which is found only in the 
Mahabharata and is certainly later than other epic forms of 
gloka. Nearly forty stanzas of this type, consisting of iambs 
only (allowing final anceps), that is, over six hundred succes- 
sive iambs — evidently a late tour de force — occur in xii, 322, 
33-71, written by a poetaster who presents old ideas in a new 
style, 1 as in this specimen : 

pura vrka bhayamkara mauusyadehagocarfth 
abhidravanti sarvato yatag ca punyagllane 
pura hiranmayan nagan 2 nirlksase 'drimurdhani 
na matrputrabandhava na samstutah priyo janab 
anuvrajanti samkate vrajantam ekapatinam 
yad eva karma kevalam purakrtaiii gubhaqubham 
tad eva putra s&rthikam bhavaty amutra gacchatah 
iha ’gnisuryavayavah qariram aqritas trayah 
ta eva tasya saksino bhavanti dliarmadarqinah 

So far as I have observed, although the prior pada may end 

either in w w or in w_ w _, the union of both in one 

gloka is unknown to the epic. This is a combination of one 
freedom with another. The forms, therefore, were felt as 
liberties and consequently were not multiplied in narrow com- 
pass. Such glokas, however, are found in the early style, and 
even the Mahabhasya gives us a sample, apparently from 
some defunct epic source, where one prior pada is aharahar 
nayamano and the following is Yaivasvato na trpyati. 3 This 

1 Found, for example, in the Vedantasara of Sadananda : satattvato 'ny- 
athapratha vikara itv udlritah, 162, etc. For the single pada, diiambic prior, 
see vii, 55, 49, cited below under Diiambus. A single pada of this sort is both 
Vedic and Puranic. 

2 See Proverbs and Tales in the Sanskrit Fpics, A. ,T. Phil., vol. xx, p. 24. 

• Cited by Weber, InJische Studien, vol. xiii, p. 483. 



may indicate that our epic has been metrically refined ; other- 
wise we should perhaps find in it the same freedom. Notice- 
able also, I may say in view of the paragraph below on the 
posterior pada, is the absence of any certain case of a hemi- 
stich ending like the prior pada in w This Gatha form 

is found in the examples from the Bhasya (compare, for 
instance, ratrim ratrim smarisyanto ratrim ratrim ajanantah J ) ; 

but the utmost freedom of the epic is vy _ at the end of 

a hemistich, except in the semi-prose example given below 
(on the Diiambus) ; a circumstance that makes it impossible 
to believe that the epic in its present form is older than the 
second century b. c. 

The Posterior Pada of the <)loka. 

Owing to the prevailing diiambic close of the hemistich 
there is little variety in the posterior pada. The first foot 
may have (sporadically) any one of seven forms, that is, with 
the exception of the unique opening of the prior pada in pro- 
celeusmaticus, the first foot of the posterior pada may be 
identical with any of those of the prior pada. The second 

foot is a diiambus, or sporadically w _, and w 


Fibst Foot. Second Foot. 

1 . ^ 

2. H 

3. ^ 

4. m 


6. m 

7. H 




\-/ — w 
— w 

r sporadic 

Of these forms, the first three and the fifth occur also as 
prior padas (with diiambic close). The seventh form is 
avoided because it is the jagatl measure ; but in general three 
final iambs are avoided. The first form is an oddity. Illus- 
trations of all the forms of prior and posterior padas will be 
found in Appendix B. The rules for this pada are given 

1 Weber, loc. cit., p. 485. 



Of the forms of the first foot (third of the hemistich), 
all except Nos. 1 and 7 are found passim in both epics; of 
the forms of the second (fourth) foot, with rare exceptions 
only the diiambus is found. The commonest forms are Nos. 
2 and 3 (ending in spondees). After the first vipula both 
of these are equally common and each is about twice as 
frequent as No. 4, and from two to four times as common 
as No. 5 (final trochees). No. 6, ending in a pyrrhic, is 
sometimes surprisingly frequent after tliis vipula; but at 
other times is lacking for whole test-sections of a thousand 
verses. After the second vipula, which usually ends in an 
iambus, as after the first vipula (also iambic), Nos. 2 and 3 
are favorites; No. 3 being perhaps a little more frequent. 
Here Nos. 4, 5, 6, are much less common; No. 6, however, 
is rarest of all. After the third vipula, No. 2 sometimes 
yields in frequency to No. 3; but in other sections tliis foot 
still holds its own, and as in the former examples is even 
twice as common as other combinations, though it practi- 
cally repeats the vipula, , m Here Nos. 4 

and 5 are about on a par, sometimes only a third as com- 
mon as No. 2, sometimes more frequent, with No. 6 half 
as common as Nos. 4 and 5. 1 After the fourth vipula, how- 
ever, No. 6 is as common as any other, sometimes slightly 
in excess, with the others about on a par; No. 4 being per- 
haps the rarest. 

Such varying ratios are not worth tabulating. They show 
that while the posterior pada is not absolutely uninfluenced 
by the form of the prior, yet the determining factor is rather 
the inevitable presence of the former’s diiambus, since the 
only marked choice is for spondees before it, as in the first 
pada before an iambus (pathya). The other cases reveal 
merely a shifting predilection for one of several forms, all 
of which are used pretty freely, the strongest influence of 
the preceding vipulas being simply that the usual prefer- 

^ For example in one text case of a thousand verses, there were twelve 
cases of No. 2 ; four each of Nos. 4 and 5 ; and two of No. 6. In another, 
nine of No. 2 ; eleven each of Nos. 4 and 5 ; four of No. 6. 



enee for a spondee before the final diiambus is changed into 

a natural aversion after a spondaic vipula, or 

_ w , but this is what might have been predicated in 

advance. After pathyas one foot is as permissible as another. 
Occasional variations here are of even less significance than 
in the case of precedent vipulas. 

As all the forms of the prior and posterior padas may 
have syllaba anceps, both initial and final, each pada may 
appear in four forms. 1 Not to speak of the important modi- 
fications introduced by a varying caesura, the syllabic com- 
binations resulting from joining any one of the four kinds of 
each form of the posterior pada with any one of the four 
kinds of each form of the prior pada results in a large num- 
ber of possible verse (hemistich) forms; while, since any 
form of the first hemistich may be united with any form of 
the second hemistich — to take only the commonest eighteen 
forms of prior pada 2 3 and the five current forms of posterior 
pada — the resultant variations in the form of the verse (hemi- 
stich) are 1440 ; in the case of the whole stanza (gloka), 
2,073,600; so that one could write twenty Mahabharatas in 
glokas (the present one in the Calcutta edition contains 
95,739 glokas) and never repeat the same metrical stanza. 
Despite this latitude, however, the poets are not at all shy 
of repeating the same syllabic hemistich in juxtaposition, 
showing that they were indifferent to the vast possibilities 
before them and cared for caesura more than for syllables. 
Thus Nala v, 45b-46a: 

Damayantya saha Nalo vijabara ’maropamah 
janayamasa ca Nalo Damayantya mahamanah 

1 In explanation of the number of examples in Appendix B, I would say 

that, for the sake of showing the truth of this anceps theory, I have given the 
four forms, syllaba anceps at both ends of the pada. 

3 That is, the first six pathyas, the first four forms of the first and second 
vipulas respectively, the first form of the third vipula, and the first three forms 
of the fourth vipula. These, by the way, are the forms “approved” by mod- 
em native scholars, according to Brown, Prosody, p. tl. 




The Diiambus 

The rule of diiambic cadence appears to be violated in the 
epic. Far from regarding this as an archaism on the part 
of epic poets, one should recognize in such cases only a 
Puranic licence or adaptation of the Gatha freedom con- 
spicuous in all popular and therefore loose composition. 
Not only is that rule for Sanskrit which allows a syllable 
to remain light before kr, pr, br, hr, valid for the later epic, 
but the extended Prakrit licence is also found, whereby al- 
most any conjunct 1 may be treated for metrical purposes 
like a single consonant. Examples are found both in the 
Mahabhiirata and the later Ramayana. For the latter epic, 
Jacobi, Das Ramayana, p. 25 ff., should be consulted, where 
are given examples in br, pr, mr, ml, tr, hr, kl, and gr, e. g., 
knit tu Ramasya prltyartham, R. v, 53, 13 ; viniigayati trai- 
lokyam, ib. 1, 65, 13. From the Mahabhiirata (in the ap- 
pended illustrations of epic gloka forms) I have drawn 
several examples which are doubtful, because they may be 
regarded either as irregular (unusual) forms without this 
licence or regular forms with it. Such are daga pailca ca 
praptani (No. 25); hate Bhlsme ca Drone ca (No. 22); 
sarvagaucesu Brahmena (No. 23) ; abhijiinami brahmanam 
(No. 41) ; manena bhrastah svargas te (No. 22) ; Ruclrasye 
’va hi kruddliasya (No. 24). But further, in a few cases, 
gr also seem to leave the syllable light behind them, as in R. ; 
e. g., adyaprabhrti grlvatsah (Nos. 15, 26, 39). Nor are we 
aided as much as we should like to be, when, turning from 
these doubtful priors, we examine the posterior padas. For 
though at first it seems decisive that such a pada appears as 
putram Ipsanti brahmanah, vii, 55, 21 ; tosayisyami bhra- 
taram, viii, 74, 30 ; yet it is not quite settled whether we have 
here a syllable to be read light because, as in Greek, mute 
and liquid really make insufficient position, or whether the 
syllable is heavy but is allowed to stand for a light. For 
there are other cases where mute and liquid are not the 

1 Colebrooke, Essays, vol. ii, p. 05, note (“ any conjunct ” in Prakrit). 



components of the conjunct. First we have double semi- 
vowels, which ordinarily make position, but fail to do so, 
for example, in vii, 55, 50, abhi t)vaitye ’ti vyftharan, which 
appears after a gloka with a diiambic prior pada (sa cen 
mamara Srnjaya). So the syllable remains light before cell 
and ks and dv, or the metrical rule is violated. In the Ramii- 
yana and in the Mahabharata, cases of liquid and ks are found 
more rarely in tristubhs, but often enough to show that they 
are occasionally allowed. Thus in R. iii, 63, 6 b, °etya 
klegam (tristubh). 1 In M. : 

viii, 37, 24 d, tyaktva pranan anuyasyaml Dronarn 
xii, 73, 7 a-b, yada hi 2 brahma prajahatl ksatram 
ksatrarii yada va prajahatl brahma 
xii, 319, 89 b, sarve nityam vyaharante cii brahma 

In sum, the cases where this licence may be assumed for 
the later epic style 3 are before dr, br, bhr, mr, kr, pr, kl, tr, 
§r, hr, ty, vy, gy, dv, cch, ks. For dv, compare striyag ca 
kanyag ca dvijag ca suvratah, iv, 37, 33 ; avartanani catvari 
tatha padmani dvadaga, xiii, 107, 26; for cch, yugesv isasu 
chatresu, vii, 159, 36, where the texts avoid the third vipula 
by writing ch for cch. But whenever a short syllable is needed 
before cch it is got by dropping c (sometimes in one text, 
sometimes in another). For ks, ca kslyate, xii, 343, 87 ; 
ranabhitag c& ksatriyah, vii, 73, 39 (apparently an interpo- 
lated passage) ; exactly as we find the same licence in Yayu 
Purana, viii, 155, where the gloka ends °sS ksatriyan, or as 
ib. v, 28, we find the common licence before br , lokan srjati 
brahmatve. For <jy, see below on the tristubh scolius ; mr, 
ml, ty, tr, I have not found in the Mahabharata. They 
seem to belong to the latest parts of the Ramayana. 

1 Jacobi, Ram. p. 27, gives cases from the later R. In G. v, 28, 5, na tyajet 
(B. correct v. 1.) ; G. ii, 27, 24, tvaya saham (B. correct v. 1.). 

2 This section is free ; but in xii, 202, 22 b, there is an upajati group where 

we find tad ev& pratyadadate svadehe {si being demanded). 

3 Examples of regular (heavy) position before mute and liquid are found 
everywhere, e. g., ix, 17, 41, 43, 44, 47, 51, 52 ; xii, 63, 8, 27 ; 64, 16, 18, etc. 
This is the rule ; failure to make position or neglect of quantity is the excep- 
tion and is characteristic rather of the later epic, as shown by the examples 



We may, I think, assume that the liberty in respect of 
liquid and consonant was first introduced into epic Sanskrit, 
and that then in the later epic this was extended, with Gatha 
freedom, to cases where the precedent syllable cannot be light, 
but is reckoned so. Therefore, while the early epic has only 

diiambic close, the later epic (like the Puranas) admits w _ 

as an equivalent; not of course generally, but sporadically, 
where the writer is late and careless, as is indicated by the 
character of the sections where such illegitimate freedom is 
found. So in the tristubh scolius, there are a few cases of 
careless writing where a heavy syllable stands in the place 
of a light one. To say that this heavy syllable is light be- 
cause it ought to be, is misleading. The weight may be 
ignored, as in Prakrit (though there mutilation explains 
much that appears of this nature), but it must exist. Even 
the Greek poets occasionally pretended that a heavy sylla- 
ble was light. In fine, w _ must be admitted as an 

occasional fourth foot of the hemistich, though it is avoided 

whenever possible. 1 For the foot w , I have only the 

hemistich etac chrutva tu Kauravyah (pibim praclaksinam 
krtva, iii, 194, 7, but this is apparently an accidental verse 
in a prose narration. 

Poetic Licence. 

In general, however, while the epic poets are here and there 
rough and uncouth in their versification, the normal epic style 
sacrifices a good deal to what is regarded as good metrical 
form. Such a sacrifice, which culminates in the classical rule 
that one may use ben for bean (masa for masa) if one only 
follows the metrical norm, is found most clearly exemplified 
in this very case of the diiambic close ; a proof that the diiam- 
bus was regarded in general as obligatory. 2 But it is also to be 
noticed in the observance of preferred vipula forms at the sae- 

1 Its restitution in Prayna ii, 6, rco vaj unsi samani, yajfiah ksatram [ca] 
brahma ca, is at least probable. 

2 Compare even in the Rig Veda the regular irregularity of yavisthiam, 
for yavistham, for the sake of the diiambus ; and see now an article by Pro- 
fessor Bloomfield on this very point, JAOS. xxi, p. 50 If. 



rifiee of (Sanskrit) grammatical accuracy. There are, indeed, 
cases where word-structure appears to be needlessly sacrificed ; 
but the vast majority of cases in which Sanskrit grammar is 
violated have to do with metrical necessity or predilection. 

As already stated, the most frequent cause of such violation 
is the well-nigh obligatory diiambus at the close of a verse, as 
in phullam Gomati-tlrajam, iv, 17, 12. The diiambic rule, as 
ordinarily stated, is included in this presentment of cloka re- 
strictions : “ The second, third, and fourth syllables of a pos- 
terior pada should not form a tribrach, anapaest, or amphi- 
macer, and the fifth, sixth, seventh, and eighth syllables should 
make a diiambus or second paeon, while the tribrach and 
anapaest rule apply also to the prior pada.” Obviously, in the 
posterior pada, the tribrach rule, forbidding 

^ \J \J \J \J vM 

is to avoid a succession of four or five short syllables ; wlfile 
the anapaest and amphimacer rule, forbidding 

is to avoid the (jagatl) close of three final iambs. The rule 
then for the even pada is simply: Posterior pailas must end 
with diiambs, but must not end with triiambs, and must not 
contain a proceleusmaticus. 

The following examples illustrate how secondary is San- 
skrit grammar to this metrical rule: yag ca g unyam upasate 
(for upaste), 1 v, 33, 39 ; na sma pagyama laghavat, vii, 146, 5 
(necessarily present) ; bharyayai gacchati vanam, R. ii. 32, 8 ; 
setihase c& chandasi, xiii. 111, 42; kathakhyayikakarikali, ii, 
11, 36, and svadha ca svadhabhojinam, R. vii, 23, 23; yatha 
lii kurute raja prajas tam anuvartafe,® R. vii, 43, 19; madliuni 
dronamatrani bahubhih parigrhyafe, R. v, 62, 9 (not in G.) ; 
apakramat, ix, 11, 62. 

1 So we find at the end of a tristubh pada, upasate yah, iii, 5, 19b. Less 
common is the second person, moksadharmam upiisase, xii, 315, 15. 

2 This is simply a case of sacrifice to metre by a pedant who imitates 
Manu yiii, 175, where prajas tam anuvartante is the close of a prior pada. 
Another form of this proverb, by the way, is shown in R. ii, 109, 9 : yadvrttah 
santi rajanas tadvrttah santi hi prajah (Sj)r. 1,013, 1,652, 5,708). 



These examples comprise different classes, where, metri 
causa, are changed (a) the conjugation or mode; (b) the 
temporal termination; (c) the feminine participle; (d) the 
euphonic rule ; (e) the gender ; (f) the syntactical combina- 
tion ; 1 (g) length of root-vowel and other sporadic cases. 

Of these, by far the commonest are irregularities in the 
temporal termination, and in the ending of the feminine par- 
ticiple. Of these two, the usual changes are the substitution 
of preterite for present endings and atl for anti; less often, 
present for preterite and anti for atl. The participial change 
is the commonest of all, and what is most important is that 
scarcely any of the irregular participial stems are irregular 
from any other cause than that of metrical preference, and 
the greater number are fashioned simply to give diiambus at 
the end of the hemistich. I lay especial stress on this because 
in the lists of such changes occasionally published either no 
weight at all has been laid on the motive of the change, or 
the motive has been only incidentally acknowledged, or thirdly 
the lists have been made with reference to the class of the 
participle, as if the conjugation were especially important. 2 
The only thing of importance, however, is the metre. What 
has been lost sight of, or not seen, is that not only the obvi- 
ous cliiambic rule but also the vipula preferences come strongly 
in play, especially in the Ramayana. A few examples will 
illustrate this. 

First for the diiainbus: ca ’nyarii gatim apagyatl, R. vi, 47, 
10; kurarim iva vagatun, Nala, 11, 20; so elsewhere in Mbh., 
abhilapsatl, cikirsati, nadayatl, aveksati, anvesati; and in 
Ram., parigarjatT, yacati, anudhavatl, janayatl mama, etc. 
Likewise in the verbal ending: adho gacchamX medinTm, i, 
13, 18; duhkham prapsyama darunain, ix, 59, 30; yuddhe 
kim kurrria te priyam, ix, 32, 62 ; katlia draksyama tam purim, 

1 See below, on dialectic Sanskrit. 

2 At the same time I wish to acknowledge my indebtedness to the follow- 
ing lists as collections of material: For four books of the Ramayana, 
Bohtlingk, Berichte d. philol. histor. Classe d. Konigl. Sachs. Gesell. d. Wiss. 
1887, p. 213 ; Holtzmann, Grammatisches aus dem Mahabharata. 



R. ii, 47, 11; na ca pa§yama Maithilim, R. iv, 50, 15; 56, 13. 
Compare also the striking example in R. ii, 91, 59: nai Va 
’yodhyam gamisyamo na gamisyama Dandakan. These ordi- 
nary irregularities might be exemplified with hundreds. 1 
Other cases are less frequent; but to the same cause is due 
the close of hemistichs in tav akurvatam, i, 176, 9; the fre- 
quent change of voice, as in svargam Ihantx nityagah, vii, 
71, 14; the change of vowel-length in upakramat, apakramat, 
parakramet, vii, 54, 58; ix, 11, 47; 11, 62; xii, 140, 25; so 
r pi niskraman, R. iv, 50, 9; Laksnuvardhanah (passim) and 
the frequent loss of augment. 2 One of the most striking 
verbal changes is in na bibhyati for na bibheti in i, 75, 53 ; na 
bibhyase, R. iii, 46, 30. 

The other half of the rule for the posterior pada is kept by 
avoiding three iambs and a succession of four breves, with a 
sacrifice of the normal quantity, in praceraso dacja (so ex- 
plained in P\V. s. v.); sakhiganavrta, Nala, i, 24; na <jrir 
jah&ti vai tanuh, xi, 25, 5 (jahanti for jahati, below); upa- 
sante mahaujasah, R. vii, 37, 19 (upasate in 20) and 21 ; 
ayatlhitam ucyate, G. iii, 44, 11; and instead of adharayam 
(mahavratam), samadharam, R. vii, 13, 25. Compare also na 
svapami nitjits tadii, Nala, 13, 61, patois for svapimi; and the 
middle draksyase vigatajvaram, ib. 12, 93, witli draksyasi in 
92 and 95; draksyase surasattamam, v, 14, 5. 3 

In the prior pada, to avoid the anapaest the same form is 
used, draksyase devarajanam, v, 11, 24; the sandhi of eso hi 

1 One of the commonest cases is the substitution of sma for smah. This 
is found oftenest in the prior pada but also in the posterior, e. g., R. iv, 65, 11, 
anupraptah sma sampratam. 

2 Compare also the endings patnisu, prakrtljanah, R. i, 37, 6; 42, 1; 
grhagrdhniinam, R. vi, 75, 14, marryunam, ib. 15 (dirghabhava arsahsays the 
scholiast) ; kopena ’bhiparivrtah, R. vii, 58, 22 (below) ; anudaram, xiv, 46, 47. 

3 Here too belongs the use of the future imperative in ix, 25, 44, drakgya- 
dhvam yadi jivati, followed by yudhyadhvam saliitah sarve. Bohtlingk, loc. 
cit., denies to the epic a future imperative. The case I have cited, however, 
is not in Holtzmann’s list (loc. cit. § 938), on which B. draws for his material, 
and it seems to me conclusive in favor of such a form (and meaning). Were 
it not for the breves the poet would have used pa^yata (not draksyatlia), as 
is shown by yudhyadhvam and the general situation. 



parsato virah, yii, 192, 18; the long vowels in Pusanam 
abhyadravata, vii, 202, 59; grata vati nama vibho, ix, 48, 2; 
and the change of conjugation in dadanti vasudham sphltam, 
xiii, 62, 46. To avoid diiambus at the close of the prior pada 
we find, for the genitive, dadarca dvairatham tabhyiim, vii, 98, 
26 ; the participial exchange noted above, kusumany apacin- 
vanti, R. iii, 42, 82; jananti, R. ii, 10, 35 and Mbh. i, 78, 6; 
and various sporadic irregularities in the latter poem: pra- 
daksinam akurvanta, viii, 72, 12 ; pusnamy ausadliayah sarvah, 
i, 78, 40 ; Duryodhanam upasante, viii, 84, 12 ; gayanam samup- 
asanti, vii, 72, 40 (so G. vii, 41, 2); valukiim, pattibhih, etc., 
R. iii, 73, 12; iv, 25, 23; gaktlbliih, R. vi, 71, 14. For a like 
reason, but to avoid a final minor Ionic, we find pagyate raja, 
R. vii, 32, 25 ; draksyase tatra, ib. 34, 1 0, etc. 

Less generally have been recognized irregularities due to 
vipulas. But here too Sanskrit grammar yields to the decided 
tendency to have an iambus or diiambus precede in three of 
the four forms and also to less marked tendencies. Even the 
pathya shows similar cases, though in this foot more latitude 
is allowed. But there often is, for example, in the pathya a 
decided preference for the opening ^ ^ rather than 

xiw , and in accordance with this we find arditah sma 

bhrgaih Rama, in R. iii, 10, 11, and agatiih sma, ib. 15, 2; 
where sma must be for small (in some cases this is doubtful). 1 
Of the vipulas, the third is naturally chiefly affected. In the 
last passage, for example, gl. 19, we read iha vatsyama Sau- 
mitre, w'hich is changed as certainly for metrical reasons as 
are the similar cases in the diiambic ending. So in R. ii, 17, 
10; 40, 22, etc. So, too, loss of augment in sa pravigya ca 
pagyacl vai ; the participle in -ant, tatha rudanthii Kausalyam, 
R. ii, 40, 44; duhkhany asahati devl, R. ii, 12, 89; kacic 
cintayati tatra, R. vii, 24, 11 (as opposed to sa cintayanti 
bucldhya ’tha, Nala, 5, 12) ; and shortening of a long vowel, 
sapatnivrddliau ya me tvam, R. ii, 8, 26 ; pitur inguclTpanya- 

1 In upasanta maharajam, iv, 18, 16, the form is chosen not from any 

aversion to ^ w w, but for variety, because this foot precedes in the same 

jloka. In R. i, 4, 4, agrhitam (“ Vedic ”) is merely an error. 



kam, R. ii, 104, 8 ; so hnaravatisamkagam, R. vii, 33, 4. The 
commonest form here is the sma just referred to : pitrmatyah 
sma bhadram te ; krtapunyah sma bhadram te, R. i, 33, 3 ; ii, 
55, 12. So, adharmam vidma Kakutstha asmin, R. vii, 63, 2. 
Offensive is the heavy third vipula preceded by a succession 
of heavy syllables, and so we find : aho trptah sma bhadram 
te, R. i, 14, 17 ; nunam praptah sma sambhedam, R. ii, 54, 6 ; 
vyaktam praptah sma tarn decam, ib. 93, 7. 1 

The Mahiibharata is not so strict in its vipula regulation, 
but even here we find the same condition of things, though in 
less careful observance. Thus, tvayy adlnnah sma riijendra, 
v, 8, 22 ; tvadadhlnah sma rajendra, xv, 3, 54 ; upagiksama te 
vrttam, xii, 16, 2 ; 2 ihai ’va vasatT bhadre, Nala, 13, 66. Both 
texts, merely in accordance with the vipula rule or predilec- 
tion, have kirn main vilapatim ekiun in Nala, 12, 55, and 91, 
which modern editors, sure of grammar but ignorant of metre, 
change to vilapantlm (compare R. iv, 20, 22, kim mam evam 
pralapatlm) ; evam vilapatim dlnam, vii, 78, 36. Other 
examples are tato rudantlm tam drstvii, Nala, 16, 33 (as in 
R. vii, 80, 18, arajii ’pi rudanti sa, to avoid the Ionic ; but 
visnmjnakalpam rudatlm, vii, 78, 39, etc.) ; mam anusmaratl 
gete, viii, 44, 17 ; paitim anvesatim ekain, Nala, 12, 34. Most 
participial changes of this sort not due to the diiambus 
(avoided or sought) are due here as in the Ramayana to the 
natural disinclination to heap up long syllables and the grad- 

1 About half the cases of sma for smah are due to metre. This word 
before sonants on account of its monosyllable would lose its character, and for 
this reason most of the cases not due to metre are before sonants to avoid 
smo. Of all the cases in Bohtlingk’s list only two are before surds. At the 
pada-end, where length is indifferent, sma stands only before sonants. With 
the exception of sma, in the first four books of the Ramayana (according to 
Bdhtlingk’s list) the only examples of ma for mah which appear to be inde- 
pendent of metre are vidma purvam and praveksyama at the beginning of 
posterior padas. The first is not in G. ; the second appears in G. as veksi/ami. 

I may add of sma, as indicative of the pseudo-epic, that the thirteenth book 
has three forms of this word, smah, sma (perhaps dialectic), and smahe. The 
last, a modern form, is found not only in xiii, 1, 13, but in 93, 41, na smahe 
mandavijnana na smahe mandabuddhayah . . . pratibuddha sma jagrma. 

2 Holtzmann, at § 548 ; but I should not entertain the notion that any of 
these forms (a3 here suggested) was other than indicative. 



ual creation of the iambic rale for the third vipula. 1 The 
change to anti, illustrated by musnantl and kurvanti in Nala, 
5, 8, and 16, 11, respectively, and ayanti, R. vii, 26, 47 ; 96, 
11 , etc., is in part explained by preferred combinations and in 
part by analogy, the great mass of verbs making the form 
anti. The best case of change for metre is furnished, how- 
ever, by the tristubh in ii, 67, 53 (^ _ w w begins a tristubh 
only before m w ) : 

tatka bruvantliii karunam rudantlm 2 

The first vipula is responsible for the form upasanta in ix, 
38, 53, tvam upasanta varadam ; the second, for aho mudkah 
sma suciram, xiii, 16, 27 ; the third, for vicarisyama loke 'srnin, 
viii, 33, 12. For the fourth I have no sure case. 

In regard to the augment, it is omitted so freely that only 
in pronounced cases are we sure that it is dropped for metre, 
especially as the endings ta and tha are interchanged (as they 
are in the later Upanishads). Thus in R. iv, 53, 8 kim na 
budliyata may be present, or, as the commentator says, stand 
for nabudhyata (diiambus) ; but again there appears to be no 
reason for samantat paridhiivata in R. vii, 28, 17, for the aug- 
mented form would serve as well. But in this category, 
besides the influence of patois, we have a more than usual 
source of pseudo-archaisms. For in many other cases we can 
but assume that copyists have tampered with the text, cor- 
recting after their wont, sometimes for grammar and some- 
times for metre, according to their individual taste ; a process 
that explains in our printed texts the frequent divergences 
that depend on these points. 3 But with the augment it is 
especially easy to give an archaic effect, since, while Sanskrit 

1 In Holtzmann’s list, for example, the only case of ati for anti that does 
not come under these rules is caratl in Nala, 12, 10 ; which may be attracted 
by anvesatl in the same verse (the latter caused by the diiambic rule). 

2 Holtzmann registers rudantl for i, 6, 5, where B. has rudatl ; and for Nala, 
17, 12, but B. has rudatyau. 

3 For this reason I have elsewhere called them “ unguarded texts,” mean- 
ing of course that they were not protected, as were the poems of sacred 
character, by artificial methods of transmission. 



kept the augment, most of the other forms dealt with are 
current side-forms as well as antique. So we find, for ex- 
ample, in R. vii, 28, 26, nanavadyani vadyanta, but in G. 
36, 26, °ny avadyanta, and here, as in parallel cases, it is quite 
impossible to say whether we have a grammatically emended 
text or a mere imitation of the antique on the part of a 

Instances of alteration in tristubh verse are of the same 
sort as those just mentioned and need not be specifically 
detailed. Here too we find the same imitation of the antique. 
One example will illustrate both cases. In xiii, 102, 55 a — b, 
occurs, buclhyami tvam Vrtrahanam qatakratum, vyatikra- 
mantam bhuvanani vigva. 1 Compare also na ca ’pi janima 
tave ’ha natham, iii, 265, 4 d ; na ’bhutikiilesu phalam da- 
dan ti, xii, 25, 7 a ; and the following examples : 

na tam vaded usatlm papalokyam, xii, 300, 8 d 
prayama sarve caranam bliavantam, i, 197, 4 d 
Kartiam bi&Aeduh sahitali prsatkaih, viii, 82, 16 c 

jahara papas tarunliii vicestatim, R. iii, 53, 26 c 
apaqyatl Raghava-Laksmanav ubhau, R. iii, 52, 44 c 
hatah sma sarvah saha mantribhiq ca, R. ii, 61, 26 b 

Here, as will be seen from the structure of the tristubh, the 
cases of grammatical irregularity are of the same type and 
character as those in qloka. The prevailing type, namely, is 
the patois substitution of ma for mas as verbal ending, and 
the alternate participial form. The change here also, as in 
qloka, induces a preferred or “ regular ” form against a more 
unusual, more disliked, or more irregular form. The last ex- 
ample above, for example, gives a cadence common to both 
epics ; but to have smah for sma would be a cadence of the 
Mahabharata, not of the Ramayana. 

To sum up for the cloka : In the occasional modification of 
accepted Sanskrit forms purely for the sake of metre and in 
the lack of a thorough observance of metrical laws, which have 

1 This form occurs also in i, 3, 57 bhuvanani vifva ; and vii, 201, 77, 
bhuvanani ’ha vifva, in the same formula. Generally sapta takes its place. 



yet obviously affected certain parts of the epic, we can see the 
rules themselves in process of making. For the greater part 
of the Bharata there is no fixed rule, but the foundation of the 
rule is there in popular liking and dislike. Thus cases do ex- 
ist, and they are not infrequent, of h w before a second 

vipula, but there is a decided tendency against such a combi- 
nation, and as a result we find bhaksayisyava sahitau, i, 152, 13 ; 
to explain which we need only say that the first vipula favors, 
while the second does not favor, this precedent foot; just as 
ib. 154, 35, Qlghram gacehama bhadraiii te is merely a present 
indicative with a preterite (patois) ending, substituted because 
the Sanskrit ending would oppose a metrical combination to 
winch there is a growing though not yet thoroughgoing 

Finally, as already abundantly illustrated, the statement 
that “ the laws of the gloka are the same in the Ramayana, the 
Mahabharata and the classical poets ” 1 is certainly much too 
strong. What is quite fixed in the last is not so rigid in 
the first, and is much looser in the Bharata than in either of 
the other two. 2 

The Hypermetric Cloka . 3 

A ninth syllable is often attached to the octosyllabic prior 
qloka pada, regularly prefixed, sporadically incorporated; the 
hypermetric syllable in the former case being, with the next 
also, a brevis, while the third is long before an iambus, the 
whole foot preceding a pathya or any vipula, thus : — 

1 Das Ramayana, 1893, p. 21. 

2 It is indeed enough if the vipula be preceded by a heavy syllable or long 
vowel, as has justly been remarked by Jacobi, in his article Ueber den Qloka 
im Mahabharata, but this rule does not mark the distinction between prece- 
dent iambs and spondees. The rule is to have a precedent iamb, and a spondee 
is always exceptional ; but in R. it is a very rare exception ; in Mbh. a very 
common exception. 

3 Analogous to the freedom in tristubhs we might expect to find also cases 
of catalectic, or more properly abridged, gloka-padas, such as, e. g., purap cakre 
dvipadah, BAU. ii, 5, 18 (cakara?); but I have not noticed any such epic 



pathya : 

anubhuyatam ayam vlrah, 1ST ala, 2, 9 
first vipula : 

prakrtir gunan vikurute, xii, 314, 15 
second vipula : 

katliam Arstiseno bhagavan, ix, 40, 1 
third vipula : 

navanltapankah ksirodah, xiii, 80, 6 
fourth vipula : 

caranfigataiii na tyajeyam, v, 12, 16 

The regular hypermeter thus coincides in its opening with 

the irregular and unusual octosyllabic pada, ^ w _ w 

For instance, akrtavranah culihair vakyaili, v, 184, 14, is 
hypermetric, while apakarinam mam viddlii, xiii, 96, 7, is an 
acatalectic pada ; for which reason, probably, the latter is so 

Such hypermeters are not unusual in the Mahablmrata and 
Ramayana, though more frequent in the former, not only on 
account of the mass, but in the same amount of matter. They 
seem to be at times rather affected by the later epic poets ; 
perhaps to give an appearance of antiquity, whereby, as often, 
the effect is overdone. I know at least of no passage in either 
epic where, as in Harivaiiga, 1, 3, 54, and 87, and 91, and 108, 
four hypermeters can lx; found in the space of fifty odd glokas. 
They are common too in the Puranas. 

Certain phrases are apt to appear in this form. The com- 
monest is abhivadayanti or some similar derivative, wliich 
often introduces hypermeters in glokas (as also in tristubhs). 
Thus, for example : 

abhivadayanti bhavatlm, v, 90, 98 
abhivadayanti vrddhahc ca, v, 47, 16 
abhivadaye tvam rajendra, iii, 291. 37 
abhivadaye tvam bhagavan, iii, 207, 13 and R, iii, 

11, 72 

abhivaditah kaniyobhih, iii, 257, 8 
abhivadya cai ’nam vidhivat, v, 179, 13 



abhivadayitva cirasa, v, 176, 28 
abbivadayita vrddhano ca, xiii, 104, 65 
abhivadayisye hrste ’ti, xiv, 68, 19 
abhivadayamas tvam sarvah, R. vii, 49, 15 

Although avamanyase mam nrpate, v, 189, 22, might sug- 
gest the possibility of pronouncing omanyase, and abhiva- 
denti in the examples above, yet this explanation is almost 
excluded by the fact that parallel examples, in overwhelming 
majority, admit of no such solution. Many of the cases have 
been collected by Gildermeister in his excellent article in the 
fifth volume of the Zeitsehrift fur die Kunde des Morgen- 
landes, p. 269. 1 It is easy to add many parallel examples. 
Thus abhisektukamas tarn raja, G. ii, 74, 55, is a parallel to 
abhisektukamam nrpatim, Mbh. i, 85, 19, and garanagatam 
is an opening used repeatedly, e. g., v, 178, 9; viii, 90, 112; 
xiii, 32, 2 and 34 (but in 38 b, caranagatasaksanam). 2 Some 
difference of texts is to be noticed. Thus in xiii, 93, 119, 
garanagatam hantu sa vai, C. omits vai, an impossible pada. 
On the other hand, in xiii, 94, 27, anrtiiu vratl jatl cai ’va, 
of C. 4,573 is converted into anrtau ca vratl cai ’va. So in 
G. v, 63, 2, abhayam dadami te vlra; but in B., abhayam 
te pradasyami. The commonest words thus employed, owing 
perhaps merely to opportunity, are abhivadayanti, or an equiv- 
alent, garanagata 0 , and Janamejaya. Those mentioned by 
Benfey, in the notes to his Chrestomatliie, are chiefly of the 
same character, but he also adduces long initials, of which 
I shall speak presently. Although, as shown above, any 
form of vipula or a patliya, may contain the hypermetric pada, 
and the fourth vipula is very common, yet the patliya is the 
usual place for it, so that the last may be regarded as itself 
the patliya or regular form of this irregularity. 

Besides the cases noticed by others, to which references 
will be found loc. cit., Janamejaya, abhisaryamanam, aditir 

1 Compare also Jacobi, Das Ramayana, p. 24 anil in the Gurupujakaumudi. 

2 In v, 12, 15, and 16 (cited above), ?aranagata ’smi te brahman, and <;arana- 
gatam na tyajeyam, respectively. But in v, 15, 33, jaranam tvam prapanno 



ditili, balavat sapat-, upajivanam, vrsallpatih, purusam tv 
idiinlm, arunodaye, tam ahaiii smayann iva rane (one of the 
repeated phrases, v, 179, 22, etc.), atithivratl (also repeated, 
iii, 260, 4, etc.), akrtavranaprabhrtayah (repeated opening, 
v, 180, 17, etc.), 1 and a few more hitherto cited, I add with 
references : 

aparajito jyotikaq ca, i, 35, 13; upaglyamuna narlbhih, etc., ii, 
58, 36 (iii, 158, 83; vii, 82, 28); kapilavatam, iii, 84, 31; (kapi- 
lasya goh, xii, 269, 5) ; bhagavan anekaqah, iii, 99, 39 ; 188, 9; 
viyunajmi dehat, iii, 142, 26; paricarakesu, iii, 200, 9 ; amitaujase, 
v, 4, 12 ; Sumanomukho Dadhimukhah, v, 103, 12 (in i, 35, 8, as 
Sumanakhyo Dadhimukhah) ; krtakilbisah, v, 165, 22 ; purusah 
sanatanamayah, vi, 21, 14 = 773, v. 1. ; 2 * madanugrahaya para- 
mam, vi, 35, 1; avamanyamano yan yati, vii, 73, 30; arunam 
Sarasvatlin prapya, ix, 5, 51 ; Garudananah kafikamukhah, ix, 
45, 83; madadhisthitatvat samare, ix, 62, 18; Cakune vayaiii 
sma deva vai, xii, 300, 4; avyaktarupo bhagavan catadha ca 
sahasradha, qatadha sahasradha cai ’va tatha qatasahasradha, xii, 
315, 2; tadanantaraiii ca Eudrasya, xii, 319, 62; arani mamantha 
brahmarsih, xii, 325, 9; Uqaua Brhaspatic cai ’va, xii, 336, 45; 
ayajad dhariih surapatim, xii, 338, 30 ; paramanubhuta bhutva tu, 
xii, 345, 15; sahasa jagrhatur vedan, xii, 348, 29; tridaqas tri- 
kaladhrk karma, xiii, 17, 62 ; animantrito na gaccheta, xiii, 104, 
143; Yiduradayaq ca, xv, 3, 76; atavibalam, xv, 7, 7; Upada- 
navi sutahl lebhe, H. i, 32, 8 ; asatlm Vapustamam etam, H. 3, 5, 
21; dhvajinah patakinaq cai 7 va, R. v, 4, 20; Amaravatim sama- 
sadya, R. vii, 5, 26; Yamalarjuuau, R. vii, 6, 35; Krtavan Pra- 
cetasas putrah, E. vii, 111, 11. 

It will be observed that Yamalarjuniiu and Amaravatim 
(these Ramayana passages have already been cited by Jacobi) 
are exactly of the same type as are dhvajinah patakinah, 
abhivadaye, and abliisektukamah, though the first two occur 
together in a late addition to the epic and the other three 
examples are in the body of the work. As the type per se 

1 These are complementary references. 

2 Ends, yatah Krsnas tato jayah, variant on the older phrase, just preced- 

ing, yato dharmas tato jayah. 



is old (Upanishads), 1 the occurrence of hypermeters denotes 
rather lack of refinement than lack of antiquity, so that the 
phenomena as a class stand parallel to the care or careless- 
ness in the making of vipulas. 

When on two short syllables a third short follows, the 
phrase is rudely adapted to metrical needs. Hence aho 
manyata for ahar amanyata in R. iv, 35, 7. 2 Some excep- 
tions occur to mar the uniformity of the phenomena, but for 
the most part they are in words or phrases which are forced 
upon the poets and which they have to handle as best they 
can. So we find a variant on the daga proverbs 3 4 in the 
form dacacrotriyasamo raja ity evam Manur abravlt, i, 41, 31, 
where there are two departures from the norm and the verse 

is a hypermetric form of the pathyii w. _ w w w 4 A 

similar case occurs in It. iii, 35, 9, where we find dagagrivo 
vingatibhujah. Here I can scarcely agree with Professor 
Jacobi in regarding daga as monosyllabic (Ram., p. 24). So 
in the case of Dagakandhara-riijasunvoh, cited by the same 
author (in Gurupuj, p. 52) from iii, 290, 19, which is like 
pratibodhaviditam matam, simply hypermetric but answering 

to the type si _ w w _ w (not to be read as Dagakand- 

hara, as Jacobi suggests). Either tins or the explanation 
offered below of suppressed a seems to me most probable. 

Hypermeters with long initial syllable are sometimes found. 
They are of two sorts and should be carefully distinguished. 
The first is where the pada corresponds exactly to those just 
discussed save that a long syllable takes the place of the first 
brevis. So far as I know, this occurs only in the later epic 
portions (also Puranic). It is a clumsy or careless form 
which, induced generally by proper names, regards only the 
mechanically counted syllables and entirely disregards the 

1 For example, pratibodhaviditam matam, Kena, ii, 4; abhayam titlrsatam 
param Katha, iii, 2. Gildermeister, loc. cit., p. 275. 

2 Compare Bohtlingk, loc. cit., p. 214 ad fin. So puno pi, Gatlia and Pali. 

3 Compare xii, 108, 16, dayai ’va tu sada ’caryah yrotriyan atiricyate ; xiii, 
105, 14, daya ’caryan upadhyayah. 

4 The partial parallel, uttarayanam from Manu vi, 10, cited by Gilder- 
meister, loc. cit., p. 272, is a later text for turayanam (see Jolly’s text). 



essence of the hypermetric light dissyllable. This consists in 
a mora measurement of two breves, or light syllables, as a 
substitute for one long vowel or heavy syllable, which is im- 
possible in padas that have such initials as 

Ekata-Dvita-Tritflq co ’cuh, xii, 337, 20 

Aqvamedhikam samasadya, xviii, 278, corrected in 
B. 6, 69 to asadya. 1 

hTaimisaranye kulapatih, H. 1, 1, 4 (C. 11) 

daksinayanam smrta ratrih, H. 1, 8, 9 e 2 

Where a short vowel follows (as in other parallel cases 
mentioned hereafter) it is practically suppressed. So asthiny 
antarato daruni, BAU. iii, 9, 28 (asthiny antar ’to) 3 and in 
the epic: 

paksivanararutajnaiq ca, i, 70, 45 (van ’ra), 

or the two breves must be read as a morarequivalent. It is a 
mark of the popular style, as in Agni Parana, iii, 11, bibhrata 
kamandalam purnam ; ib. x, 28, brahmana Dagarathena tvam. 
Prefixed extra metrum is aum in xii, 348, 38, aum, namas te 
brahmahrdaya, and elsewhere. 

The cases of long initial cited from the older epic are of 
quite different character from the form with initial long. 
The supposed parallel from Manu vi, 10, adduced by Gilder- 
meister, and cited above, being removed in the revised text, 
there remain only a few padas of entirely different formation. 
Instead of having a long syllable prefixed they follow a dis- 
tinct type of tristubh. The pada does not begin with a long 
syllable and then continue with a short, but begins with two 
long vowels or heavy syllables, or a short followed by a long : 

(a) retodliah putra unnayati, i, 74, 111 ; H. 1, 32, 12 

(b) Bhlsmo vasunam anyatamah, v, 185, 18 

(c) craddharn pitrbhyo na dadati, 4 v, 33, 35 

1 Compare Amaravatlm samasadya, v. 1. asadya, R. vii, 5, 26. 

2 In Manu i, 67, ratrih syad daksinayanam. Compare the similar “ Pur- 
anic ” verse, daksinena ’ryamnah panthanam, cited above, p. 6, note 2. 

3 Compare the subsequent padas : retasa iti mi vocata : dhanaruha iva vai 
vrksah, though here we may read a(h) + i =: e, as also occasionally in epic 

4 Cited by Gildermeister, loc. cit., p. 273. 




One case (cited like these by Jacobi) is found in the later 
Ramayana, vii, 21, 14, 

samtaryamanan Vaitaramm 

with the first syllable short and second long, e. g., v, 43, 11, 
(d) katham samrddham asamrddham 

It will be noticed that the caesura is after the fifth syllable. 
The forms in the corresponding (a, b, c, d) tristubh padas, 
where the initial length is indifferent, may be illustrated by : 

(a) na cen mam Jisnur | ahvayita sabhayam 

(b) amantraye tvam | brtihi jay am rape me 

(c) yasya ’vibhaktaiii | vasu rajan sahayaih 

(d) samanam murdhni | rathayanaiii viyanti 

Many cases of these forms will be shown in the next section 
on tristubhs. The two formations are evidently identical ; but 
what occurs passim in the tristubh is sporadic in the gloka. 
The pada in each case consists of a complex of two metrical 

groups, ^ ^ and _ v ^ _ or ^ w 1 

An extra syllable in the posterior pada is indicative merely 
of late carelessness under the power exerted by names and 
titles which are hard to coerce into normal metrical form ; as 
in the spurious verse cited by Professor Jacobi from R. vi, 
105, 10, Hiranyareta divakarah. Such cases as Pulastyovaca 
rajanam or Laksmanas tu tatoviiea indicate not a precedent 
hypermeter but the looseness of epic sandhi. They are very 

There is, however, a more regular interior hypermeter which 
is old. Thus in Katha Upanishad, vi, 8 and vi, 11, respect- 
ively, we find 

avyaktat tu parah purusah 
apramattas tada bhavati 

1 The references for the tristubh padas will be given below. The pada 
cited from the Mahabhasya, IS. vol. xiii, p. 459, avidvansah pratyabhivade 
is without parallel, I believe, in the epic. The same rule appears in Manu ii, 
123 with abhivada, which may have stood here originally, unless abhi was 



It was suggested by Gildermeister, loc. eit. p. 274, that in 
such instances in the epic, bhavati might be read as two syl- 
lables, but he seems inclined to reject the notion. Professor 
Jacobi, on the other hand, favors this reading, and says of 
such cases, “All is in order if one pronounces bhavati as 
bhoti ” (Gurupuj., p. 52). But he is forced to add immedi- 
ately, “It is more difficult to decide how one could have 
managed with kimsvit suptam na nimisati and katham sam- 
rddharn asamrddliam.” 

The explanation lies, I think, in the fact that mora- 
measurement was at work in syllabic verse. This is very clear 
in tristubh; in fact, it is the only possible explanation for a 
mass of forms which from a syllabic point of view are wildly 
irregular but with this admission of mora-measurement are 
easily understood. The (jloka cases are generally found at 
the end of padas, where cajsura aids the reading of two breves 
as equivalent to one long. In the case of bhavati itself and 
a few similar forms, where we know that bhoti or hoti is a 
dialectic equivalent, there is, to be sure, no great objection to 
reading bhavati as bhoti, but the general explanation of the 
phenomena as a class is not that uuis contracted, for some 
of the intervening consonants would make this impossible, 
but measured as the metrical equivalent of one long. In the 
examples above bhavati and purusah and nimisati are thus 
parallel cases. In Katha iii, 5-6, both padas are hypermetric : 

yas tv avijnanavan bhavaty 
ayuktena manasa sada 
yas tu vijnanavan bhavati 
yuktena manasa sada 

I see no reason to separate these cases from their epic ana- 
logues. 1 Here we have the oft-cited examples of prior padas 
ending in -triyo bhavati, priyo bhavati. nivartayitum, unnayati, 
iii, 313, 45-48. 2 In the cases cited above from this passage, 

1 For more examples from the Upanishads, compare Gildermeister, loc. 
cit., p. 275, ff. 

2 The irregular use of svit in this passage probably explains the impossible 
pada, kena [arid] dvitiyavan bhavati, ib. 47. In the following question, svit 



813, 61, and from v, 43, 11, the same principle is extended, 
exactly as we shall see it in tristubh verse, where the second 
foot after the first dipody, t=t _ m may be resolved from 

\j into KP wuH. So here, kirn svit suptarn na nimisati 

may be on the tristubh model, w, wwww, which 

passes into and appears as , yy w w w, as in the 

tristubh, v, 16, 5, prapte kale pacasi punah samidclhah, tvam 
eva ’gne bhavasi punah pratistha. So we shall find labhate 
in a tristubh, where it must be equal to ^ just as in the 
gloka of the Dhammapada, No. 131, we find pecca so na 
labhate sukham, where the two breves must be measured as 
one long (so the MSS., but changed in the new text), but is 
not contracted (compare in prior, prajapatic carasi garbhe, 
Pragna ii, 7 ; grig ca prajnaih ca vidliehi nah, ib. 13). 

A very interesting phase of this question is the relation of 
the Sanskrit to the Pali. We have a proverb in R. ii, 
103, 30, 

yadannah puruso bhavati tadannas tasya devatah, 

which Professor Lanman at the Meeting of the Oriental Society 
in 1899 argued was from the Pali form because there hoti 
actually occurs in the same proverb. 1 But against the cer- 
tainty (though not the probability) of this conclusion stand 
the facts that the form of the verb is undetermined in Pali 
and the hypermeter of this sort is just as common there as in 
Sanskrit. It is clear, for example, that in such verses as na 
tena bliikkhu hoti, Dhammapada 266, must be read (as the 
text now stands) bhavati (compare tatrayam acli bhavati, sic, 
in 375, and in other verses of the same collection) ; while on 
the other hand, in 387, sannaddho khattiyo tapati (= tap’ti) 
stands parallel to similar uneontractile forms in Sanskrit gloka 

is omitted, as it should he here. The other cases are all parallel to kena svic 
chrotriyo bhavati, frutena frotriyo bhavati, 47-48. 

1 Since publishing an article on the Parallel Proverbs of the two epics in 
A. J. Phil., vol. xx, p. 22, if., I have found a parallel to this yadannah proverb 
in the Mahabharata, viz. yadanna hinara rajans tadannas tasya devatah, where 
tasya is still preserved though the plural noun precedes! It is (of course) 
from the careless pseudo-epic, xiii, 60, 61. 



and tristubk forms. There is then no real necessity for 
changing the latter to khatyo (a possible form.) 

Nevertheless, in the case of bhavati itself, which like bhos 
may have been current as bhoti in Sanskrit as well as in 
dialectic form, the latter may have been used, and a dual 
pronunciation may be accepted and given as a probable reason 
for its frequent recurrence in apparent hypermeters. 1 In 
other words, padas with this word may possibly not be true 
hypermeters, as must be other forms which are not thus con- 
tracted or contractile. That a hoti in Pali may stand for 
an original bhavati, may be seen by comparing Dhammap. 
260 with Mbh. iii, 133, 11: 

na tena thero hoti [bhavati] yen’ assa phalitam siro 
na tena sthaviro bhavati yena ’sya palitaiii qirah 

Compare Manu ii, 156, na tena vrddho bhavati (v. 1. sthaviro 
in some of the commentators). Another of these numerous 
bhavati proverbs is found in Dhammap. 268, na monena muni 
hoti, Mbh. v, 43, 60, rnaunan na sa munir bhavati. 2 

Dialectic Sanskrit. 

Accepting bhoti (= hoti) as a possible dialectic Sanskrit 
form, I have next to show that the masa for masa principle, as 
illustrated in the paragraph above, is subject to an important 
restriction. It would be quite wrong to suppose that the 
mass of grammatical irregularities are of a form entirely 
arbitrary, or that, in general, a grammatical modification that 
is found repeatedly in one category may be utilized for 
metrical purposes in any other of the same outer appearance. 
I say in general, because I admit that here and there in the 
epic occur grammatical monstrosities and forms not subject to 
metre, though irregular, but what is of moment is that most of 
the grammatical irregularities in the epic are merely dialectic 

1 Thus xii, 233, 12, fariram frayanad bhaTati, murtimat sodaijatmakam, 
and often. 

2 On the variant to the yadannah proverb contained in the words yaccittas 
tanmayo bhavati, see p. 42. 



variations. For this reason in the paragraph above, headed 
Poetic Licence, I have been careful to state that the modi- 
fications were those of Sanskrit forms, not that they were 
absolute alterations of received forms, independent of any 
grammatical basis. I believe the latter cases to be exces- 
sively rare, while on the contrary there is some sort of gram- 
matical authority for most of the changes so abundantly 
introduced. Metre surpasses Sanskrit gr amm ar but not 
grammar altogether. What then? Where Sanskrit gram- 
mar fails, the poets had recourse to patois. 1 

As I have already shown, a large majority of the cases 
under consideration are comprised under the head of feminine 
participles and first plurals of verbs, with a smaller number of 
various forms. 2 

Some of these, like brumi, are at once dialectic and yet 
accepted as Sanskrit. There is no reason why we should not 
regard kumii, Gatha kurumi, in tatliii kurmi and kim kurmi ’ti 
krtafijalih, iii, 142, 44; II. 3, 14, 12, as on a par with brumi. 
The latter occurs not only in R. vi, 9, 20 (where G. reads 
bravlmi, v, 80, 22), but also in R. ii, 19, 4; iii, 13, 17 ; iv, 7, 
14. So R. ii, 12, 36, anjaliih kurmi; vii, 78, 20, aharaih gar- 
hitarii kurmi. So too vedmi and dadmi, e. g., Ii. ii, 53, 21 ; 
vi, 124, 17, aham apy atra te dadmi, which in the later Bharata 
is more and more frequent. Others appear to be gross violar 
tions of grammar, like °nati and vidusah, nominative, as in 
parallel forms, tasthusam purusam, xii, 317, 17, etc., 3 but they 
may be not only Vedic but dialectic, as Pali °ati and vidu 
(= vidvan) may imply. Doubtless some are pure archaisms, 

1 So far as T know, this important subject has only been touched upon in 
a note by Kielhorn, JRAS., 1898, p, 18, who says : “ In the so-called epic 
Sanskrit there are not a few forms and constructions which seem to me to be 
Pali rather than Sanskrit.” 

2 Lengthening of a vowel metri gratia is called arsam almost invariably 
by the commentators. Some of the cases are really archaic ; others are 
clearly a sacrifice of form to metre, generally for the diiambus, as in R. v, 
36, 21, sukhanam ucito nityam asukhanam anucitah. 

3 To Prof. Holtzmann’s list I add (the reduplicated forms, § 803) tasthusi, 
x, 8, 70, and nedusam (apsarasam), ix, 57, 68. 



as in vigva, lack of augment, va for iva, and varying final vowel 
length (atha pari, na, etc.) ; but when we consider that the 
participle is indifferently bhavatl and bhavanti, and that the 
first plural verb ends regularly in via in all forms, 1 that, for 
instance, asma is regular, we shall hesitate to speak of any 
general grammar-sacrifice save that of Sanskrit. Thus kra- 
mati (for kram) is Prakrit. 2 In the older epic, arbitrary 
changes were not introduced at will, but dialectic forms were 
borrowed. Even upasante for upasate (compare the older 
hinsate for hihste, R. iv, 53, 16) is merely a dialectic change of 
conjugation, just as is the case with the forms dadanti, 
jahanti (compare Dhammap., hinsati and dadanti, okam okarii 
jahanti te, etc.). These forms, it is important to observe, can- 
not be explained on the assumption that epic Sanskrit precedes 
the differentiation of correct (Sanskrit) and vulgar (Prakrit) 
forms, because, were that the case, they would appear passim ; 
whereas they appear usually, as in svapami for svapimi and 
grhya for grhltva (cited above, pp. 205, 247), only when 
the metre requires them. Take, for instance, the clear case 
of patois, geha for grlia. It occurs in iii, 69 (Mala 17), 
15-16 to prevent a diiambus at the end of a prior piida (though 
grlia is used in the preceding verse) ; again at v, 36, 34, to 
prevent the minor Ionic; in ii, 68, 1, to prevent a third vipula 
from following a brevis, bhavanti gelie bandhakyah; in iii, 
303, 13, to prevent an anapiest, mama gelie mayii cii ’sya 
(for the same reason in R. vii, 68, 20) ; in xii, 336, 25, to 
avoid triiambus in an even pada. Dialectic are further, in all 
probability, the exchange of weak and strong perfect forms 

1 The change is not really grammatical but phonetic, as Dr. Thorp has 
shown, since the preterite is not used for the present but the primary ending 
is reduced from mas to ma (and may be contracted, as in na janime 'ty atlia 
’bruvan, v, 120, 21). 

2 l'ischel, Grammatik der Prakrit Sprachen, § 481. For svapami, compare 
ib., § 407 ; for asiya as na syat, § 404 ; for neuter instead of masc., § 357. 
Professor Pischel’s mine of wealth came to hand only after this book had 
gone to press, or I could have given a more systematic as well as fuller treat- 
ment of a comparison based chiefly on Sanskrit and Pali, and such few dia- 
lectic forms as chance furnished. But I think the more the epic is studied 
the more Prakrit will be found. 



and perfects without reduplication, when needed for metre, 
akarsatuh, i, 153, 44 ; bibheduh, viii, 82, 16 (to avoid a brevis 
before a second vipula) ; the exchange of nominative and 
accusative, ausadhayah (ace.), 1 though this is also Vedic. 

But the epic took long in making, and while the earlier 
poets drew on dialectic forms (thereby creating a sort of 
Gatha dialect, though not so gross as the genuine article), the 
later poets did exactly what the later Greek hexameter poets 
did, viz., copied their predecessors instead of borrowing from 
the life. Consequently they made blunders. The early poets, 
for example, used, metri causa, optative for indicative, viii, 89, 
22, and often (as in late Upanishads, e. g., Cvet. v, 5) a vulgar 
confusion ; and ma for mas and dadanti for dadati ; because 
they knew that these were spoken forms, if not the polite forms 
(which they used by preference when convenient) ; but the 
later poetaster knew only that the old epic poets had mixed 
up ma and mas and anti and ati, and so he used the un- 
Sanskrit forms not only more frequently but more incorrectly. 
Thus he said apatjyamas, ix, 1, 20, and did not hesitate to use 
bhavati for bhavanti, of course only in the later epic, as in iii, 
211, 9 (a late chapter, above, p. 34), anyonyam na ’tivartante 
samyak ca bhavati, dvija. Compare the wisdom to be learned 
at Mitliila, in the preceding copy of Yalmiki's proverb, striyo 
liy avadhyah sarvesam ye dharmam abhivindate, iii, 206, 46 
(na hantavya striya iti, vii, 143, 67). So in xiii, 145, 20 
(alpabuddhayah), bubliusate (for diiambus) ; and, in the later 
Ramayana, prajas tarn anuvartate, R. vii, 43, 19 (v. 62, 9, 
interpolated? above, p. 245). 

1 Both in Mhb., pusnamy ausadhayah sarvah, i, 78, 40 ; and R. draksyasy 
osadhayo diptah, vi, 74, 32. Compare sarvah prakrtayah ganaih . . . sam- 
jahara (Jatugrha Parvan) and ib. 145, 4; with R. vi, li2, 19, santvayitva 
prakrtayah. Carelessness in the length of vowels in declension is also a mark 
of patois (epic examples above). The Ramayana has some genders which 
may be dialectic. They certainly are not Sanskrit : parikhan (!) purayantag 
ca, R. vi, 42, 10; ciksipur vividhan gastran (!), R. vi, 53, 20 (both lacking as 
such in pw.), etc. As remarked above, some of this may he scribe’s work. 
Thus yada vedagrutir nasta, xii, 340, 105 ; vedagrutim yatha, G. iv, 5, 4 ; but 
in R. 0, 5, nastam devagrutim (“arsa”) iva. But merely for metre is dosam 
for dosah, R. v, 28, 5 ; G. vi, 33, 30. 



In the careless writing of the pseudo-epic, Sanskrit grammar 
is flung to the winds. I do not mean that irregular forms are 
not found outside of it. Substitution of the ar-conjugation is 
found in adadat, iii, 173, 8 ; 275, 40 ; ix, 51, 10 ; though the last 
is an evident interpolation, and as the forms are not required 
metrically in the other cases it is still open to question 
whether they do not contain just such copy-slips as are found, 
e. g., in the Vayu Purana, where viii, 163 has vyadadhat pra- 
bhuh, while 165 has adadat prabhuh. The cases in the older 
epic are, however, not frequent (in xi, 25, 5, jahati is 3d sg.), 
but in the late epic they flourish like reeds (compare jahanti 
in i, 172, 8; dadanti in xii, 25, 7 ; 341, 16; xiii, 62, 46, etc.), 
and it is just here that new irregularities are found. Thus 
vigvedevan apnoti, xii, 318, 5; vi§vedevebhyah, xiii, 97, 14. 
Even such a syntactical monstrosity as the Gathaism iti viii 
menire vayam (with similar cases there) is not shunned, xii, 
337, 38, to say nothing of the syntactical confusion in a§vi- 
bliyara pataye cai ’va marutam pataye tatha, xii, 341, 103. In 
the thirteenth book, besides kurvanas, xiii, 17, 131, we find 
smalie, xiii, 1, 13 ; 93, 41 ; stam for astam, ib. 98, 7 ; the first 
instance of a finite negative verb, 1 another Gathaism (compare 
ajanehi for ma janaya), afterwards somewhat affected : drey ate 
'di’Qyate ca ’pi, xiii, 14, 160. Here also, another Gathaism, 
the popularized change of the r-declension, apaharta and 
harta (together with Atharva, which, however, is in late Upa- 
nishads, Munch i, 1, epic atliarvaya namah), srastaraya namah, 
ib. 309-310 and 313-314. So etan for etani, xiii, 62, 55. 
Such neologisms go far beyond the current interchange in 
upasante and vilasinyah (acc.), 2 also found here, xiii, 104, 19; 

1 With the infinitive, e. g., xv, 11, 15, na ’datum. The negative finite verb 
(given here in C., and required by the sense) is not recognized in the grammars 
as occurring before the classical period. 

2 In Gita 10, 16 and 19, atmavibhutayah may be nominative. The form as 
acc. can scarcely be a Vedic reversion. The Gita still uses no = na u, and so 
in iii, 34 , 11 : but in xiii, 51, 10, vad etad api no mulyam, no is simply late and 
careless for na. Editors or copyists have tried to change bhavati and acc., 
the text in C. xv, 376 (= 11, 21), but they cannot in xiii, 62 , 30 , and in bhumir 
bhavati bhumidam, it still governs the accusative. 



107, 39, and bring us into the field of slovenly adaptation 
from any source, which characterizes the slipshod Sanskrit of 
later epic and Puranas alike. 

Prose-Poetry Tales. 

In the Verhandlungen dcr Philologenversammlung in Gera , 
1878, attention was called by Professor Windisch to a “pre- 
epic phase of poetry,” consisting of prose narration inter- 
spersed with gathas or verses of popular form which helped 
on the story. One epic tale, which has gone over into later 
verse-form, has been shown by Professor Oldenberg, in his 
article on the old-Indic Akhyana, 1 to exist in a prototype of 
this kind. Such mingling of prose and verse, as remarked 
by the latter writer, is found in the epic itself, in i, 3. There 
is also, though not of epic content, a kind of rhythmic prose 
which is half metrical, as hi xii, 190, 5 ff. : tatra yat satyam 
sa dharmo, yo dliarmah sa prakiigo, yah prakagas tat sukham 
iti . . . yat tamas tad duhkham iti, atro ’cyate (three glokas) ; 
tat khalu dvividliam sukham ucyate (. . . to 13) : susukhah 
pavanah svarge, gandhag ca surabhis tatha, etc. Here the 
epic Upanishad glides in and out of metre, the last verse be- 
fore the resumption of gloka being again metrical, in a form 
of tristubh found elsewhere in the epic: na cal ’te dosah 
svarge pradur bhavanti. 

The next chapters to this have alternate prose and glokas, 
the latter appearing either, as at the end of 191, without warn- 
ing, or introduced with the words “ there ’s a stanza about 
that,” bhavati ca ’tra glokah. In 192, one unannounced gloka 
follows the introductory prose, then more prose, and with the 
words bhavanti ca ’tra glokah follow one gloka and two 
tristubhs. 2 after which glokas are again resumed. 

It hap tie ns that a late poet runs on in tristubhs till he 

1 ZDMG., vol. xxxvii, p. 54 fi. 

2 The gloka here, xii, 7006, is another form of a proverb {liven elsewhere 
in the epic, abhayam sarvabhutehhyo dattva, and may he added to Ppriiche, 
485, 486. Qlokah here scarcely connotes tristubhs (as in the Brahmanas),but 
includes them with the gloka. 



stumbles and ends in prose, xii, 336, 10, after several tristubhs : 
gvetah pumanso gatasarvapiipag caksurmusah papakrtam nara- 
nam, vajrasthikayah samamanonmana divya(n) -vaya(va)ru- 
pah gubhasaropetah, etc., in pure prose. There is, further, a 
good deal of plain prose narration in the first, third, and 
twelfth books and in a hymn in H. 3, 68 (praise by titles). 

But a tale of the prose-verse variety exists complete in the 
story’ of the Frog-girl, iii, 192. In this apparent prose there 
are not only metrical and half-metrical padas and hemistichs, 
such as ramanlyarii saro drstva, but even regular epic padas, 
such as muda paramaya yutah, the latter being indeed a stereo- 
typed epic phrase, as in iii, 256, 20; 295, 16. The verses here, 
as was to be expected, are freer than in the regular epic style. 1 

The tale begins : 

2. atha ’casta Markandeyah (apurvam idaih gruyatam) 

The opening line of C., 13,113, is not in B. From the 
openings in the following tales, parv. 196 and 198, the phrase 
atha ’casta Markandeyah was stereotyped and united with the 
preceding, thus : 

bhuya eva mahabhagyam katbyatam iti abravit 
atha ’casta Markandeyah 

In the present tale the former appears as: bhuya eva briih- 
manamahabhagyaih vaktum arhasl ’ti abravit. 

In the following mixture of prose and metre it is sometimes 
difficult to say whether the rougher metrical parts ought to be 
touched. For instance, at the beginning, Iksvakukulodvaliah 
parthivah Pariksin nama mrgayam agamat may have been 
prosed out of Iksvakukulavardhanah Pariksin nama parthivah 
mrgayam gatavan nrpah, or some such turn. So in the next 
sentence, tarn ekagvena mrgam anusarantam, from tam agvena 
’nusarantam ; while for the ninth stanza or paragraph it would 
be a sin of omission not to note how easy it is to read : atha 

1 In another case, iii, 194, the section begins and ends in prose, but has 
flokas between, the last hemistich of which, before the narration closes in 

prose, has the free measure cited above, p. 244, , w ., 

w w vo , etac chrutva tu Kauravyah f’ibirn pradaksinam krtva. 



kanyam gayantxm ca puspani ca ’vacinvatlm ; apagyad, atha sa 
rajnah samipatah paryakramat ; all with freedom not unknown 
to the epic gloka. But any change would in the first place be 
pure guesswork, and besides why should glokas have become 
prose? Again, these tales are built with prose bricks and 
metrical mortar and it is not strange that the mortar occasion- 
ally runs over the brick. 1 I therefore abstain except in two 
or three cases (in some, as will be seen, where the length of 
prose invites verse) from the temptation to make gloka padas 
out of clauses more or less metrical, and write the story as it 
stands (with prose omissions as indicated below) : 

1-4, Ayodhyayam Iksvakukulodvahah parthivah Parlksin nama 
mrgayam agamat, tarn ekagvena mrgam anusarantam 
mrgo duram apaharat (5, prose) 

6, ramamyam saro drstva 
sagva eva vyagahata 

7, madkuram grtam agrnot 

8, sa grutv£ ’cintayan ne ’ha 
manusyagatini pagyami 

kasya khalv ayarii gttagabda iti. 2 9, atha ’pagyat kanyam para- 
marupadarganlyam puspany avacinvatim gayantlhi ca, atha sa 
rajnah samlpe paryakramat. 10, tam abravid raja 

kasya ’si bhadre ka va tvam (iti) 3 
sa pratyuvaca kanya ’smi (iti) 

1 That is to say, as in the case given in the last note, a more or less regular 
verse may incidentally and accidentally he shaped in prose narration with- 
out its being intended as regular verse, though the poetic style of the en- 
vironment may have induced such prose-poetry subconsciously. As for the 
metaphor above, except as illustrating my meaning very roughly, I cannot 
defend it. On the contrary, as the verse-element in tales was fixed and used 
in many buildings, while the prose was crumbled up and renewed in each new 
edifice built of the same brick, it would not be quite unliistorical to invert it 
and speak of poetic bricks and prose mortar. 

2 Was this : kasya khalu ayam gabdah 1 

3 This or ka ’si kasya kuta? ca tvam is an ordinary epic (verse) formula. 
With the preceding, compare (Sita) kusumany apacinvanti {prior pada), and 
kusumani viclnvati, E. iii, 42, 32 ; 43, 1. 



tam rajo ’vaca arthi tvaya ’ham iti. 1 11, atho ’vaca kanya 

samayena aham tjakya 
tvaya labdhum na anyatha 

iti, raja tam samayam aprcchat, kanyo ’vaca 

no ’dakam me dar^ayitavyam (dargetavyam ?) 

iti, 12, sa raja tam badham ity uktva tam upayeme, 2 krtodvahaq 
ca raja Parlksit kridamano 

muda paramaya yutah 3 

tusnlm samgamya taya saha ’ste. 13, tatas tatrai ’va ’sine 
rajani sena ’nvagacehat(a). 14, sa seno ’pavistam rajanam pari- 
varya ’tisthat, paryacvasta^ ea raja tayai ’va saha cibikaya prayad 
avaghotitaya sva(iii) nagaram anuprapya rahasi taya saha ’ste. 4 
15, tatra ’bhyaqastho 'pi ka^cin na ’pa^yad atha pradhanamatyo 
'bhyaqaearas tasya striyo ‘prechat. 5 16, kim atra prayojanam 
vartate (vartata) ity, atha ’bruvaiis tah striyah. 6 

17, aptirvam idam pacyama 
udakaih na ’tra niyata(e) 

ity, atha ’matyo 'nudakam vanaih karayitvo ’daravrksam, etc. 

18, vanam idam udarakam 7 
sadhv atra ramyatam iti 

1 Perhaps samarthl tvaya bliadre ’ham (compare 33). 

2 More natural would be : sa raja badham ity uktva tam kanyam upayeme 

3 A regular epic phrase in various forms, muda, yriya, pritya, etc., with 
yutah or yuktah, according to the pada. Compare the references above and 
ii, 53, 23; Nala, 20, 40; ix,27, 6; 36,42; pritya paramaya yuktah, ix, 55,4; R. 
i, 52, 11, etc. 

4 The texts give ’nvagacchat and ’nvagacchata, svanagaram and svam 
nagaram. This may point to a corruption. Leaving out the fine palanquin : 
tatas tatrai ’va ’sine (tu 1 ) rajni sena ’nvagacchata sa (tu) seno ’pavistam (ha) 
parivarva atisthata, parya^vastaf ca (sa) raja anuprapya svanagaram rahasy 
aste taya saha. The long stretch of prose favors this. Compare uvaca ca 
taya saha, an epic phrase, e. g., i, 73, 20. 

5 There is no object to the first verb. Was it not : tatra ’bhyaQastho pi 
ka§cin rajanam na apaqyata, atha pradhanaraatyas tu tasya striyah aprcchata ? 

6 The more probable form is vartate kim prayojanam; kim prayojanam is 
a regular epic close of a hemistich. Compare for example, xiii, 93, 81, kasya 
’rthe, kim prayojanam. 

7 Sic, B. ; C., udaram anudakam. 



After this, prose to 23-25, 

kruddho ajnapayamasa (sa raja) . . . 
yatha vrttam nyavedayan 

. . . 27, iti, §lokau ca ’tra bhavatah (28-29). Compare v, 
64, 5, where, although the whole text is in §lokas, one stanza 
is especially mentioned, §lokena ’nena, Kauravya, papraccha 
sa munis tada. 

30, tam evaid vadinam istajanaQokaparltatma raja ’tho ’vaca 

31, na hi ksamyate tan maya 
hanisyamy etan etair duratmabhih, etc. ; prose to 

32, sa tad vakyam upalabhya 
etc., prose to 33. 

In the following I omit references to the intervening prose 
and give the metrical padas in their order : 

33, tam abravld raja taya 
samarthi, 1 sa me diyatam 

34, athai ’nam rajue pita ’dad 2 
abravic ca enam enam 
rajanarii quqrusasve ’ti 3 

35, evam uktva duhitaram 

36, harsena baspakalaya 
vaca * prapatya ’bhipiijya 
mandukarajam abravld 
anugrhito 'smi iti (sc. te, omit iti) 

37, yathagatam agacchat(a) 

1 In C., asmy aham arthi. 

a In C., dadau. Perhaps sa dadau. 

8 Perhaps : abravic ca duhitaram enam rajanam fufrusa, iti. 

4 A stereotyped phrase, either straddling the padas of a verse, Nala, 9, 25 ; 
or in a pada (after one syllable), as in sa, iv, 20, 28; B. ii, 82, 10. Perhaps 
here : sa baspakalaya vaca pranipatya ’bhipujya ca. 



38, atha kasyacit kalasya 1 

tasyam kumaras (te) trayas 
tasya rajnah sambabhuvuh 
Qalo Dalo BalaQ ce ’ti 
tatas tesaiii jyestham Qalam 

samaye pita rajye 'bhisicya 2 tapasi dhrtatma vanaiii jagama, 
prose through 39. In the following Tale of ^ala : 

40, sutarii co ’vaca, tpghram mam 
vahasva [iti], sa tatha uktah 3 4 * 
suto rajanam abravlt 

41, na kriyatam anubandho 
nai ’sa qakyas tvaya mrgo 
'yam grahltum, yady api te 
rathe yuktau vamyau syatam (iti) 
tato "bravld raja sutam 

42, athai ’nam evam bruvanam 

[abravld raja] 

Vamadeva^ramaiit yahi (iti) * 

43, bhagavan, mrgo [me viddhah] palayate 
sambhavayitum arhasi 

[vamyau datum, iti, tam abravld rsir 
dadani te vamyau] 

krtakaryena bhavata 
mamai ’va 6 vamyau niryatyau 
[ksipram iti] 

. . . antahpure asthapayat 

44, atha ’rsiq cintayamasa 
taruno rajaputro ('sti) 
kalyanam pattram asadya 

1 An epic phrase with variations, kasyacit tv atha kalasya, H. 3, 5, 11, etc. 

2 Possibly : pita rajye "bhyasecayat tatah tapasi dhrtatma vanam jagama 
(sa raja) ; or : pita rajye 'bhisicya ca. Both are formulas, as in i, 74, 126 and 
75, 55. 

3 The text has : vahasveti sa tatho ’ktah, perhaps as much of a verse as is 
the form above. As in 36, the iti padas are, I admit, particularly bad. 

4 B. prayahi. 

6 So B. ’ 



ramate na (me) pratiniryatayaty, aho kastam iti (prose to 48, ff. 

Though far from epic verse, this is not exactly prose, 1 
which, though often rhythmical, is not metrical to such an 
extent as this. Further, the actual presence of epic padas 
in the narrative shows beyond question that it is meant to 
be couched more or less in metrical form. Of what sort 
then is this metrical prose? It is, I think, an early form 
of popular verse, older than the present epic gloka, which, 
as I have remarked above, is probably more refined than it 
was when first written and is less free even than the Maha- 
bhasya epic gloka. It is not, however, necessarily antique, 
nor necessarily modern. It is, in short, the instrument of the 
perpetual story-teller, a naive form, running in and out of 
prose like rhymes in fairy tales. 2 

1 Benfey, Panchatantra (translation), vol. i, p. 259, says that with the excep- 
tion of the two flokas (28-29), “ the rest of the narrative is in prose.” 

2 The same tendency to the creation of pada verse (not arranged in floka 
form) may he seen in the prose tale of i, 8, where, besides the regular verses 
in the prose narration, are found such metrical combinations as : 

Janamejaya evam ukto 
devafunya Saramaya . . . 
etasminn antare kaycid 
rsir Dhaumyo nama ’podas . . . 
sa ekam yisyam Arunim 
Pancalvam presayamasa . . . 
sa upadhyayena samdista Arunih, 

the last being a respectable tristubh pada. If, however, this and the tale of 
Suyobhana be regarded (as Benfey says) as pure prose, what difference is 
there between the other parts which will not give any rhythmical cadence 
and such a rhythmical complex as, e. g., ramaniyam saro drstva, sayva eva 
vyagahata, kruddho ajnapayam asa, and yatha vrttam nyavedayan? And 
how does it happen that kasya ’si ’bhadre ka va tvam, and muda paramaya 
yutah and . . . baspakalaya | vaca are actual verses found in the epic ? 
There is a literary product which is neither prose nor poetry, but a middle 
genre, a sort of dog-trot between walking and running, into which a narrator 
may drop without the conscious campu alternation of padya and gadya (poetry 
and prose) found in more precise literature. It is perhaps not extravagant to 
say that beneath the cultured verse of the literati this kind of style may have 
existed for centuries and even have been the foundation of the earliest literary 



The Epic Tristubh 

i. The Regular Tristubh in the Mahabharata 

The rarest forms of the epic tristubhs are those that in 
the corresponding syllables answer to the commonest forms 
of the gloka, namely the pathya and first and third vipulas. 
The commonest forms of tristubh are those that answer to 
the second and fourth vipulas (decadent in the more refined 
§loka) and to the minor Ionic, a form of tjloka almost extinct 
in the later epic style. Both metres have besides the diiam- 
bic and major Ionic forms, but in both they are exceptional. 

Measured by their precedent combinations, the tristubh 
forms thus corresponding to the 9lokas in second and fourth 
vipulas and minor Ionic, outclass the others as decidedly as 
they do in the number of their occurrences; for whereas 
before the tristubh feet corresponding to the pathya and first 

vipula forms stand only ^ _ and ^ , before the 

second and fourth vipula forms stand five, and before the 
minor Ionic form stand seven combinations, respectively. 

In thus grouping the tristubhs 9loka-wise I have wished 
merely to contrast the general structure of this metre with 
that of the 9loka, 1 and have included only the hendekasylla- 
bic tristubh. For the sake of convenience, I shall call regu- 
lar all forms of the eleven-syllable tristubh (pada), however 
unusual, in distinction from other forms, and will now give 
a scheme of these regular tristubh forms (omitting the scolius 
or terminal amphibrach). 2 

product. That any of it has been preserved is a mere accident, not antece- 
dently to be expected. 

1 Of course, as previously explained, the syllaba anceps of the eighth 
syllable must be given up ; but the initial syllable is anceps, as it is in the 
gloka, in the usual forms. 

2 The jagati occurs in the same forms as the tristubh and needs no special 
table (though separately discussed below). Mechanically, it is merely a 
tristubh with an extra syllable added, making the close with diiambus instead 
of amphibrach. 




Combinations of the Regular Epic Tristubh in the Mahabhabata. 

First Foot 

Second Foot of Trigtubh 










































^ V_/ 








8 ? 















— vw 


Eor the abbreviations, compare the table above, p. 236. For \j 

as a second foot in a hypermetric pada, see the paragraph in the list of illus- 
trations in Appendix C, under No. 11. For w w w as second foot, see 

under No. 15. The hypermetrie forms indicated in Appendix C, when refer- 
ences are not given, will be found illustrated in the following paragraphs. 
Tristubhs of catalectic and hypermetric form are not included in this table. 

The Illustrations in Appendix C give a full discussion of 
the occurrences of these forms as they appear in combination 
with the ctesura, now after the fourth now after the fifth 
syllable. Here I will point out that, as is shown by the table, 
all cases of pyrrhic and most cases of trochee in the syllables 
immediately preceding the fourth syllable are merely sporadic, 
whacever be the caesura; but that the trochee before the 

vatormic middle, w w , is not uncommon; and add that 

the caesura is here after the fourth syllable (No. 15). The 
prevailing types of the great epic are ( as is also shown by the 
table) an iambic or spondaic opening, w _ w followed by 



_ \j w _, _ w , w w , all three of which are found in 

the same stanzas. They are always commingled in the older 
parts of the epic and even in later parts, but, on the other 
hand, the first, or choriambic middle, is the stanza-form often 
exclusively employed in late sections, as is shown below in 
the paragraphs on the Stanza. 

Bird’s-eye View of Tristubh Padas. 

The regular Mahabharata tristubh, which is of the hendeka 
variety (i), appears then in three (four) principal phases 
(all others being rare or sporadic), thus: 

( (a) .w w. u u u yj passim, but restricted as in (b). 

j < (b) id w WWW w id t 

(c) w w id w w id [ common 

(d) w w ww w — id 

Besides these, as will be shown below, there are other Bharata 
types, thus: 

ii w w^ w | id 

iii id id id | w w 


catalectic, dekasyllabic. 

ir w w — id id | id w id — w id 

/(a)(w id id | id w id. w id 

|(b)(w. w | w w w w w id 

viv w w i d idlww w 

hypermeters, dodekas (with 
still other sporadic ar- 
rangements of syllables). 

vii w w w | w w 

viii id id ididl id w id w id. 

ix id w Iwwwid w id 

Jagatl forms of these padas will be discussed below. 

The epic tristubh, then, is not (as has been affirmed by a 
distinguished scholar) of one uniform type. On an aver- 
age, about one-fifth of the Bharata tristubhs of the regular 
mixed type have twelve-syllable padas, which, however, are 
not jagatls, since they have the tristubh finale. A noticeable 
point is the common (not passim) occurrence of the trochaic 

opening, _ w , in some sections of tristubhs, and also in 

such sections the comparative rarity of the choriambic tristubh 
as compared with the tristubhs which have forms of §alinl, 

— w , or vatormi, w w , character (though not strictly 

§alinl or vatormi padas). Thus in the hundred odd padas 

Double hypermeters, thirteen 



that complete in tristubh form the story of the Frog-girl given 
above, there are only a dozen of choriambic form ; while only 

one stanza out of the twenty-five is of upajati (^_ w 

form throughout, though two others have two consecutive 
choriambic padas. 

The Ramayana Tristubh. 

Very different is the scheme presented by the Ramayana. 
Here the upajati is almost exclusively the form of tristubh 
employed, and all the variegated padas of the Bharata are 
practically reduced to one type. In fact, the exceptions, given 
under Nos. 7, 13, 19, 23, of the Illustrations, Appendix C, are 
so few as scarcely to modify the statement that the Ramayana 

employs only one kind of tristubh, 1 which is uu_u_y, 

with variable caesura, as in 
R. vi, 128, 122 : 

ayusyam arogyakaram yaqasyam 

saubhratrkam buddhikaraiii eubharii ca 
qrotavyam etan niyamena sadbhir 
akhyanam ojaskaram rddhikamaih 
R. ii, 82, 32: 

tatah samutthaya kule kule te 
rajanyavaiqya vrsalaq ca viprah 
ayuyujann ustrarathan kharaiic ca 
nagan hayane eai ’va kulaprasutan 

1 I pass over some obvious errors, noticing their place : typographical, G. 
iv, 43, 69, vicetum; R. vi, 59, 12, pataka ; G. vii, 7, 48 (agani in E). These 
affect the fourth syllable. E. iv,28, 66, affects the eighth, nigrhefor nigrahe. 
Other palpable errors affecting the metre are : G. ii, 80, 24, ksudlia ca tandrya 
(ca 1) vipannatam gatah, not in E. ; G. iii,63, 28, jahau tada trtsamudbhavam 
klamam (in E., ksudha duhklia 0 ); ib. 29, pada ends e9as tada (compare end 
of E. iii, 63, 6 b, etya klefam, where, however, kl probably does not make 
position) ; G. v, 14, 66, priyam aviksamano Raghunandasya, corrected by R., 
priyam apajyan Raghunandanasya tam ; ib. 19, 34, evam sa tarn hetubhir 
anuviksya, for anvaveksya (the form, though with v. 1., in R.) ; nagati for 
nagyati in v, 80, 24, is noticed under No. 19 ; G. vii, 20, 44, tam arcavitva 
nigacaro jagau (not in E.) has apparently lost a ca (cf. d) ; G. vii, 40, 19, 
Hanumatah kah sthasyati purastat, for sthasyati kah (R. 36, 46). In R. vi, 
59, 12, nanapataka dhvaj achat rajus tam (gastra in G. 35, 6), ceh becomes ch 
as in Mbh. i, 3,658, prchami tvam. Contrast sagvadhvajacchatramahapatakam, 
E.ib. 135. 



R. iy, 11, 93 : 

yatha hi tejahsu 1 varah sada rayir 
yatha hi qailo Himavan mahadrisu 
yatha catuspatsu ca kesarl varas 
tatha naranam asi vikrame varah 

This uniformity of metre, resulting in an almost classical 
tristubh, places the Ramayana on the same plane, when com- 
pared with the Bharata, as we saw it occupied from the point 
of view of the cloka. The more antique forms of regular 
tristubhs are found in the Bharata. 2 

Yet if this is the case in the regular tristubh, still more 
striking is the difference between the two epics in respect of 
the catalectic, hypermetric, and other irregular tristubhs, which 
are antique and found in the Bharata, but are unknown to the 
Ramayana. But before taking up these three classes as they 
appear in the great epic, I have a few words to say in regard 
to the final amphibrach or seolius. 

The Seolius. 

The many examples given in Appendix C sufficiently 
illustrate the fact that after the long eighth syllable (very 
rarely short) 3 the ninth syllable of the tristubh is regularly 

1 In G. 11, 11, yatha hi tejasrivaro divakaro, etc., followed by a stanza not 
in R., with na sarvayakseyadhaneyvaro vibhuh, the other padas haring caesura 
after fourth or fifth. 

2 One cannot, howerer, claim as evidence of antiquity the antique yalini 

and vatormi type of pada, either pure or in parti-form, w ± j _ w 

and c/ , without noting that these are also Puranic, though 

rare here, and chiefly loans. Thus in a pure single (separate) upendra 
stanza at Vayu P. r, 19, stands pravartate codyamanah samantat. So ib. ix, 
113, where a, b, d, have falirii form, and c has : diyah yrotre caranau ca ’sva 
bhumih. Most of this is epic, e. g., ib. xvii, 7 d, na jayate mriyate va ka- 
dacit (Gita, 2, 20). Still rarer (as in Gita, 8, 9) is the form in the same 
Purana, xiv, 7 c, kavim puranam anugasitaram. I take this opportunity of 
stating that I shall hereafter use upendra and vaik;astha as shorter forms 
of upendravajra and van<;asthabila, though I believe only the latter has 

s See Appendix C, under No. 15, ekarh sama yajur ek&m rg eka, xii, 60, 

47 c. 



short, the tenth is long, and the eleventh is anceps. This rule 
is seldom violated, but in the (dbicarita, in, 197, 8, we find: — 

gadami vedan vicinomi cchandah 
sarve veda aksaraso me adhltah 
na sadhu danaih crotriyasya pradanam 
ma pradah qyenaya na kapoto 'smi 

Here we find, in pada a, the phenomenon discussed, above, 
in relation to the close of the gloka. Before cchandah the 
vowel should weigh heavy, but it is doubtless reckoned light. 
In b, me 'dhltah is more probable than the (hypermetric) 
pada, as it appears in both texts (above) ; but since this is a 
possible form, the pada cannot be cited for a long ninth. 
Pada c is regular. In d, the pada may be corrupt, the 
necessary ma (— mam) apparently being lost after the pro- 
hibitive ma, though a long ninth cannot be avoided in any 
circumstances with the rest of the text as it is. I suspect 
that gyenaya has taken the place of a vocative, and that the 
verse read originally : ma ma pradii na ’smi rajan kapotah ; 
but it may be a specimen of the group of six before ccesura, 
like yatra dev! Ganga | satatam prasutli, and the other 
cases of the sort cited below, if the hiatus may be assumed 
to leave a short vowel, ma pradah, qyenaya na kapoto asmi 
(hypermetric), as in xiv, 9, 9 a, just below. The tale, how- 
ever, is a popular story, doubtless handed down in rough verse, 
and since the long ninth is actually found in such verse, it is 
not necessary to assume that the pada must be correct. In the 
following stanzas, in the same way, we find the vowel appar- 
ently reckoned as still short (light) before gy. The cases are : 

iii, 197, 15 c, yasmin dege ramase 'tiva, gyena 

ib. 18 b, saumyo hy ayarii, kith na janasi, gyena 

ib. 24 b, prcchami te, 1 cytkune, ko nu cyenah 2 

1 Perhaps accusative. I refer to C. only when the reading differs. 

3 On gyena as giena in 19 c, see the paragraph on Defective Tristubhs be- 
low. Above I have cited cases where the vowel is short (light syllable) before 
mute and liquid in glokas and also given examples in tristubh, where c& 
brahma, °tt ksatram, and “nil Dronarn make the scolius. The latter is, as it 
were, strengthened to make position in vii, 179, 47 b, antarmanah kurusu 
pradravatsu (C. 8,161, pra). 



In v, 44, 24 d tlie long ninth is admitted into an old pada : 
na ’nyah pantha ayanaya vidyate, in YS. 31, 18; Civet. Up. 
3, 8 : vidyate (a)yanaya (perhaps in the epic for : na anyah 
pantha ayanaya vidyate). 

Another apparent example is found in the stanza 1 xii, 
270, 23: 

caturdvaram purusam caturmukham 
eaturdha cai ’nam upayati vaca 
bahubhyam vaca udaracl upasthat 
tesaih dvaraiii dvarapalo bubhuset 

But here the first pack is perhaps a jagatl, either with ca lost 
before purusam or (but this is unlikely) with resolution of 
the semivowel: caturduaram purusarii eaturmukham (as in 
RV. iv, 51, 2, vi u vrajasya tamaso duara) ; though as it stands 
it is a metrical duplicate of na ’nyah pantha (above). 

Two metrical irregularities appear in xiv, 9, 4 c : 

samvarto yajayati ’ti me grutam 

This pack also is of the same form as the two last, with the 
irregular U v_vas second foot and _ w ^ as the scolius ; yet 
to read grutarn me corrects them both. But in iv, 8, 8 a, gr 
certainly fail to make position, though not before a scolius. 
The first section has another example, xiv, 9, 9 a, aham ga- 
cchami maghavan duto 'dya, where hiatus, as in the first 
example above, may perhaps be assumed with a short vowel : 
aham gacchami [ maghavan duto adya, unless an inversion has 
taken place, adya dutah, with maghavo (or bhagavo, C.) be- 
fore it. Below, ib. 31 b, saha ’gvibhyam somam agrhnad ekah, 
B. saves the metre and C. 249 saves the grammar. 

In the Harivanga is found one case at 7,593 c, which is cor- 
rected in B. : 

prahur vipras tvarii guninam tattvajiiah 

Though of the same class with the Anugasana piida (cited 
below) ending in prayaechat, yet, while the latter may be 
easily emended, tattvajnah is intractable, and the hypermeter 

1 For the meaning, compare ib. 28 ; v. 1. in 300, 28 : catvari yasva dvarani 
suguptany amarottamah, upastliam udaraiii iiastau vak caturtlii sa dharmavit. 



of B. 2, 74, 32, is probably correct : prahur vipras tvam | guni- 
nam tattvavijiiah. Another apparent case in H. 14,732 d, 
where yada ve ’gvarah ends a tristubh, is a mere misprint for 
yadavegvarah, 8, 82, 13. But xii, 292, 22 d, antye madhye va 
vanam agritya stheyam, has a clear case of w for w _ w. 

A secondary csesura is more likely not to be found before 
the scolius than to be found there. Examples of both cases 
are given (incidentally) in the examples of the different sorts 
of tristubh. Calling the scolius an addition is, then, merely a 
mechanical device, to show the pada forms free of their uni- 
form close. In reality, the scolius, because it is always the 
same, is the most important part of the pada, since it seals the 
tristubh. To show how the second caesura does not divide off 
the scolius as a sort of tail tied on to the pada proper, may be 
taken vii, 179, 13 a-b: 

Ssthaya tarii kancanaratnacitram 
rathottamam sihhavat saiimanada 

The form ^ _ w is then the only form of the epic scolius, 
except for a few cases of seeming carelessness, as in prayacchat 
and vidyate, where special reasons may have induced the ex- 
tant form, or, as in cases before ks, cell, etc., where advantage 
appears to have been taken of a Gatlia freedom in reckoning 
a heavy syllable as light in certain eases. Of the scolius type 
_ w, which Fausboll (previously) set up for the Dhamma- 
pada, the epic has parallel examples, but I doubt whether 
the single example to be found in the Dhamma, vs. 306 : 

yo va ’pi ka- | tva na karo- | mi ’ti ca ’ha 

will be found on second thought really to support this interpre- 
tation. F or in this case, as in all similar epic examples, the 

division is not, as Fausboll assumed, I | 

but (as a hypermeter) ^ | w |u_u, exactly as 

in the common hypermeters of the epic, e. g., sa vai rajan na | 
’bhyadhikah kathyate ca, where the only difference between 
the scansion and that of the more usual hypermeter, e. g., 
yasya ’vibhaktam | vasu rajan sahayaih, is that in the latter 



case the caesura is normal, while in the former it is neglected. 1 
On such cases, see the section just below, on Hypermeters. 

The epic, then, as a whole, has passed far beyond the Vedie 
stage, where the final syllables of a tristubh are ( jo-) kj _ • 
nor is it likely that the few cases above are to be explained as 
archaisms rather than as further examples of such slovenliness 
as has been met before in the examples already given. For 
even the Rig Veda poets are already tending to a stricter form, 
w _ ii, as is shown, for example, by the substitution of maslya 
for maiislya, RV. x, 53, 4, merely to win an amphibrach. 

Catalectic and Hypermetric Tristubhs. 

A short form of tristubh is where a syllable is omitted, but 
in such a way as to preserve the characteristic final cadence, 
giving the pentad form familiar to the Rig Veda; as in 
Mbh. iii, 195, 3, tain tvam prcchami | kathaih tu rajan, like 
RV. i, 67, 8, ya Irii ciketa | guha bhavantam. Although 
catalectic is a name more properly applied to a pada cut off 
at the end, I shall yet call the double pentad a catalectic 

In a jagatl, by the addition of a syllable, the final trochee or 
spondee of the tristubh’s amphibrach is converted into a di- 
iambus ; in a hypermetric tristubh, the final cadence is preserved 
intact, the tristubh’s nature is not lost, but a syllable is pre- 
fixed or inserted elsewhere. It may be said that any dodeka 
is a jagat! pada. I shall not quarrel with this (native) defini- 
tion, but the difference here is one of metrical character, and 
must be strongly marked in name. Admitting then that it is 
somewhat arbitrary, I shall designate as a jagatl only the 
diiambically closed pada ; the other, as a hypermetric tristubh. 

1 This interpretation, anyway, seems to be merely a slight oversight on the 
part of the learned editor. In No. 329, eko care matafig’ aranne va nago, the 

first foot is correctly given as . The choriamb doubtless caused 

the different interpretation ; but the middle foot w is parallel to 

\j , as shown in the examples cited below. [The new text in 306 

omits iti : but I keep the remark above, written prior to the new text's appear- 
ance, as the old text has authority and need not be changed metri causa.] 



Besides the prefixed or inserted syllable, which gives two 
varieties of the hypermetric tristubh, a tristubh pada may 
have both the prefixed and inserted syllables. The tristubh, 
then, as shown in the bird’s-eye view on p. 275, may consist 
of ten, eleven, twelve, or thirteen syllables, without losing its 
characteristic cadence. Unique, however, and not typical (I 
may add) is a fourteen-syllable tristubh. Apart from all these 
forms lies the matra-tristubh, of thirteen syllables, but with two 
breves reckoned as equivalent to one long .syllable. Post- 
poning the examination of these forms, I take up now, reckon- 
ing the regular tristubh (above) as i, the catalectic and simple 
or dodeka hypermetric tristubh, ii-vi. 

ii-iii. The Catalectic Tristubh. 

ii. In this form the caesura falls after the fifth syllable. The 
pada is one of a tristubh stanza. Examples are iii, 113, 23 : 

Arundhatl va subhaga Vasistkam 
Lopamudra va yatha hy Agastyam 
Nalasya vai DamayantI yatha ’bhud 
yatha C'acI Vajradharasya cai ’va 

Here b can be scanned only as Lopamudra va | yatha hy 
Agastyam. Another case, referred to above, is found in the 
stanzas at iii, 195, 3-4 : 

3, vidvesanam paramam jlvaloke 

kuryan narah parthiva yacyamanah 
tarn team prcchami kathain tu rajan 
dadyad bhavcin dayitam ca me ’dya 

4, na ca ’nuklrtayed 1 adya dattva 

ayacyam artham na ca sari^rnomi 
prapyam artham ca samrjrutya 

tarn ca ’pi dattva susukhl bhavami 

In 3 d and 4 a, the caesura is shifted, and the padas can be 
read as 

dadyad bhavan da- | yitam ca me 'dya 
na ca ’nuklrta- | yed adya dattva 

1 Tliis seems better than anukirtaye (he) dadya (N.). 



In 4 c, there is a gloka pada ; unless d be reft of its opening, 
to leave another pentad : prapyam arthaih ea samgrutya tarn 
ca, which would leave d as : api dattva su- J sukhi bhavami. 

The dekasyllabic pada is particularly striking when united 
with the hypermetric pacla (10 + 12.) An example occurs in 
the same story, iii, 197, 26, c-d: 

etad vo laksma | givam karomi 

hiranyavarnam ] rueiram punyagandham. 

The ten-syllable pada ib. 17 b, has, perhaps, lost a syllable, 
(tarn) te pagyantu : 

(a) uksanam vehatam aniinam nayantu 
(b) te pagyantu purusa mamai ’va 
bhayahitasya dayam mama ’ntikat tvam 

praty am nayantu tvarn hy enam ma hihslh 

(a) ww w _ w (No. 13, hypermetric) 

(b) ( ) w ww w — w (No. 20) 

For c and d, see No. 23 and No. 7, in the Illustrations of 
Appendix C. It is possible, however, that b belongs under 
another head (below). Giving a patois pronunciation, pasi- 
antu, would make the verse quite smooth. In the subsequent 
stanza, 19 c, there appears to be a case of resolved semi-vowel 
(giena for gyena), a regular pada: 

yatha giena priyam eva kuryam, 

though it may be read as cataleetic. 1 

A case in C. viii, 4,545 d, is corrected in B. 89, 22 : 

C. : vayavyastrena, tatah sa Karnat 
B.: vayavyastrena ’patatah sa Karnat 

In xii, 322, 72 = 12,115, where C. has kirn te dhanena 
bandhubhis te, B. has the dekasyllabic pada: 

kim te dhanena, kirn, bandhubhis te, 

the other padas being hendekas. A combination of hyper- 

1 For the verse in the same stanza, yatha mam (hi) vai sadhuvadaih pra- 
sannah, see below, The Hypermetric Tristuhh. 



metric, catalectic, and hyper-hypermetric pada occurs in 
H. 7,448: 

yasmad bhfitanam | bhutir anto 'tha madhyaxh 
dhrtir vibhutih | grutig ca liudrah 
graha (sic) ’bhibhutasya purusasye ’qvarasya 

Compare 1 H. 8,399 : 

tam kurdamanam madhusudanah sa 
drstva mahatma | harsanvitas tab 
cukurda satya sahito mahatma 

balasya dhlman | harsagamartham 

iii. This pada is what may be called csesurally catalectic. 
Like the last, it is antique, in Veda and Upanishads, and the 
epic has but few examples. The pause follows the fourth 
syllable, which is usually heavy. Here the caesura, so to 
speak, costs a syllable and, unless read with sufficient time 
allowance, the tristubh appears to be crippled. Of this 
sort are: 

i, 3, 61 d, maya ’(jvinau samanakti carsanl (so 66 c) 
i, 92, 14 a, prcchami tvam, sprbanlyarupa 

In the latter example there may be corruption. Compare 
i, 88, 10 c, tat tvam prcchami sprhanlyarupa, but the open- 
ing phrase, prcchami tvam is stereotyped, i, 93, 21 a; v, 48, 
1 a, etc. We may compare RY. i, 120, 4, vi prchami pakia 
na devan. 2 The next case is 

iii, 197, 27 b, surarslnam atha sammato bhrcam 
Although this pada has eleven syllables, it is not a tristubh, 
but a catalectic jagatl, analogous to the tristubhs of the same 
nature. The whole stanza consists of syllables 13 + 11 + 12 
+ 11, but a is doubly hypermetric (explained below), so that 
there is no alternate symmetry but ehiastic symmetry, thus : 

13 (= 11) +12 + 12 + 11 

1 In the Bombay edition, 2, 72, 59 : dhrtir bhutir yag ca guha rrutii; ca 
guha ’bhi 0 , etc. (on this, see below). The following 8,399 = 2, 89, 17, also 
avoids the same cadence by reading : drstva mahatma ca mudanvito 'bhut 
. . . harsagamartham ca balasya dhlman. 

3 C. in 3,664 has prchami (sic) tvam. 



It is, however, possible, perhaps, to resolve the -am. 

v, 42, 5 a, pramadad vai asurah parabhavan (jagatl) 
v, 42, 21 a, va etad va bhagavan sa nityo 

In this ease, although there is no possible objection to 
reading the pada as it stands, it is possible that a bhati has 
been lost after etad. The sense is yaj jagad iva bhati sa 
nityo 'vikarl bhagavan (N.). Compare 43, 7, jagad bhati. 

v, 46, 3 c, atandritah Savitur vivasvan 

The same criticism. Before Savitur, sa may have been 
dropped, as in C. viii, 3,343 c, gete papah suvibliinnagatrah, 
where B. restores the metre with gete sa papah. So C. omits 
su in the aparavaktra, xii, 9,035 b, but corrects it in repeat- 
ing the verse at 10,530. Nevertheless, I prefer the text as 
it stands, especially as any correction would have to be ex- 
tended into the next stanza, where we find : 

ib. 4 b, dicah gukro bhuvanam bibharti 

Here it is easy to suggest sambibharti, but emendation is 

v, 48, 37 c, Matsyaih sardliam anrgangarupaih 

The next stanza has jyestham Matsyam anrgansiiryarupam, 
which makes it rather doubtful whether this form may not 
have stood in 37 c. 

v, 67, 6 c, anayasva pitaram mahavratam (jagatl) 
viii, 68, 7 a, apy agisma vayam Arjuna tvayi 

C. 3,386 has atha ’gisma. Possibly agisama should be read 
but it is not necessary. The brevis is noticeable (compare 
above, in § ii, iii, 197, 17 b). 

xiii, 76, 7 a (after the injunction in the half-gloka, vs. 6) : 

6, pravigya ca gavam madhye imam grutim udaharet 
7 a, gaur me mata vrsabhah pita me 

divam garma jagatl me pratistha, etc. 



xiii, 102, 55 d: 

budhyami tvam Vrtrahanam catakratum 
vyatikramantam bhuvanani viqva 
kacein na vaca vrjinaiii kadaeid 
akarsam te manaso ' bhisangat 

iv-ix. The Hypermetric Tristubh. 

iv-vi. Simple Hypekmeteks. 

The first form, iv, is the initial hypermeter ; a light syllable 
appears to be prefixed to an iambic opening. The same effect 
is produced, in some cases with the same words, as that already 
described in the account of the gloka. The pada starts with 
an anapaestic slide. The difference is one of frequency, since 
in the case of the tristubh the initial hypermeter is not very 
common. Most of the cases have a brevis and in fact, to my 
ear, the long (heavy) initial belongs in another category (vi) ; 
but I admit that in yatra devi Gangii satatam prasuta and 
the few similar cases it is doubtful how we should regard the 
extra syllable. I have noticed with short initial the following 
cases (iv) : 

i, 3, 147 b, vayatas tantiin satatam vartayantyau (No. 13) 
i, 76, 55 a, asur&ih surayam bkavato 'smi dattah (No. 1) 

Here the preceding pada ends in i, but it is scarcely possible 
that the two tristubhs should have been read as a unit. The 
same thing occurs occasionally in the examples of hypermetric 

i, 92, 6 c, kuta ayatah katarasyam diqi tvam (No. 13) 

iii, 5, 10 a, tata utthaya Viduram Pandaveyah (No. 15) 

v, 42, 6 c, pitrloke rajyam anuqasti devah (No. 20) 
v, 44, 18 b, dhanam acaryaya tad anuprayacchet (No. 20) 

xii, 63, 4 c, vrsallpatih picuno nartanag ca (No. 12) 

xiii, 76, 14 d, pratigrhnau vai gopradane vidhijnah (No. 7) 
xiii, 102, 19 a, atithivratah suvrata ye jana vai (No. 6) 

ib. 35 c, (jagatl), Varunasya rajnah sadane mahatmanah 
xiii, 126, 38 a, bahule samange hy akutobhaye ca (No. 1) 

H.2, 72, 33 b, krtinam viram (C, 7,422 dhiraih) dana- 
vanam ca badham (No. 7) 



All these cases have an anapaestic opening ; all but one have 
the fifth syllable heavy. 1 Some have been given under 
the examples referred to above. 

v, a. Much more frequent is the inserted fifth. I do not 
mean, of course, that a regular tristubh is first made and a syl- 
lable is then inserted, but that the cadence does not have the 
rhythm of iv, to wit, but (with the caesura regu- 
larly after the fifth syllable) “ini , so that the effect 

is that of a syllable inserted at the place of caesura. This 
measure produces rather a pleasing alteration and is frequently 
found in regular tristubh stanzas, scanned exactly like the 
other paclas with the modification thus indicated. The form 
is Vedic, and is found also in the Upanishads and in the 
Buddhistic texts. Examples are: 

i, 71, 40 d, yatha tvadartharii | raksita ’ham careyam 2 
v, 48, 101 d, safiiyudhyamana | Dhartarastra na santi 

The effect of this measure I have endeavored to reproduce 
from the following extract, v, 48, 75-7G: 

ayaiii Gandharans tarasa sampramathya 
jitva putran Nagnajitah samagran 
baddham mumoca vinadantam prasahya 
Sudarqanam vai devatanam lalamain 
ayam Kapate 3 nijaghana Pandyam 
tatha Kalingan Dantakure mamarda 
anena dagdha varsaptigan viniitha 
Varanasi nagarl sambabhuva 

And you Gandharas, at a blow Krishna vanquished, 

And conquered all Nagnajita’s descendants, 

Their plaining victim, as he lay bound, releasing 
(Of gods the jewel, “Beautiful” called, a fair man); 

1 On this case (tata utthaya), see below, p. 200. 

2 Compare with this example, Rig Veda, i, 120, 3, ta no vidvansa | manma 
vocetam adya, and for other Vedic parallels, Oldenberg, Hymnen des Rig 
Veda, yoI. i, p. 60 ff. (ZDMG. vol. xxvii, p. 7o). 

3 v. 1., kapatena jaghana. Below, the scholiast explains dantakure as in 
battle rather than as a proper name. Perhaps Dantakrurarh jaghana (a be- 
fore kr), as in vii, 70, 5. 



He at Kapat slew in a war the Pandya, 

He smote Kalingas, Dantakur’s men a-fighting, 

He too, that hero, burned and enslaved a long time 

Benares town, city sans help unaided. 

It will be observed that the first part of tills measure is that 
of the regular tristubh with the cm sura after the fifth, as in 
Yamo 'bravln 1 mam: na mrto 'si saumya, xiii, 71, 18a, which 
form may have led to the establishment of the liypermeter 
on the one hand and the eaesurally catalectic pada on the 

The texts sometimes show variations, like those found in 
the simple tristubh forms. 2 Thus in vii, 179, 45 d, where 
C. has the hypermeter, 11. omits the extra syllable : sainpag- 
yanto (vai) vijayam raksasasya. 

Of the different hypermetric forms, the commonest are those 

in which the fifth syllable is followed by _ w or w w ; 

less often by All three occur at i, 7 6, 50 ff. : 

50, kacasya margam pratipatsye na bhoksye 

53, guror hi bhlto vidyaya co ’pahutah 

54, smaratni sarvaih yac ca yatka ca vrttam. 

The extra syllable, like the initial, may be heavy or light ; 

but except when followed by w w the latter is rare. The 

second and fourth syllables are rarely light. I give below 
examples of the different forms. First of the common va- 
rieties (but w w w _ as second foot is the rarest of these) : 

yaqo na nacyej, jfiatibhedao ca na syat, iii, 4, 8 a 
vadhaya rajan, Karnasutasya saiiikhye, viii, 85, 36 b 
ma vai dvitlyam rna trtlyam ca vance(t), iii, 297, 25 c 

1 fin page 186, note 1, I have referred to Tama's world as portrayed in 
Sabha in contrast to “ elsewhere.” The remark is correct, but elsewhere is 
not everywhere else; e. g., this account of Naciketas represents it as blissful. 
Usually, of course, it is a hell. 

2 These changes T have discussed in A. J. Phil., xx, p. 18 ff. as affecting vii, 

163. In vii, 179, 24 a, B. has w for w w in C., with several 

similar changes close by ; strikingly in 32 d = 8,140, where B has no babhiivuh 
(C., na). 



So in v, 44, 24 c; vii, 2, 33 b; viii, 42, 17 c; xii 278 (7), 
6 a, etc. 

A case of fifth brevis and also fourth brevis is found in i, 
1, 217 c, dvyuna vii'njatir ahatii ’ksauhinlnam ; and fourth 
brevis in iii, 197, 12 d, na trunniii labliet tranam icehan sa 
kale; where, however, C. has labhate (labh’te) which may l»e 
correct. 1 All five syllables are heavy in ii, 77, 10 b; kan- 
yiim Pailcallm Panda vebhyah pradaya. Unique (I tliink) 
are breves in the third and fourth syllables : datvii ’naduhaiii 
suryalokaiii vrajanti, 2 iii, 186, 8 b (No. 10). 

Preceded by brevis (fifth syllable) : 

samanam mtirdhni rathayanaiii viyanti, i, 3, 64 b 
tatha titiksur atitiksor viqistah, i, 87, G b = xii, 300, 15 b 
yas tv evam brahma tapasa ’nveti vidvan, iii, 192, 56 c 
dharmam puriinam upajivanti santah, viii, 45, 16 c 
taiii vai manyeta pitarain niataraiii ca, xii, 108, 22 c 
garbho 'mrtasya jagato ‘sya pratistha, xiii, 76, 10 b 

So i, 1, 212 c ; 1, 213 c ; 89, 6 c ; 232, 16 c ; iii, 4, 13 a ; viii, 
42, 16 b ; etc. 

Preceded by a heavy syllable : 

hataiii saihgratne Sahadevena papam, i, 1, 20S c 
idaiii ca rajan hitam uktaiii na cet tvam, iii. 4, 12 c 
tatha caktlr apy adhamaiii ghorarftpah, v, 181, 9d 
tatha vayvagul pramimanaiii jagac ca, vii, 201, 67 b 
yasyfi ’vibhaktaiii vasu rajan sahayfiih, iii, 5, 20 a 
tan aha sarvan rsimukliyan Agastyah, xiii, 94, 9 a 

So iii, 5, 18b; 113, 6b: v, 42, 15a; 48, 46c; vii, 179, 
42 a; viii, 37, 30 b: 42, 9d, etc. 

Cases of fourth brevis are ii, 56. 15 c, pay cat tapsyase tad 
upakramya vakyam; and i, 1, 216 b, tatha lumdhubhih pitr- 
bhir bhratrbhig ca. 

1 Compare also iii, 13,291 a, Tatha marii hi vai sadhuvarlaih prasannah, 
where, however, B. 197, 19, omits hi, which makes when retained, a bhujarii- 
gaprayata pada; q. t. below, under the head of Ak'araechandas. 

2 Compare BV. viii, 59, 7, indravaruna | saumanasam adrptam, cited by 
Oldenberg, loc. cit., p. 08. 




For v, 516, prayaecha mahyam bhavatsahyam karisye, B. 16, 
82 d, has tava sahyam. In vii, 200, 82 a, B. has tasya ’syatas 
tan nigitan pitadharan, where C. 9,389 has sunigitan. All five 
syllables are heavy in ii, 77, 7 a ; eitran samnahan avamunc- 
antu eai ’sam. 

Preceded by brevis : 

na cen mam Jisnur ahvayita sabhayam, ii, 58, 16 b 
tans te dadani ma prapata prapatam, i, 92, 11 a = 93, 3 

Preceded by a heavy syllable : 

gomayur uccair vyaharad agnihotre, ii, 71, 22 b 
amanyamanah ksatri} 7 a kimcid anyat, v, 42, 15 c 
amantraye tram bruhi jay am rane me, viii, 67, 22 c 
anarthakam me dargitavan asi tvam, viii, 68, 8 c 
prayaecha ’nyasmai Gandivam etad adya, viii, 68, 28 a 1 
nai ’ko bahubhyo Gautami raksitavyah, xiii, 1, 30 b 

There is, I believe, only one other case of this form in the 
thirteenth book, 103, 42 c. It is mre as a tristubh hyper- 
meter, but it occurs also (see below) as a jagati. 

Besides these forms are found : w w w _, of which I have 
but sporadic examples : sa yatre ’cchasi, Vidura, tatra gaccha, 
ii, 64, 11 c (note to No. 20) ; ahaih karte 'ti, Vidura, ma ca 
mahsthah, and na tvam prcchami, Vidura, yad dhitam me, ii, 
64, 7 a and c (C. has ma ’vamahsthali) ; pratas tri varga ghrta- 
vaha vipapma, xiii, 26, 88 c (No. 19, ad fin.). 

Between divisions iv and v stand a couple of cases in which 
the initial syllable is heavy but the second is light. They 
belong neither to iv with its anap;estic opening, nor to v with 
its iambic or spondaic opening : agvinav indum | amrtam 
vrttabhuyiiu, i, 3, 63 a; atra Kaunteya | saliito bliratrbhis 
tvam, iii, 134, 41 a. Compare above p. 286, tata uttliayS 
viduram Pandaveyah (No. 15). 2 

1 This pada is followed by tvatto yo 'strair abhyadhiko a narendrah, with 
the CEesura ignored. Pada a is virtually repeated in viii, 09, 72 c-d, anyasmai 
tvam Gandivam dehi Partha, tvatto 'strair va viryato va vi(,:is!ah. 

2 Such Vedic cases as this last are grouped by Oldenberg, loc. cit., with 
those just mentioned, e. g., abhi krsnena rajasa dyam rnoti, liV. i, 35, 9, 



Quite exceptional, though corresponding to recognized (but 
unusual) forms of the regular tristubh, are further : 

dadarca ’sinani dharmatmanam vivikte, iii, 5, 6 c 
(No. 27) 

kirn vidviso vai mam evarii vyahareyuh, ii, 71, 7 c 
(No. 26) 

dhrtayudhah su-krtlnam uttamaujah (v. 1. sukrtinam), 

H. 7,442 c (No. 24 ; v. 1. in ii, 72, 53) ’ 

Compare also a case of No. 23, below, p. 294. 

The hypermeter beginning with an anaptest, iv, is found 
also in popular Buddhistic poetry, where also a long syllable 
rarely takes the place of the initial brevis. There are, for 
example, in the Dhammapada, half a dozen cases with anapaest, 
but none with long initial (vs. 40 has naga-, in the new text). 

Examples of jagatls like the tristubh hypermeters given 
above 1 are : 

athai ’va cyeno vajrahastah cacfpatih, iii, 197, 25 b 
bhitam prapannam yo hi dadati catrave, iii, 197, 12 c 
svadhyayaQlla guruquqrusane ratah, xiii, 102, 33 a 
satye sthitanam vedavidam mahatmanam, xiii, 102, 34 c 
balena tulyo yasya puman na vidyate, ii, 65, 25 a 
(a has 13 syllables; b, 12; c-d, 11 each) 

Occasionally a tristubh and jagatT occur in the same stanza 
in hypermetric form, as in iii, 134, 39: 

tato 'stavakram matur atha ’ntike pita 

nadlm samangam (jlghram imam vinasva 
(provaca cai ’naiii sa tatha viveca) 

The unique tristubh-pada of fourteen syllables, of which I 
spoke above, runs, ii, 64, Id: 

balan iva ’sman avamanyase nityam eva 

perhaps better so than with the initial hypermeter, as the latter, except for 
this example, is characterized by a heavy fifth, as stated above. 

1 Also Vedic, e. g., vifvasu dhursu vajakrtyesu satpate, JRV. x, 60, 2 (in- 
cluded under tristubhs in Oldenberg’s list, loc. cit. ). 



The scholiast, who rarely touches on purely metrical phe- 
nomena, 1 explains this as “ redundant and archaic,” recogniz- 
ing the pada as it stands. But it is impossible to suppress the 
suspicion that avamanyase stands for an original manyase, a 

regular hypermeter (_,_w ), “thou regardest us as 

children,” strengthened by some one to “ thou despisest us.” 
Another, but doubtful, example is given below. 

The distribution of these hypermetric forms, va, is somewhat 
uneven. The examples run in groups, showing clearly the 
effect of different styles. A baker’s dozen of hypermeters, for 
example, are found in the seventh book, which has three hun- 
dred and twenty tristubhs ; but half of the dozen are in the 
fifty-seven tristubhs of adliy. 179. On the other hand, the 
fourth book, which has two hundred tristubhs, has no ex- 
ample. 2 The second book, which has only one hundred and 
fifty-five tristubhs, has thirty examples. 3 In the thirteenth 
book the older parts have most examples. Thus in the few 
tristubhs that tell of the seers’ oath, adhy. 94, there are twelve 
hypermeters in thirteen tristubhs, a much greater proportion, 
as the tale is much more ancient, than is found in any other 
part of equal length in this book. 4 

As an illustration of the epic free tristubh with hyper- 
meters may be taken the following stanzas from the continu- 
ation of the story of the Frog-girl in iii, 192, 48 ff. : 

[Vamadeva uvaca] 

prayaccha vamyau mama parthiva tvam 
krtaiii hi te karyam abhyam acakyam 

1 He seldom comments on unusual rhythms, although often remarking on 
archaisms real or fancied, as for example on prasthe dattva yipinam brahma- 
nebhyah, at i, 93, 23 b, explaining prasthe as for pratasthe “with Yedic loss 
of reduplication.” 

2 The fourth book is writ like the Ramayana, in the refined style, and has 
scarcely a dozen padas of the free tristubh type, almost all its tristubhs being 

3 Two such hypermeters in one stanza are not unusual in old tales, e. g., iii, 
192, 63 a-b, janami putram dajavarsam tava 'ham jatam mahisyam (Jyena- 
jitam narendra. 

* Compare what was said above, in the note on p. 221, regarding the jlokas 
in this section. 



ma tva ’vadkld Varuno gkorapaqair 
brakmaksatrasya ’ntare vartamanam 

[rajo ’vaca] 

anadvakau suvratau sadkudantav (_ vj ) 

etad vipranam | vakanam Vamadeva (— , — w _ ^) 

tabkyam yaki tvam | yatra karoo makarse (_, _ w ) 

cckandahsi vai tvadrqam sariivakanti (_ w ) 


ckandansi vai madrcarii samvakanti 
loke ‘musmin partkiva yani santi 
asniiks tu loke mama yanam etad 

asmadvidkanam | aparesiuh ca rajan (_, W W ) 


catvaras tvam. va | gardabkak samvakantu (_, _ w ) 

qrestkaqvataryo | harayo vatarankak (_, w w ) 

tais tvarii yaki ksatriyasyai ’sa vako (— ^ ) 

mamai ’va vamyau | na tavai ’tau ki viddki (_, w w ) 


gkorark vratam brakmanasyai ’tad akur 
etad rajan yad ika ’jlvamanak 
ayasmaya gkorarupa makantaq 
catvaro va yatudkanak suraudrak 
maya prayuktas | tvadvadkam Ipsamana (_, _ w w _) 
vakantu tvain qitaqulao eaturdka 

And so on (the last stanza has six padas, as not infre- 
quently happens ). 1 

As seen in some of these stanzas, there is sometimes accord 
between the hypermeter and its environment. This is not 
rare. Thus in ii, 58, 9, three padas have the form w _ w _ 
w w and these are followed by pada d as a hy- 

permeter of the same sort; ity agato 'ham nrpa te taj ju- 
sasva. The hypermetric cadence to close a passage is not 
unusual. Thus to close a stanza, xiii, 159, 11 : sa eva pur- 
vam nijaghana daityan, sa purvadevag ca babhuva samrat, 
sa bhutanam bhavano bhutabhavyah, sa vigvasya ’sya jagataq 

1 That is, it is a strophe of two three-pada tristubhs (above, p. 194). 



ca ’bhigopta. Again, in i, 90, 5 d : bhuya(j ce ’danlm vada 
kirn te vadami; then Astaka uvaca. As hypermeters I 
should explain the difficult padas, 1, 3, 123 c-d, the latter 
having (affectation of the antique ?) choriambic opening be- 
fore KJ (No. 18 has v — \j w — w) ♦ 

van navanltam | hrdayam tlksnadharam(iti). 

The preceding pada is, I think, to be read as : 

tad viparltam j ubhayam ksatriyasya, 

instead of tad ubhayam etad viparltam. Then all the padas 
are metrical, after a fashion. 

There is a regular tristubh with the movement _ w 

w and hypermetrie in bhayahitasya dayam mama 

’ntikat tvam (cited under No. 23). Like this, but with a 
different hypermetrie opening, is the apparent pada found in 

1, 3, 63 c : hitva | girim a<jvi- | nau ga muda carantau, 

_ w w with neglected caesura. 

This brings me to the comparatively few cases of different 
caesura in this form of hypermeter. As shown in the exam- 
ples given above, the caesural pause comes after the fifth 
syllable. When this is neglected (but the practice is ob- 
served in a large majority of the cases), we have an approach 
to the shifting caesura of the former division, iv, and, as I 
have said already, it may seem simpler to regard such cases 
as initial hypermeters with long instead of short initial. But 
the difference of cadence between the opening ^ w m and 
seems, as in the case of the §loka, to mark an 
important though not a radical distinction, between these 
groups. While the ictus of the former, as in vrsallpatih, is 
w A. v J-, that of the latter, as in hatam samgrame is u i, 

l Nor does the shift of ctesura in asuraih surayam 

bhavato 'smi dattah, etc., change this. But when the second 
class shifts the csesura to the sixth syllable, as in yatra devl 
Ganga satatam prasuta, then, instead of coinciding with the 
ictus of iv, we still have necessarily the same opening with 
that of v, but still differentiated in the following. For in 



the whole tristubh, we certainly cannot read hatam samgrame 

Sahadevena papani as ui L d,, etc., whereas in the other 

case the only way, as it seems to me, to read the pada is yatra 
devi Gan | gii | satatam prasuta. I prefer, therefore, not 
to call these cases long initial hypermeters, but to class them 
separately, as vi. There are, as I have shown, cases which 
bridge the distinction and connect these classes in their ex- 
treme varieties, so that some may choose rather to consider 
them as radically identical openings ; but it is certainly con- 
venient to distinguish these forms. Of vi I have the follow- 
ing examples, the type being antique, as in Muncl. Up., iii, 1, 
6, yatra tat satyasya paramaih nidhanam, as distinguished 
from ib. ii, 2, 10, na tatra suryo bliati na candratarakam, ne 
’mti vidyuto bhanti kuto 'yam agnili, etc. I unite with 
them the sporadic cases where the caesura, instead of coming 
after the sixth, where it is usually found, is neglected or 
falls after the fourth syllable, except where, in the latter case, 
two light syllables follow : 1 

i, 89, 3 b, sa vai rajan na ’bhyadhikah kathyate ca 

(No. 13) 

i, 197, 10 d, yatra devi Gaiiga satatam prasuta (No. 3) 

ii, 64, 11 b, vicesatah ksattarahitam manusyam (No. 19) 

iii, 134, 7 a, evam Astavakrah samitau hi garjan (No. 3) 

ib, 27 c, balesu putresu krpanam vadatsu (No. 19) 

iii, 13,193 a (B. 192, 54) marnai ’va tau vamyau parigrhya 


B. omits eva, but both texts immediately after have — 
iii, 192, 55 b, na tva ’nucasmy adya prabhrti hy asatyam 


v, 42, 9 b, tatra ’nu te yanti na taranti mrtyam 2 (No. 19) 
v (42, 17), 1,592 d, etad vidvan upaiti katham nu karma 

(No. 2) 

B. has no ’paiti — 

v, 44, 10 a, gurum qisyo nityam abhivadaylta (No. 20) 

v, 44, 28 c, rathariitare barhadrathe va ’pi rajan (No. 6) 

v, 48, 77 c, vegenai ’va cailam abhihatya jambhah 

(No. 20, note) 
2 C. 1,584 has te tatra ’nuyanti. 

1 For these cases see below. 



vii, 2, 1 b, bhinnam navam iva ’tyagadhe Kurunam 

(No. 9) 1 

vii, 179, 26 b, Qaktyrstiprasamusalany ayudhani (No. 13) 2 * 
But C. 8,140, has tjaktyah prasa (regular) — 

viii, 4,546 b, praduc cakre vajrapratimaprabhavam (No. 2) 

Here B. 89, 23 has vajram atiprabhavam, but C.’s form 
(words) is a stereotyped tristubh ending, as in viii, 89, 61 d ; 
ix, 17, 19 d ; 35, 37 c ; xii, 112, 21 b, etc. ; e. g., in the last 
case, pura mahendra pratimaprabliava. 

[xii, 108, 33 a, etat sarvam anirdeqenai ’vam uktam 8 ] 

xiii, 94, 13 d, na hy utsahe drastum iha jivalokam (No. 19) 

xiv, 9, 34 c, sahasraih dantanam qatayojananam (No. 2) 

H. 2, 72, 31c, virupaksaih sudar^anam punyayonim (No. 7) 

ib. 32 d, somapauam marlcipanam varisthah (No. 8) 

ib. 44 a, vi-anjano jano 'tlia vidvan samagrah 

(Note to No. 9 in Appendix, with the pada tri-ambakam 
pustidam, etc., another case of resolution.) 

Compare also the pada cited above p. 278, ma pradah 
gyenaya, etc. 

In the explanation of the packs given above, I have partially 
accepted 4 * * * the analysis of Kiihnau, who in his book, Die 
Trishtubh-Jagatl Familie, has divided yatra tat sdtyasyd | 
paramdm nidhanam ; but I cannot cany this out in tans te 
dadani, ma prapata prapatam, and therefore separate the 
classes, reading the latter as tans te dadani | ma prapata 
prapatam. The pada with ccesura after the sixth syllable, 

1 Perhaps va for iva (as below). 

2 On this pada also, see below. 

8 This extraordinary verse, though anirdefena is vouched for by the com- 
mentator, seems by metre and meaning to have been originally a sample of 

No. 27 (with nirdeyena in its usual sense), , ^ yj. 

As it stands it must have fourth brevis (hypermeter), ww 

w w - 

4 Kuhnau’s schemes (loc. cit., pp. 104, 159) find a place even for the pada: 

yada ’rrausarh Dronah Krtavarma Krpai; ca, which does indeed stand in C. 

196 a, but is corrected in B. i, 1, 198, ’frausam having been taken over from 

the circumjacent padas, but being properly omitted (as in C. 201, yada Drone), 

leaving a regular tristubh. See, however, viii, below. 



examples above, may, however, be grouped for mechanical 
clearness with the regular tristubhs, the numbers of which 

I have added to the various specimens. 

As in the case of tristubh versus gloka pada, one cannot 
always say just which measure one has in hand when regular 
and hypermetric tristubhs run together. Thus in xiii, 80, 

II a-b: 

dhenum savatsam, 

kapilam bhuricniglm. 

vasanottarlyam ; 

or in iii, 34, 21 c-d : 

mitrani eai ’nam | aeirad bhajante 
deva ive ’ndram | upajlvauti cai ’nam, 
like a vaitallya. 

The hypermetric syllable may be only apparent (elision) 
in some cases. In the older epic I have noticed only a elided 
thus, as in v, 44, 10 d : 

esa pratk ’mo brahmacaryasya padah 

In the later epic, such elision takes place as well in the case 
of u and i, unless we assume a freer use of hypermetric sylla- 
bles ; as in : 

i, 55, 11 d, tvaiii va Varuno dharmaraja Yamo va 

vii, 201, 05 b, paraevadhinam gadinaih ca ’yatasim 

ib. c, Qubhraiii jatilam musalinaiii candramaulim 

vii, 9,455 d (=ib. d), vyaghrajinam paridadhanaiii dandapanim 

But here B. has parigliinam. 

xiv, 10, 2 a, Dhrtarastra ! prahito gaccha Maruttam 1 
H., 2, 79, 9 c, where the whole stanza reads : 

a, apo devya | rslnarn (hi 2 ) vicvadhatryo 

b, divya madantyo yah | qamkara dharmadhatryah 

c, hiranyavarnah | pavakah qivatamena 

d, rasena creyaso mam jusantu 

1 Read gacch’ (a common type. No. 14). 

2 C., 7,794, omits hi, and in b reads dharmaratryah. 



If yah followed rasena it would improve both padas ; but on 
this see the nest paragraph. In c, hypermetric, givatamena 
must be read as giv’tamena. In the next stanza (after apam 
esa smrto mantrah, intervening), C., d, has (sc. ma) 
bhartur bhaveyaih rusatl syam ca vagaga 

but here B., 11, has syam vagamga, which smacks of B.’s 
usual improving process. 

vii-ix. Double Hypermeters or Tristubhs of Thirteen 


vii. Sporadically appears an “ inserted fifth ” in addition 
to the initial liypermeter : 

xiii, 94, 3 a, rsayah sametah | pagcime vai prabhase 
xiii, 102, 39 a, catavarsajlvi | yag ca guro manusyah 

If the reading is right, this is found, but with different open- 
ing, in 

iii, 197, 27 a, etasam prajanam | palayita yagasvl. 

viii. But in the last case (though tasam may be suggested 
for etasam) a combination seems to be at work which is like 
that wrought by the caesura after the fifth, in cases where 
the tristubh then builds up its second half independently. 
Thus palayita yagasvl would be a regular second half and 
etasam prajanam would be a rough metrical equivalent of 
the type yatra dev! Gangii. The cases are : 

(1) ii, 67, 4 c, sa tvam prapadyasva | Dhrtarastrasya vegma 1 

(2) iii, 5, 20 c, sahayanam esa | samgrahane ‘bhyupayali 

(3) v, 46, 27 c, a jag caro diva- | ratram atandritag ca 2 

(4) viii, 76, 18 a prasag ca mudgarah | gaktayag tomarag ca 
(6) xiii, 159, 26 a, sa eva parthaya j gvetam agvam prayacchat 

(read prSyacchat ?) 

(6) xiv, 9, 10 b, balani sarvani | vlrudhag ca ’py amrdnan 

1 Possibly, however, prapadya has been altered here by a grammarian. 

2 In 30, ajaf caro divaratram atandrito 'ham, where C., 1,790, has aja? ca 
’horatram. The stanza is Upanishadic : angusthamatrah puruso mahatma na 
dryyate 'sau hrdi samnivistah, ajaj (etc.), sa tarn matva kavir aste pra- 
sannah (as in Ivatha vi, 17, etc.). 



(7) H. 2, 72, 32 a, bhunkte ya eko (pronounce yaiko) vibhur | 

jagato viQvam agryam 

(8) ib. 47 d, abhi trivistapam | tjaranam yami Eudram 

(9) ib. C. 7448 c, guha ’bhibhutasya | purusasye ’gvarasya 1 

And so, perhaps, in the case cited above from H., 2, 79, 9 b, 
divya madantyo yah | (jamkara dharmadhatryah (when, after 
rasena in d, tah may be supplied). 

The number of cases (all I have found) is considerably 
reduced by reading in the etasam verse above, 

tasam prajanam 

in (1) 

sa tvam prapadya 

in (4) 

pnisacj ca mudg’rab 

in (5) 

saiva parthaya 

in (6) 

balani sarva (analogous to viqva) 

in (8) 

trivistapam (omitting abhi) 

But the type seems to be established by bhunkte yaiko vi- 
bhuh in (7), and guha ’bhibhutasya in (9) ; so it may seem 
better to stick to the text than to adopt an explanation which 
would demand still further changes, such as omitting esa in 
(2), and vibhuli in (7) ; or rejecting the form of (9). Other 
examples of thirteen-syllable tristubhs exist, but they seem 
to belong to another category, as shown below, where, however, 
chandovidas te | ya uta na ’dliltavedah differs from adyai’va 
punya ’ham | uta vah Pandaveyah only by ctesura, the latter 
(from i, 198, 5 b) belonging here. 

Defective Tristubhs. 

Considering the extent of the epic, the number of defective 
(impossible) tristubh padas is small. Some of these I have 
already noticed incidentally, and need not take up again. The 
others I group in their order : 

i, 197, 23 d, adya ’(jesasya bhuvanasya tvam bhava ’dyah 

Omit Bhava, (Jiva (Ho. 13, hypermetric). 

1 Here B. (59) has purusejrarasya. 



v, 42, 15 d, 

na ’dhlylta nirnudann iva ’sya ca ’yuh 
Read va for iva. 1 

v, 44, 3 c, 

anarabhyam vasatl ’ha karyakale 

Omit iha (= atmany eva). 

v, 44, 25 a, 

abhati quklam iva lohitam iva ’tho 

krsnam atha ’iijanam kadravam va (v. 1. in 26 a, 
krsnam ayasarn arkavarnam). 

Read va for iva ’tlio and atho krsnain aiijanam. In 26 a, a 
like change. So v, 48, 86 d, akage ca 'psu ca te kramah. syat, 
for ca apsu. 

v, 44, 28 a-c, nai ’varksu tan na yajuhsu na ’py atharvasu 
na dreyate vai vimalesu samasu 
ratliaiiitare barhadrathe va ’pi rajan 

For c, see the list above, p. 295. In a, read naivarksu tan 
na ’pi yajulisv atharvasu, or as hj-pennetric with yajusu? 

viii, 3,338 c, ditsuh Karnah samare hastisatkaih yah 

B. 66, 30, has hastisadgavam and omits yah 

xii, 60, 46 c-d, adharo vitanali saihsrsto vaiqyo brahmanas trisu 
varnesu yajnasrstah 

The preceding padas make metre and sense. These make 

xii, 226, 18, na tat sadah satparisat sabha ca sa 

prapya yam na kurute sada bhayam 
dharmatattvam avagahya buddhiman 

yo 'bliyupaiti sa dhuraihdharah puman (v. 1. narah) 

Read (?) 

na tat sadah satparisat sabha ca sa 

samprapya yam na kurute sada bhayam 
tad dharmatattvam avagahya buddhiman 
yas tv abhyupaiti sa dhuramdharo narah 

The sa has caused the loss of the following sam, a copy-error. 
Just so, bhavatmakam parivartamanam has lost sam before the 

1 The form va for iva is found everywhere, e. g., xiii, 90, 42 c, sa vai 
muktah, pippalam bandhanad va (cyavate). So R. vii, 34, 15; 36,42. 



last word, xii, 10,544 a = 287, 13. The parallel proverb, v. 35, 
58, has na sa sabha yatra na santi vrddhah (Manu, xii, 114). 

xii, 285, 26 d, mam adhvare Qaiiisitarah stuvanti 
rathaiiitararii samagaq co ’paganti 
mam brahmana brahmavido yajante 
(d) mama ’dhvaryavah kalpayante ca bhagam 

Varied readings in xiii, 159, 16, where d appears as tasmai 
havir adhvaryavah kalpayanti, but tasmai here is offensive. 
Read me 'dhvaryavah. 

H. 2, 74, 27 b, qaqvae ehreyah kanksibliir varadameyavlrya(h) 1 

(se. ptijyase) 

v, b and ix. Mora-Tristubhs. 

v, b. In the form of the hypermetric tristubh shown above 
in tatha titiksur atitiksor vigistah or na tvam prcchami, Vidura, 
yad dhitam me, the scheme is 

^ — v, w wXl _ w _ ^ 

Now, as soon as the caesura in such a combination of syllables 
shifts back to the fourth syllable, ^ 
as in 

tesaiii kraman kathaya tato 'pi ca ’nyat, v, 42, 26 c, 

it is evident that, although such a piida may be mechanically 
equated with No. 19 (as a hypermeter), it is on the other 
hand nothing but a mora-equivalent of the form (No. 1) 

w w w w m. Again, in the case of neglected 

caesura (above), where two light syllables follow the “ extra ” 
syllable, we may as well take gaktyrstiprasamusalany ayudhani 

as an equivalent of ^ , like the regular 

pada with _ w in the second foot (No. 6) ; or, to give an 

example where the caesura is clearly marked, sa mam jihmam, 
Vidura, sarvam bravisi, iii, 4, 21 a, may be scanned as 

vj , w w _ w. Such padas stand parallel to the 

regular forms, as in the Gita, 2, 29, imitation of Katha Up. 
ii, 7: 

1 The commentator asserts that this is really a “ fourteen-syllable pada,” 
but, as nityada precedes, yayvat may be omitted, leaving a dodeka hypermeter. 



agcaryavat pacyati kaqcid enam 
accaryavad vadati tatbai ’va ca ’nyab 
aQcaryavac cai ’nam anyah crnoti. 

As resolution may take place in several places, we get quite 
a variety of rucira-like packs. The common alternation of the 
_ vj and w w packs is tlius represented : 

xiv, 10, 19 a-b, ayam indro haribhir ayati rajan 

devaib sarvais tvaritaib stuyamanab 

ib. 10, 23 c-d, ayam yajnam kurute me surendra 
Brbaspater avarajo vipramukhyab 

But the ehoriambus-equivalent is more common, as in 

iii, 134, 28 e, bastl ’va tvam, Janaka, vinudyamanah 

xiv, 26, 1 ff. (refrain), yo hrccbayas, tam abam anubravlmi 

Two or three of these packs together are not unusual : 

iii, 132, 9 d-10 a, bharyam ca vai duhitaram svarii sujatam 
tasya garbbah samabbavad agnikalpah 1 

viii, 68, 7 d and 8 a-b, phalarthinam vipbala iva ’tipnspah 
praccbaditam badicam iva ’misena 
sarhcbaditam garalam iva ’qanena 

So in the jagatl-pacla iii, 133, 10 d, kasmacl balah sthavira 
iva prabhasase. Here it needs only the iambic opening to 
make a true rucira, — KJ — \J \J KJ — \j — — , and this pada 

is found repeatedly, not in complete rucira-stanzas alone, but 
in jagatl stanzas. For example, iii, 3, 31 is a vanqasthabila 
stanza, where three packs are regular, but b has : 

praklrtayec chucisumanab samabitah 

On the other hand, in i, 34, 26, the first pack alone is of vafiga- 
stha type, while three rucira packs follow, e. g., pack d : 

mahatmanah patagapateh praklrtanat 

These are both tag-stanzas, embellishing the close of a chapter 

1 The naive padas 10,606 b-7 a, following this stanza, are omitted in B. 
The embryo here says : vedan saugan sarvagastrair upetan adhitavan asmi 
tava prasadat, etc. ! 



and of benedietive content. 1 A similar case occurs in iii, 3, 
75a, where, after praise, is said: 

imarii stavaiii prayatamanah samadhina 

pathed ilia ’nyo ‘pi varaiii samarthayan, etc. 

But this arrangement is found also apart from such employ- 
ment. So in viii, 66, 47, a, b, d are of this rueira type ; c is 
of upendra form, thus c-d : 

hato maya so 'dya sametya Kama 

iti bruvan pragamayase ( v . . 1. me) ‘dya Phalguna 

Here eleven syllables do not equate twelve (thirteen), but 

d equals ^ w w In viii, 84, 20 a, B. has sphatikacitra, 

where G. 4,281 has (tato dhvajam) sphatikavicitrakaneukam, 
probably the original, as B. is apt in varied readings to have 
the more uniform (improved) types. 

As upendra and vahgastha padas alternate, so rueira padas 
alternate with vahgasthas. Thus in xii, 244, 29, a and c are 
of rueira form ; b and d, of vahgastha form. In a stanza of 
mixed upajatis, xii, 341, 119 b has 

mahatmanah purusavarasya kirtitam 2 

The seventh book has a number of these combinations of 
rueira padas and stanzas and upajati padas and stanzas, usu- 
ally as pada tags at the end of chapters, for example, adliy. 
26, 29, 30, 32 ; but it has also incorporated complete ruciras 
as parts of an upajati system, as in 2, 15 and 16. 

I give now — reverting to the tristubh — a few more 
examples : 

ii, 58, 16 a, na ca ’kamah Cakunina devita ’ham 

iii, 4, 17 a, tvaya prstah kim aham anyad vadeyam 
iii, 4, 18 a, etad vakyaiii Yidura yat te sabhayam 

1 In xii, 219, 52, two or throe padas in a benedictiye stanza are of this type. 
The first pada in C. begins imam vah pathati vimoksanigeayam, for B.’s imarii 
hi yah pathati (ri’) moksaniycayam. In xii. 111, 21, a benedietive stanza, 
rueira padas appear in a and d, e. g., the latter: na vanmayarii sa labhati 
kiriicid apriyam. xiii, 77, 32 has a whole rueira in benediction. 

2 Compare Gita, 8, 10, sa tam param purusam upaiti divyam, etc. 



In this example, ii, 71, 17, the much affected pada sym- 
metry is shown, b and d having v w , a and c having 

w : 

atidyutam krtam idam Dhartarastra 
yasmat striyain vivadadhvaih sabhayam 
yogaksemau nacyato vah samagrau 
papan mantran Kuravo mantrayanti 

Similar is ib. 3, only the first pada is jagatl. But the second 
foot corresponds to that of the third pada ; and so the fourth 

pada has w w corresponding touu^_ in the second 


anyam vrmsva patim a^ubhavini 

yasmad dasyam na labhasi devanena 
avacya vai patisu kamavrttir 
nityarn dasye viditam tat tava ’stu 

Contrast this, for example, with the following padas, 20 a, 
24 a, 26 c, of the same section : 

Bhlmasya vakye tadvad eva ’rjunasya 
tato Gandharl Vidurag ca ’pi vidvan 
Krsnam Pancallm abravlt santvapfirvam 

The last is a pure vai<jvadevl piida, as above nityam dasye 
viditarii tat tava ’stu is a pure vatormi pada, and yogaksemau 
na§yato vah samagrau is a pure QalinI pada. 

In padas of the rucira or rucira-hke type, the same word 
appears in the tristubh, which has caused a discussion in the 
§loka : 

iii, 192, 56 d, tena qrestho bhavati hi jlvamanah 

v, 44, 18 c-d : sa tam vrttim bahugunam evam eti 
guroh putre bhavati ca vrttir esa 

xii, 300, 27 d, moghah qramo bhavati hi krodhanasya 

Here bhavati need not be pronounced bhoti, as it is a perfect 
parallel to bahu gu- in this stanza and to pacasi (bhavasi) in 
the following : 



i, 232, 14, srstva lokfuis trln. irnan havyavaha 

kale prapte pacasi punah samiddhah 
tvam sarvasya bhuvanasya prasutis 
tvam eva ’gne bhavasi punah pratistha 
A monosyllabic pronunciation cannot be claimed for all these 
cases, though it might be maintained for special words : 
i, 197, 42 a, tarn cai ’va ’gryaiii striyain atirupayuktam 1 
iii, 4, 1 c, dharmatmanaiii Viduram agadhabuddhim 

iii, 4, 3 a, evaiii gate Yidura yad adya karyam 

iii, 26, 11 d, labdhva dvijam nudati nrpah sapatnan 
iii, 34, 9 b, yathakamabi viditam Ajataqatro 
iii, 34, 20 c, mahagunaiii harati hi paurusena 
iii, 111, 10 d, vratam brahman^ carasi hi devavat tvam 

xii, 302, 114 b, maharnavaih vimalam udarakantam 

xiii, 71, 16 a, drstvai ’va mam abhimukham apatantam 
xiii, 93, 136 a, adhvaryave duhitaram va dadatu 2 

xiii, 102, 36 b, tathe ’stluaiii daqacjatam prapnuvanti 
xiii, 103, 35 b, tathai ’va ’cyan anaduho lokanatha 
H. 2, 72, 33 a, Atharvanam suqirasam bhutayonim 
H. 2, 74, 23 b, khyato devah pacupatih sarvakarma 

But the great objection to a monosyllabic pronunciation is 
that the rucira pacla interchanges up to tlrree padas with the 
ordinary tristubh pada, and must therefore be identical in 
structure ndth the same pada when four times repeated, in a 
perfect rucira stanza. But in the rucira stanza, no one can 
maintain for a moment that uwu_istobe read with crasis. 
Why then when a stanza has three packs of the same type or 
even one ? 

It may be said, however, that the mora tristubh pada differs 
in no respect from the “ inserted fifth,” when the latter is a 
light syllable. For example in this stanza: 

iii, 4, 21, sa mam jihmam, Yidura, sarvarn bravlsi 
manam ca te 'ham adhikaiii dharayami 
yathe ’cchakam gaecha va tistha va tvam 
susantvyamana ’py asati strl jahati 

1 All the other padas here are of strict yalini type, , w 


2 ib, 94, 44, idem, but ya fails. 




Here it is clear that susantvyamana ’py is a complete foot of 
the inserted fifth variety ; but pada b is indifferently an in- 
serted fifth or a inora pada, the caesura pointing but lightly to 
the latter explanation. One reason, however, against such an 
identification is that the mora explanation in almost all cases 
is indicated, as in most of the examples given, by a plain 
caesura before the fifth. Another is that this explanation 
brings the various padas of a stanza into symmetry, as in iii, 

192, where _ ^ is employed with predilection throughout, 

and we find in 

iii, 192, 69, yatha yukta | vamadeva ’ham enam 
dine dine ] samdiqantl nrqaiisam 
brahmanebhyo ( mrgayatl sunrtani 
tatha brahman | punyalokaiii labheyam 

Here mrgayatl su — , as w , accords with the structure 

of the other padas. So in jagatis, e. g., 

vii, 26, 65-66, sa nagarajah pravarankuqaliatah 

pura sapakso 'drivaro yatha nrpa 
bhayaih tada ripusu samadadhad bhrqam 
vanigjananam ksubhito yatha ’rnavah 
tato dhvanir dviradarathaqvaparthivaih, etc. 

vii, 50, 14a-b, tatha tada yodhanam ugradarcanaih 

niqamukhe pitrpatirastravardhanam 1 

vii, 109, 37 c-d nicamya tam pratyanadahs tu Pandavas 

tato dhvanir bhuvanapatha ’spread bhrqam 

Compare the close of vii, 155, four stanzas of ruciras and 
of vancasthas, with the same mora-padas. 

A third point to be noticed is that the “ inserted fifth ” as 
brevis, and with its caesura there, is always a rarity (as indi- 
cated in the fists above) unless followed by two (or three) 
other breves, so that we have finally two chief classes to ex- 
plain, one with caesura after the fifth heavy syllable, and the 
other with caesura after the fourth, followed by breves equiva- 

1 Variant on the old stereotyped yamarastravardhana, of battle, hero, etc. 
as in vii, 145, 97 d ; ib. 98 d. 


lent in morse to the rucira pada. There are a few eases 
bridging these classes and showing that the metrical equation 
was not always in harmony with the caesura, but this is no 
more than was to be expected. We are not to imag ine that 
the poets set themselves to compose padas by categories ; but 
we can hardly escape the conclusion that a pada identical with 
a rucira pada was felt to be the same with it, though the 
characteristic pause of the rucira may be absent ; for in the 
regular rucira the sense-pause and rhythmical pause are not 
always identical. Hence, when we find samanam murdlini 
rathayanam viyanti in one stanza, and yuvarii vaman vikurutho 
viqvarupan in the next, i, 3, 65 a, we may explain them as 
belonging to two categories cacsu rally distinct, or put them 
into one category, remarking that usually the caesura is after 
the fourth in such syllabic combinations ; for even with two 
breves following (the commonest case with the caesura after 
the fifth) the examples are rare in comparison with the rucira- 

like or true rucira pada, 

w v/ w ^ (rucira-like) ; w _ w KJ KJ W W 

(rucira). It is perhaps in each case merely a question of how 
the pada is naturally to be read. Some will scan only one 
way, e. g., marge bhagnam qakatam iva ’calaksam in iii, 133, 
23 d, irrespective of the stanza ; virile others may be read 
either way, as in the stanza ib. 19 : 

so 'ham crutva brahmananam sakaqe 
brahinadvaitaih katliayitum agato 'smi 
kva ’sau band! yavad enam sametya 
naksatranl ’va savita naqayami 

or when united with the five-syllable foot, as in i, 89, 20 : 

tatra sthitam mam devasukhesu saktam 
kale ’tlte mahati tato 'timatram 
duto devanam abravld ugrarupo 

dhvanse ’ty uceais trihpluteua svarena 
ix. The matra or ati-tristubh pada may even be combined 
with the pada having inserted fifth, where the breves follow- 
ing the c a; sura seem to be only rucira-like resolution. It is 
a treiskaideka measure: 



i, 89, 23 b, samlksya ce ’maxii | tvaritam upftgato 'smi 
(i, 198, 5 b, adyai ’va punya ’bam j uta vah Pandaveyah 
v, 43, 50 e, cbandovidas to | ya uta na ’dhltavedah 
xvii, 3, 13 b, yad dattam istarii | vivrtam atho hutarn ca 

In xiii, 1, 32 d, ksipraih sarpam jalii, mil bliut te vigankii, as 
compared with gaktyii rakso jalii Karnii ’dya turnam, vii, 179, 
48 e ; tapantam enam jalii papaiii niglthe, ib. 49 b, te may be 
thought to lie an intrusion, but it lias a sort of parallel in iii, 
4, 22 d, ne ’dam astl ’ty atlia Viduro bhasainanah (where C. 
has atho !). 

The mora rhythm in general is early, being found not only 
in the epic but in the I panishad and Buddhistic verse. But 
it is found also in imitative parts of the Puriinas, as in Vayu 
P., xiv, 7, in a section where upendra padas interchange with 

the galinl-like piula ( w _, _ w ). Here in 7 b-d : ma- 

hatmanam paramamatiiii varenyain, kavim puranam anugasit- 
iiram, where, as often in the epic, w ^ w w _ stands with 
and (e. g., 9a) as the equivalent, 

of the latter. On the last verse above, see the note on p. 
277. The measure appears in tristublis as an ati-tristubh of 
twelve ; in jagatls, as an ati-jagati of thirteen syllables. 

In the Rumayana I know of only one ease where this re- 
solved form is found, and that is peculiar. In R. vii, 81, 22, 
an extraordinary gloka closes the section, and in G. 88 a 
tag-tristubh of the form above is made out of it. The ex- 
traordinary gloka is: sa tiiir brahmanam abhyastam sahitair 
brahmavittamaih, ravir astam gato Rama gaceho "dakam upa- 
sprga, “ the sun has set (after accepting as a laudation) the 
secret worship by the assembled Veda-versed (seers)," ac- 
cording to the commentator. The parallel in G. indicates a 
brahmanair abhistutah instead of brahinana = upanisad or 
pujii. The tag-end in G. vii, 88, 22 seems to lie from a 
phrase just preceding (found in G. and R.), samdliyfun upiisi- 
turii vlra (Rama). The whole tag reads: 

1 As remarked above, p. 290, this, thoush inserted here on account of its 
likeness to the next example, belongs rather to the group of Double Hvper- 



abhistutah suravarah siddhasangiir 
gato ravir surucirain astaeailam 
tvain apy ato Raghuvara gaccha saiiidhyani 
upilsitum prayatainana narendra 

This may U* called a rueirii-tristubh. On the rucira stiinza, 
see the section on aksaracchaiidas in 'low. 

The Tristubh Stanza. 


As stanzas, the fonns that liegin with a iliiamlt and con- 
tinue with a choriamb are not particularly common. They 
are generally mm lilted as upajdtis, by combination with the 
iwira varieties, which Ix-gin with a spondee, indiavajrii and 
indra vailed. Sometimes the j>erfr< t form appears as a mere 
later addition. Thus in iii, 23, only one stanza, 14, is ujien- 
dra in sixteen upijfitis ( pada a has final brevis). So iii. 111, 
IT— IS = 10,044 : while in iii, 2‘.'.>. 9 and 10 are two jterfcct 
vatieasthas, interpolated among elokas. In iii, 232, 14, an 
almost perfect 1 njxmdra is eii'i om ed in a stilt i of Skauria, 
where the environment is upajati. Again, in iii, 23*!. in an 
upajati system of thirty-one stanzas, one. Id, is pure upendra, 
except that pada a ends in a brevis ; and 19. 2d, 27 are also 
pure ujiendras, 2 except that in 27. [Ada a ends in brevis. In 
xii, 201, out of twenty-seven tiiytubhs, two, 0 and 23. are 
pure upendras. A pair of pada' oeeiirs in viii, S9, 47. tato 
mahliii sagaramekhalaiii tv.uh sapattanaiii giamavatliii simrri- 
dlifitn. Rut two pftdas together is a large nmiilrr ex< ept in 
late passages, like iii, 170 and 177. when- thev are not uncom- 
mon (170. 7, Id. 10; 177. 11. 21, 22); vanea'thas in vii, lop, 
36-37, with a rucira pada, etc. 

As the vaneasthafbila ) is merely an ujiendra with a sylla- 

1 Tin- third pada, however, uni- in a brevis On tlu* [mint is to be notice! 
that such a brevis is not uncommon in the Bharat a, lmt in the Itamavana is 
rare enough to deserve a special notiie of K, vi, Tt, Vt, where , v, ry pada 
ends in brevis, flere the stanza it«elf i e nperelra. but the syspni is upajati. 

2 Here only ebyht padas are not of upajati form, but 



ble added, 1 so the jagati corresponds to the different forms 
of the tristubh. Thus in i, 197, 25, it takes the place of a 
vatormi, tatra hy ekam bhavitaro na samgayo, yonim sarve 
manuslm avigadhvam, tatra yuyam karma krtva ’visahyam, 
bahun anyan riidlianam prapayitva ; and just below, 53 b, 
pancanam eka svakrtene ’ha karmana, where it is hyper- 
metric. Here a and d have eleven, c, 12, and b, 13 syllables. 
A near approach to a perfect vaiigastha is found in i, 198, 8, 

where all four padas are normal, except that in b, w _ 

takes the place of the opening diiamb. In ii, 64, 5, all padas 
are perfectly regular. The interchange of an occasional 
vangastha pada with the other padas of an upajati tristubh 
is too common to call for further remark. Two instances 
will be found in i, 193, 20 and 22. In the former, the stanza 
would be a perfect upendra, but pada c is of vangastha form ; 
in the latter, which is an upajati tristubh, pada c again is of 
pure vangastha form. So in i, 197, 11, an indravanga pada 
heads and closes a tristubh stanza. The caesura is after the 
fifth or fourth, passim ; or after the sixth, as in i, 197, 17 a, 
yada tu paryiiptam ilia ’sya 2 krldaya; or a second occurs, 
as in iii, 5, 19 c, saiiivartlhayan stokam iva ’gnim atmavan. 
The sixth place is often half as common as the fifth. 

The caesura in the padas of the upajati system is found 
most frequently after the fourth or fifth. The former, per- 
haps, in isolated padas, as in xii, 64, 18 d, tatas te 'ham dadmi 
varan yathestam, and i, 92, 9 a and 11 d; but the forms in the 
Bharata, though inclining largely to the fifth place, vary con- 
stantly, as they do in the Ilamayana. Examples from the 
latter have been given above hi the introductory paragraph. 
I add some specimens from the other epic : 

tad vai nrcarisaiii tad asatyam ahur 
yah sevate dharmam anarthabuddhih 

artho 'py anlgasya tathai ’va rajan, i, 92, 5 a-c 

nllotpalabha suradevate ’va 

Krsna sthita murtimati ’va Laksmlh, iv, 71, 17 c-d, 

1 That is for \j \j , mechanically considered. 

2 On the light syllable before mute and liquid, see above, p. 242. 



where a-b have caesura after the fifth and fourth respectively. 
Not infrequently where the tristubhs pause after the fifth, 
the jagatl, in the same stanza, pauses after the fourth, as 
in iii, 268, 19: 

saqankhaghosah satalatraghoso 
gandlvadhanva muhur udvahanq ca 
yada qaran arpayita tavo’rasi 
tada manas te kim iva ’bhavisyat 

But in pure vahqasthas, the caesura is apt to vaiy almost with 
the pada, as in xii, 103, 40 : 

na samadandopanisat 1 praqasyate 
na mardavaiii catrusu yatrikam sada 
na sasyaghato na ca samkarakriya 
na ca ’pi bhuyah prakrter vicarana 

So in viii, 18, 12, the c a: sura of two padas falls after the fourth 
and fifth respectively, and then comes the pada : atlva cukso- 
bhayisur janardanam; while the fourth pada is cut after 
the fourth syllable. Alternation is common, as ib. 14-15 

(w _ w and W \J alternately). Sometimes there is 

no caesura: 

qarasicjaktyrstinipataduhsaliarn, viii, 88, 3-4 

or it is irregular : 

alam virodhena ! dhig astu vigraham, ib. 21 b. 

krtyam atharvangiraslm ivo ’gram, viii, 91, 48 = ix, 17, 44. 

Upajatis are sometimes used to close systems, as are also 
upendras and vanqasthas. Pure vahqasthas may end a system 
of upajati tristubhs, as in viii, 76 and 79, xii, 167, 49-51, just 
as upajatis close a scene composed in old tristubhs. The 
analogy with the tag-measures (discussed below) is here com- 
plete; the scene is set off with something better than the 
ordinary. As an example of the way in which upajatis are 
thus used may serve the end of iii, 154; or in i, 197 and 198, 

1 Upanisad is here secrecy. So perhaps in xii, 271, 30, (apetatrsnanam, 
etc.) caturthopanisaddharmah sadharana iti smrtih. 



where the first part of the wedding scene at Drupada’s is in 
irregular old tristubhs, but regular upajatis conclude the 
scene ; the latter beginning just where the actual wedding is 
described, and taking in the statement that the heroine was 
first married to Yudliisthira, then to the other brothers ; that 
she preserved her virginity day by day ; and that Drupada in 
conclusion gave most extravagant gifts. The smoothness of 
the statement babhuva kanyai ’va gate gate 'hani, etc., 198, 14, 
stands with its surrounding verses in at least metrical contrast 
to the part that goes before, where tristubhs of vatorml and 
cfdim padas and every sort of irregular combination is the 
rule. Whether the uniform upajatis conform to the uniform- 
ing of the poem is certainly a proper question to raise, though 
no signed and sealed statement to that effect is extant. 

Another interesting example will be found at the end of the 
gambling scene, where from ii, 67, 24, almost regular upajatis 
continue to the end. This happens to be the passage where 
the heroine puts the legal question to which Blnstna is un- 
able to reply, and where Kama joins in the laugh. The 
question is implied in what follows (68-70), but the passage 
in its present form is certainly open to the suspicion of having 
been rewritten by a more modern hand. 

The first chapter of the Rsya^-ruga episode is in old tri- 
stubhs. With the beginning of the sensuous description in the 
second chapter begin the upajatis, iii, 111, 112. 

In the systems of the older epic, i£_w_, w w , _ w , 

and _ w w are used as interchangeable second feet. So uni- 
versal are _ w and w vy that they must be considered 

as the chief tristubh measure of the older epic, greatly in 

excess of \J W But in the fourth book and most later 

parts, these recede before the upajati forms. Jagati padas are 
inserted occasionally in all the free tristubh sections. 1 It is 
perhaps worthy of remark that, for example, in the Dyuta 
Parvan, the diiambic opening, or even, it might be said, the 

1 The process elsewhere of making a jagati pada is sometimes patent, as 
in viii, 90, 72 d : bhindhi tvam enam Xamucirii yatha Harih (for yath ’enurah) ; 
here in an upajati system of jagatls. 



whole upendra form, is found par excellence in the final 
pada(s), though found also in a, b, c, especially as the section 
gradually passes (towards its end) into regular upajatis, 67, 
26 ff. For example, at the beginning of the scene, 56, 12-16 : 

12, v-/ w , \j w v-/; w , \j \j \j w ; 

\j \j ; — w , \j w w 

13, w , \j kj \y ; w — , — w — o' ; 

\j , w w w o; w , w o o w 

14, , w o ; o' — , — o' o' o' w ; 

o' , O' O' O' ; O' O' O' o/ O' 

15, o' , o' o' ; o' — , — w o' — o' o' ; 

O' , o' o' o' ; o' — o' , — o' O' o' o» 

16, w ^ o' ; o' — w — , — o' o' o' ; 

o' o', w w o' o' ; o' o' , — o o' o o' 

I have remarked in the list of examples given above that 
some of the older forms of the tristubh are practically confined 
to the early parts of the epic. The fourth and seventh books 
are considered to be late, or, what amounts to the same thing, 
modern expanded forms of older material. The middle foot 

w w _ occurs not infrequently in the older epic, but in the 
whole fourth book it occurs but once, and in the seventh only 
twice in 1280 padas. Upajati systems, except, as just ob- 
served, as a sort of tag, are not frequent in the older epic, 

where the systems are of the type _ v and w w with 

interspersed choriambs. The latter part of the third book, 
however, and all of the fourth book prefer the upajati system 
(the coesura being after the fourth in only one-third of the 
cases in the latter), and blocks of upajatis appear in the much- 
expanded battle-books. As a system, the upajati marks late 
passages, such as the song of Cii in the eleventh section of the 
thirteenth book, ancl the praise of gifts in the fifty-seventh sec- 
tion of the same book, where only two padas are not upajati. 
This book is also marked by the large number of its §alinl 
stanzas (not single paclas), which keep up an old measure in 
a new fixed form. Old as is the choriambic pada, the stanza 
form of the choriambic tristubh employed in great groups to the 
exclusion of other forms of tristubh appears to be an innovation. 
A form once given persists, and so we have late passages with 



— v as the second foot, just as this and w w are still 

met here and there in the Pur anas, 1 but when the choriamb is 
employed continuously in a long system 2 the passage may be 
set down as late, or, if one prefers the expression, as more 
refined, as in the whole Ramayana and in the later books of 
the Mahabharata. 

Another mark of lateness appears to lie in the absence of an 
indiscriminate mixture of tristubh and jagatl forms. Later 
passages are rather apt to show uniformity in this regard ; 
earlier passages show none, though an harmonious com- 
mingling in alternate or chiastic form (12 + 11 + 12 + 11 or 
11 + 12 + 12 + 11) is at all times somewhat affected, and late 
passages sometimes show no uniformity ; but the tendency is 
in the other direction. 

The Syllaba Anceps. 

In respect of the pada syllaba anceps, the epic permits this 
not only in free tristubhs, but also in upajatis, and even in 
isolated pure choriambic stanzas. But even the classical poets 
share this freedom. That is to say, as Professor Capeller has 
shown, although the rule is that pure upendras and indravajras 
or the corresponding jagatis shall have final anceps only at the 
end of the hemistich, yet if these stanzas, though complete 

1 Solitary falini stanzas also occur in the Puranas. For example, Vayu P. 
vi, 71, repeated in ix, 113, where occurs the stanza: vaktrad yasya brahmana 
samprasutah, yad [tad] vaksatah ksatriyah purvabhage, vaipyaf co ’rvor 
yasya padbhyam ca fudrah sarve varna gatratah samprasutah, a pure palini. 

2 A choriambic verse or stanza is a different matter. This may be as old, 
or older, than a corresponding stanza of other form. For example, the prose 
proberb of Gaut. xxiii, 29, appears in the form panca ’nrtany ahur apatakani 
first in Vas. xvi, 35, as an upajati stanza. The oldest version in the epic is in 
i, 82, 16, where there is no exception in the case of a teacher, as in Gautama 
(for an untruth here is a mortal sin, not venial), nor is the priest included, as 
in Vas.; but the five venial lies are in case of wounds, about women, in case 
of marriage, death, and robbery, couched in upajati. A second form occurs, 
however, in xii, 165, 30, where the teacher is mentioned in the same way as 
is the priest. The other difference between the epic versions is that the latter 
begins na narmayuktam anrtam hinasti ; the former, vacanam hinasti, as cho- 
riamb. Spruch 3,321 has only one of these forms (ascribed to a Purana), 
Manu, viii, 112, is in floka. 



in themselves, form part of a general system of upajatis, the 
freer form is permitted. Thus in Raghuvan§a vii, 9, a pure 
indravajra occurs with the third pada ending in brevis, but it 
is in an upajati chapter. Examples from epic poetry are : 

ii, 56, 21 a (a tristubh stanza), tato vidvan Viduram man- 

ii, 63, 10 a (ditto), jammahe devitam SaubalasyS 

So in these pure choriambie stanzas, found in a general 
upajati system: 

iii, 176, 7, tava pratijnam Kururaja satyarn 
ciklrsamanas tad anupriyam ca 
tato na gacchama vanany apasyft 
Suyodhanam sanucaram nihantum 

and ib. 15, 

tava ’rthasiddhyartham api pravrttau 
Suparnaketuc ca Cincq ca napta 
tathai ’va Krsnah pratimo balen£ 
tathai ’va ca ’ham naradeva varya 

iv, 11, 9, c, Brhannalam mam naradeva viddhl 
ib. 54, 17 a, cacara saihkhye vidiqo diqaq ca 
This is veiy rare in Virata. In jagati : 

iii, 268, 19 c, yada qaran arpayita tavo’rasl 
xiii, 70, 9 c-d, tvaya pura dattam it! ’ha quqruma 

nrpa dvijebhyah kva nu tad gatahi tava 

Examples in the Ilarivaiipa may be found at 2, 95, 1 ff. (= 
8781 ff.) ; ib. 6a; ib. 10 and 11c: ib. 14a and c; ib. 24c; 
ib. 29 c (na vetsi) ; 2, 124, 53 a (= 10,625), etc. 

Epic usage, however, keeps the final syllable long in the 
prior padas. Exceptions like those just given are not uncom- 
mon, but are distinctly exceptions. I have no statistics, but 
perhaps the general condition may be stated well enough in 
saying that one has to hunt for final breves in prior padas of 
pure upendra and indravajra stanzas and does not have to 
hunt for final longs ; while in upajatis the final breves are not 
so uncommon as in the pure stanzas of uniform type. 



In this regard I see no special difference between the two 
epics. Perhaps the Ramayana poet is a little more shy of the 
brevis but it occurs there also, not only in pure upendras 
standing in an upajati environment, but even in isolated tag- 
stanzas where the upendra stands alone. Thus where G. ii, 
33, 27 has a varied reading which converts the stanza to an 
upajati, the Bombay text of R. ii, 33, 29, presents (in an 
upajati enviromnent) a pure upendra stanza, with the first 
pada ending in brevis, pratlksamano 'bhijanam tadii ’rt&m. 
Another example will be found in vi, 69, 92 = G. 49, 77. In 
upajatis it will be enough to refer to R. ii, 15, 44 a; 21, 52 c; 
37, 34 a; 36 a; v, 28, 4 ff., etc. In the case of isolated tag- 
upendras, examples may be found in R. vi, 61, 39, where c 
ends in a brevis, although the isolated stanza is pure upendra, 
and in R. ii, 115, 24 (not in G.), where both a and c end in 
breves : 

tada hi yat karyam upaiti kiriicld 
upayanam co ’pahrtam maharham 
sa padukabhyam prathamam nivedyS 
cakara paccad Bharato yathavat 

One fact seems certain from the treatment of upajatis 
versus upendras and indravajras or vahgasthabhilas and indra- 
vaiiQiis, namely, that the native metricists in calling the upajati 
a mixture of upendra and indravajrii or of vahr-astha and 
indravahya, and treating it as a derived form are historically 
incorrect. Of course, the upajati stanza is a stanza in which 
some padas are of one type and some are of another; but it 
is not a mixed development from pure stanzas of either type. 
On the contrary, the upajati is the prius, and the pure upendra 
and pure indravajrii stanza is a refinement on the mixed type. 
Historically the choriambic tristubh begins with syllaba anceps 
like the f^loka, 1 and upendras and indravajras are differentia- 

1 For this reason, in the Illustrations, though giving examples of each, I 

do not separate (as is usually done) the types of opening, e. g., vy \j 

and \j . Only in complete forms of stanzas, like the calinl, vatormi, 

and rucira, is the first syllable fixed. In the free tristubh and upajati stanzas 
the initial syllable is quite indifferent. Then comes the upendra stanza, 



tions of the earlier mixed types. They had the same devel- 
opment as had the galinl, which began, as in the epic, with 
id _ bL and settled at last into as a first foot. 

The upajati stanza in its turn is derived (as a more refined 
form) from the mixed tristubh of the early epic type, which 
unites into one stanza not only padas of the choriambic type 

and of the types _ w , W W f WWW j but also of the 

type of the rucira or mora-pada ; of which, together with the 
special stanzas of fixed form derived from these measures, I 
shall speak hereafter. It is to be observed that this mixture 
of vatormi, galini, choriambic, and resolved-syllable padas in 
one stanza is Vedic and Bliarataie, non-classical and non-Rama- 
yanan, 1 but also, in a very limited degree, Puranic. That is to 
say, the Bharata, the oldest extant Purana, on the one hand 
preserves the old Vedic type, winch is still kept up in a 
measure in the later Puranic diction, while on the other it 
has the clear-cut upajati system favored by Valmiki, the 
former both in early and late parts; the latter only in late 
parts, according as the different poets preserved the old style, 
or, like Valmiki, cut loose from it and wrote only in upajati 

Emergent Stanzas. 

Of peculiar interest is the growth of the completed stanza 
of other tristubh forms. In the great epic, we can, as it were, 
see the gradual emergence of the complete §alinl, vatormi, and 
vai§vadevl stanzas (of four identical padas) from the single, 
double, and triple pada of this form in tristubh stanzas, 
till at last a few complete galinl stanzas are found and one 
perfect vaigvadevl. 

The occasional pada is indefinitely antique. It is the four- 
fold-combination that is emerging; just as upajatis emerge 
from mixed tristubhs, and upendras from upajatis. In the 
completed refined pada the opening is spondaic ; in the erner- 

v w , as distinguished from the indravajra, w f both secondary, 

not as padas, but as stanzas, to the upajati. 

1 The Vedic usage is illustrated in Kiihnau, Die Tristubh-Jagati Fainilie, 



gent type it is indifferently iambic or spondaic like other 
tristubh forms. This sporadic appearance calls for no special 
remark here, as examples may be found in the list of ex- 
amples of tristubh padas. The first stage beyond this is 
where two padas appear of half-Qalini form but with iambic 
opening. This is either “ regular ” or hypermetric, 1 as in 

iii, 5, 16 c-d : yatha ca parne puskarasya ’vasiktam 

jalarii na tisthet pathyam uktaih tatha ’smin 

The hypermetric Qalini pada of this sort (vaigvadevl) is 
common, as in i, 55, 12 b, trata loke 'srnihs tvam tathe ’ha pra- 
janam (so ii, 77, 10 b, etc.), as shown below. 

Again, in mixed tristubhs, where we have half a stanza of 
almost pure Qalinl form, as in vi, 3, 65 c-d ; or even an almost 
complete stanza, as in 

i, 58, 19 : etac chrutva priyamanfih sameta 

ye tatra ’san pannaga vitamoliah 
Astlke vai prltimanto babhuvur 
ucus cai ’nam varam istaiii vrnlsva 

Here the Qalini is complete save for the last pada. So in 
iii, 4, 4, there is a perfect Qalini save for the first syllable of 

а. In iii, 5, 13, the stanza is nearly vatormi, but three padas 
begin with a short syllable and the first has the Qalini trochee. 
In v, 40, 29, three padas are pure Qalinl and one is vatormi. 
These forms are often symmetrically united. Thus in i, 58, 
20, the padas run Qalinl + vatormi + Qalinl + vatormi, save 
that in b and c the third syllable of the first foot is brevis. 
Sometimes the arrangement is chiastic, as in i, 197, 30, where 
the padas are vatormi — Qalinl, Qalini — vatormi, etc. These 
forms are again mixed freely with upajati padas, as in i, 187, 

б, this combination being too common to need further illus- 
tration. The vatormi or Qalini pada often closes the stanza 
in such a combination. Thus in i, 76, 47, a is upajati, b is 
jagati, c is Qalini, d is vatormi ; ib. 64, d is Qalini, the others 

1 This form is sometimes effaced by varied readings. Thus in vii, 54, 43, 
papena ’tmanam maj jayisyaty asantam, of C. 2116, appears as pape ‘tmanam. 



are upajati padas. Alternation of upajatis, 9 alinl-pada tri- 
stubhs, and §lokas is found in the dramatic (Jamgopakhyana, 
i,' 232, 8 ff. 

A goodly number of specimens of stanzas showing a close 
approach to the §alinl is found in vii, 2, where finally, in 26, 
appears one whole 9 alinl stanza : 

aqvan agryan pandurabhraprakaqan 
pustan snatan mantraputabhir adbhih 
taptair bhandaih kancanair abhyupetan 
qlghrau Qlghram sutaputra ’nayasva 

So in vii, 54, 40 ff., there is a number of almost complete 
and quite complete qalinls. 

A complete 9 alinl occurs in i, 58, 21 ; another in v, 33, 
115 (toavard the end of adhy. 40 there are 9 alinl padas). 
The usual order in the epic, however, is a mixture of single 
padas. The pseudo-epic, on the other hand, heaps up com- 
plete qalinl stanzas. Thus in a little system of ten stanzas at 
xii, 24, 25 ff., qalinl, vatormi, and upajati padas are all mixed 
up together but lead up to perfect qalinl stanzas in 29, 30, 32. 
In Anuqasana, complete stanzas are common, e. g., xiii, 73, 
39 ; 77, 31 and 33 (with a rucira between), on giving cows 
to priests. In (Siiti may be compared also xii, 63, 9-10 (two 
complete 9 alinl stanzas); 259, 39-42; 319, 86 ff. (five out of 
seven stanzas). The prior pada of the hemistich may end 
in brevis, as in some of the last examples, e. g. in 319, 89, 
where the stanza from a Brahmanic point of view is as late 
as the sentiment: 

sarve varna brahmana brahmajae ca 
sarve nityam vyaharante c;l brahma 
tattvam qastram brahmabuddhya bravlmi 
sarvam viqvam brahma cai ’tat samastam 

The vatormi stanza, if I am not mistaken, is not yet com- 
plete in the epic ; but its padas come near to making a com- 
plete stanza, as in vii, 201, 78: 



astausarii tvam tava sammanam icchan 
vicinvan vai sad ream devavarya 
sudurlabhan dehi varan mame ’stan 
abhistutah pravikarslQ ca mayam 

still closer in ii, 58, 12 : 

ke tatra ’nye kitava dlvyamana 
vina raj no Dhrtarastrasya putraih 
prcchami tvam Yidura bruhi nas tan 
yair dlvyamah qatacjah samnipatya 

The hypermeter is not so common as that of the galinl. A 
case occurs in iii, 134, 14 b : sapta cchandansi kratum ekam 
vahanti; and another, ib. 12 b: yajnah paiicai ’va ’py atha 
pancendriyani. So in ii, 77, 7 a ; v, 35, 42 a. The last case 
reads : 

nai ’naih chandaiisi vrjinat tarayanti 
(in 43, 5, as : na cchandansi vrjinat tarayanti) 

As said above, the isolated viiigvadevl pacla is not unusual. 
Such padas are reckoned as tristubh padas, as in i, 1, 205 c ; 
216 a; and so very often elsewhere. For example, xii, 319, 
91 d : 

ajnanatah karmayonim bhajante 

tam taiii rajans te yatha yanty abhavam 
tatha varna jnanahlnah patante 
ghorad ajnanat prakrtaiii yonijalam 

In i, 1, 212 b, there is a pada identical with this save that it 
has initial brevis, hatan Pancalan Driiupadeyaiig ca suptan, 
followed in 217 d by a pure pada, tasmin samgrame bhairave 

ksatriyanam. In i, 89, 12 b, w , _ w w _ w, the 

vaigvadevl appears as an irregular hypennetric Qalini. This 
stanza is almost a vai<jvadevl : 

anityatam sukhaduhkhasya buddhva 
kasmat samtapam Astaka ’ham bhajeyam 
kim kuryam vai kirn ca krtva na tapye 
tasmat samtapam varjayamy apramattah 



Half a complete stanza appears in xii, 292, 22 (a tag) : 
rajna jetavyah gatravac co ’nnatag ca 
samyak kartavyam palanaiii ca prajanam 
_ agnic ceyo bahubhic ca ’pi yajnair 

antye madhye va vanam aqritya stbeyam 
(where the scolius is w, above, p. 280) 

A vaigvadevl pada appears also in a benedictive verse in 
viii, C. 5,045 d, gudra arogyam prapnuvanti 'ha sarve, but 
B. 96, 63 has gudrii ’rogyam. The complete stanza occurs but 
once in the great epic and twice in the Rtimayana, as will be 
shown in the next section. 

The Fixed Syllabic Metres. 

The title aksaraeehandas or its equivalent, varnavrtta, 
“ syllabic verse ” covers, properly speaking, all metres fixed 
by syllabic measurement, but it, is used only of such stanzas 
as have a fixed number of syllables arranged in a fixed order 
in each pada, all four padas being alike. The cloka, therefore, 
is not included, nor the free tristubh of the Mahabharata. 
On the other hand, the tristubh in several of its fixed forms, 
when these are used throughout the stanza, is an aksara- 
cchandas. Such are the upajiiti forms, the giilinl, viitorml, 
vaigvadevl, and ruc-ira. In the scheme of classical metres, 
there are from twenty to thirty each of such hendekas and 
dodekas, called tristubhs and jagatls because of the number 
of syllables in them. 

Of this large number, about a dozen are found in epic 
poetry. They include those just named, in regard to which it 
wall be necessary to speak further only of two, the vaigvadevl 
and rucira. Besides these, the additional epic rhythms of this 
class will now be reviewed, arranged, according to their sylla- 
bic value, as tristubh, jagatl, atijagatl, cakvari, atieakvarl, and 
atidhrti, that is in stanzas of four padas, each pada having 
eleven, twelve, thirteen, fourteen, fifteen, and (in the epic 
case) nineteen syllables, respectively. They are distributed 1 
between the tw'o epics as follows : 

1 On their numerical distribution in the Mahabharata, see below. 




In M alia.bha.rata 
and Ramayana 

' (upajatis and their components) 
■ rucira 
, malinl 

In Mahabliarata 

gardulavikri dita 

In Ramayana 

( mrgendramukha 
\ asambadha 

The upajatis, including their four components, as also the 
galini and vatormi, have already been sufficiently discussed. 
The vatormi does not reach stanza form, but its pada is fre- 
quently found alone, duplicated, or trebly; the last case, 
which is rare, giving three-fourths of a complete vatormi. 
The galinl is found not only often in pada form but occasion- 
ally as a complete stanza, sometimes grouped in small numbers 
in the later books of the great epic. The jagati pramitaksara 
padas, isolated in the (ainti Parvan, will be spoken of be- 
low, under the head of matrachandas ; where will be discussed 
also the free praharanakalitii found in the same part of the 


Having eleven syllables to the pada, this metre is called a 
tristubh. Its scheme is __ w w ^ w w — w — ; for ex- 
ample, tasya taj janayati ’ha sarvatah. It may be regarded, 
therefore, as a jagati without the initial syllable, its final 
diiambus giving the true jagati cadence. Compare under No. 
19 : (ku-)lambharan anaduhah gataih catan. There are three 
and one half stanzas of this rhythm, though the actual occur- 
rences are more numerous ; but the same stanza appears re- 
peated. Thus xii, 250, 13-14 is a repetition of xii, 194, 61-63. 



Here there are two and one half stanzas, arranged in B. in 
groups of four, two, and four padas ; in C., as four, four, and 
two ; as if the hemistich were a whole in itself. In xii, 286, 
46, one of these stanzas is repeated again with slight changes. 
In the first instance, the group forms a tag with an apara- 
vaktra, as it does also in the second instance ; while in the 
third it appears in the same way after a puspitagra. The 
third separate stanza of this sort is found as a tag after glokas 
in xii, 247, 23. All these cases are regular ; only the hemi- 
stich ends in brevis. The metre is found only in ^anti Moksa 
and not in the Ramayana. The last case may serve as an 
example : 

yac ca te manasi vartate paraih 
yatra ca ’sti tava sarinjayah kvacit 
cruyatam, ayam aham tava ’gratah 
putra kim hi kathayami te punah 

The (mecaningless) diiambic name may at least be a reminder 
of the rathoddhata’s presumably original opening, and its 
diiambic close. 

Bhuj am gapray ata. 

This twelve-syllabled rhythm is called a jagatl, but it has 
the final tristubh cadence. The latter part of the pada is in 
fact identical with that common tristubh form which has the 

middle and end _ w w m ; but before this are five 

syllables, the fourth being a brevis w w Such a form 

as this, however, is actually found in one text as the pada of a 
hypermetric tristubh, as already pointed out (p. 289), and is 
nearly equalled (long initial) in the corresponding pada, 

na tranam labhet tranam iccban sa kale, 

But the specimens in the epic show that the caesura is not that 
of the pada just cited, but rather that of a series of bacchii : 

sa adih | sa madhyah | sa ca 'ntah | prajanam 
anadyo hy amadhyas tatha ca ’py anantah 

This metre appears once as a tag in a Tirtha story, ix, 41, 40, 
and twice in the twelfth book in an identical hymn in the 



middle of two chapters, xii, 341, 100 and 343, 90, the first 
and third padas of each version being those just cited, one 
being a repetition of the other with variations. 

There is only one case in the Ramayana, vi, 77, 24. In R. 
this is part of a tag after a puspitagra, which appears in both 
texts, while the bliujamgaprayata itself is lacking in G. Here 
also the cadence is distinctly baechiic : cacale 'va co ’rvl | 
papate ’va sa dyaur | balarii raksasanam | bhayam ca ’vivega. 1 
In both epics, the hemistich alone has final brevis (anceps), 
as above, and in ix, 41, 40 d, dhrtatma jitatma samabhyaja- 
gama. This metre is expressed by its name bhujariiga-praya- 

tam, ‘ the snake-shde,’ w w which, in the stanza, is 

repeated (as a whole) eight times. 


This measure, having twelve syllables to the pada, is called 
a jagatl. But although it ends as well in a diiambus, it is yet 
far from the cadences already examined under the name of 
jagatl. The rhythm is in fact dactylic, so that the trisyllabic 
measurement suits it ; but the first foot has a tribrach as a 
substitute for a dactyl, and the final syllable is long: w w 

__ w __ w w , _ Only two of these stanzas are found 

in the great epic, and none in the Ramayana. The two are 
together in vii, 184, 47-48 ; the latter, for example, as follows: 

pravisrtah kuirmdakarabandliavah 

These are not exactly tags, but they are close to the end of 
the chapter. The prevailing ccesura 2 may indicate that the 
metre is a catalectic form of tristubh with resolved opening; 

1 A rough English equivalent would he (of the hymn): “Beginning and 
midst he, and end of creation (of the description) : “ and terror then entered 
the huge host of demons.” The trisyllabic native measurement is here the 
most accurate. 

2 The last pada above may of course be read as anapaestic with anacrusis ; 
the preceding, more naturally, with dactylic cadence. 



but this genesis is by no means so certain as in the case of 
other tristubh derivatives. To judge from the epic, it is a 
later metre, and may be either an experiment in resolution (of 
No. 2), or a new independent invention. It is not necessary, 
I conceive, to derive every metre from some other, and I 
incline to the latter view. All the piiclas in the two epic 
specimens end in heavy syllables. The sound drutavilambitam, 
www — w w, umy serve to remind of the opening cadence; 
but the other form of the name (in “tarn) really agrees with 
the meaning, “rapid and dilatory,” indicating the beginning 
and end of the pada. 


Rare in both epics, this metre occurs but once in stanza 
form in the Mahabharata, a tag followed by a supplementary 
tristubh. The first hemistich end in brevis. Sporadic padas 

of the vai§vadevl type, , _w w__xi, are not 

infrequent. The twelve syllables do not make a jagati, 
though the metre is so called, but a hypermetric tristubh of 
the type described above (see No. 7). The native method 
of measuring by trisyllables in all cases is well shown in this 
metre to be absurd. For example, in the pada cited above, 
Krsnam Pancalim abravit santvapurvam, the caesura and 
natural division is in groups of five and seven syllables 
respectively. So in the one stanza of the great epic, xii, 
291, 25 = 10,721 (Moksa) : 

bhlru rajanyo, brahmanah sarvabhaksyah 
vaiqyo 'nlhavan, hlnavarno 'lasaQ ca 
vidvanq ca ’qllo, vrttalunah kullnah 

satyad vibhrasto brahmanastrl ca tusta 1 

(26, rSgl yuktah pacamano 'tmahetor 

murkho vakta nrpahmaiii ca rastram 
ete sarve qocyatam yanti raj an 

yac ca ’yuktah snehahlnah prajasu) 

1 This is the reading of B. In C., brahmanah stri ca dusta. 



In the Ramayana, a single pada is found in R. (above), and 
one whole stanza also (lacking in G.), v, 65, 28 (both hemi- 
stichs ending in brevis). There is, further, a half stanza in v, 
63, 33, united with a hypermetric tristubh hemistich, not in G. 
but following a tag-tristubh common to both texts ; an inter- 
esting example of the equivalence of the vaigvadevl and free 
tristubh padas : 

pntisplutaksau samprahrstau kumarau 
drstva siddharthau vanaranam ca raja 
angaih prahrstaih karyasiddhim viditva 
bahvor asannam atimatraih nananda 

For the two padas of the second hemistich, see Nos. 6 and 13 
in the Illustrations of tristubhs. The only difference between 
them and the vaigvadevi lies in the syllables marked short. 
For another form of vaigvadevi, see the malinl below. 



Of the fifty-one stanzas of ruciras in the Mahabharata, 
almost all are regular. One or two slight irregularities 
occur in the thirteen cases found in the Ramayana. Inde- 
pendent packs of this type scattered among ordinary tristubh 
padas are not uncommon in the former epic. They have been 
discussed above as mora-jagatls and tristubhs. The type of 
the pure rucira, w _ w has long been 

held 1 to be merely a jagatl with resolution, and, as was said 
above, this seems to be the only possible explanation of the 
pada, whether it happens to occur four or three times, twice, 
or only once in a stanza. 

Less common than the substitution of a rucira pada for a 
tristubh or jagatl pada, yet still not infrequent, is the har- 
monious alternation of padas. The converse of the former 
case is found in the occasional substitution of a vaiigasthabila 
pada in regular rucira stanzas, as in the group of ten tag- 

1 Gildermeister, in Lassen’s Anthologia Sanscrita, 2d ed., p. 124 ; Jacobi, 
ZDMG., vol. xxxviii, p. 607. 



me inis at i, 19, 22-31 (hemistichs end in brevis; so in xii, 
52, 34). Here the stanzas are all regular rueiras, four padas 
each of the type given above (final anceps only at the end of 
the hemistich), with the exception of stanzas 27 and 30, in 
which the second padas are van§asthabila-padas ; thus, 29-30 : 

tato mahim lavanajalam ca sagaram 
mahasurah praviviqur arditah suraih 
viyadgatarii j valitahutaqanaprabliam 
sudareanam parikupitaih nicamyate 
tatah surair vijayam avapya Mandarah 
svam eva degum gamitah supujitah 
vinadya khaiii divam api cai ’va sarvaqas 
tato gatah saliladhara yatkagatam 

In i, 23, 21-26, there are six tag rueiras, as stuti, but . in 
stanza 23 only one plida is of rucira form, the others being 
upajatis; while in i, 34, 26, one vangastha pada is followed 
by three rucira padas. 

It is very unusual to find this stanza except as a tag, as in 
the examples just given. 1 In i, 56, 1, however, is found a 
stanza consisting of one rucira pada and three tristubh padas, 
the first being peculiar in opening with a spondee : biilo 'py 
ayam sthavira iva ’vabhasate, na ’yam balah sthaviro 'yam 
mato me, etc. Such a pada in such a stanza confirms the 
view that the whole rucira is merely a resolved jagatl. 

The alternate arrangement, referred to on the last page, may 
be seen in the tag at vii, 29, 51 : 

nihatya tam narapatim indravikramam 
sakhayam indrasya tad aindrir ahave 
tato parans tava jayakanksino naran 
babhanja vayur balavan druman iva 

1 At iii, 25, 5, a rucira stanza appears among the group of tristubhs with 
which the chapter begins. Its first pada is an echo of the one cited above, 
tam agatam jvalitahutafanaprabhain. In vii, 2, 15-10, two rueiras appear 
in the same way among van^astlias. At the end of vii, 148, the tag-effect 
is done away with by the addition in C. 6,443 ff. of five flokas (not found 
in B.) after the two vangasthas, which in B. complete the tag begun by the 
rucira, 5G. 



The same arrangement has already (p. 303) been noticed in 
xii, 244, 29. Another stanza in this book, xii, 52, 34, forms 
the finale of a short system of pure (tag) vahgasthabilas. 
One fifth of all the rueira stanzas in the epic are in the Hari- 
vahga, inserted as tags, and they are all perfectly regular, 
with the substitution of one vahgastha pada each in 2, 123, 
35 c and 3, 34, 48 d, respectively. In addition to these, there 
is half a rueira at C. 10,274 (after vaiigasthas and before 
glokas), the prior pada of which ends in brevis : namo 'stu 
te mahisamahasurardini, namo 'stu te bhayakari vidvisam 
sadii. Here B. 2, 120, 43, inverts the padas, permitting the 
brevis at the end ; but it also lias a varied reading, bandhana- 
moksakarini, which leaves only one rueira pada. 1 The other 
cases call for no special remark. The caesura is after the 
fourth syllable. 

In the Ramayana there are but four ruciras common to the 
two texts, R. and G., two of which are in the seventh book ; 
but there are four in G. not found in R., and five in R. not 
found in G. As in the Mahabharata, the caesura is regularly 
after the diiambus, the gana division _ w w, 

w _ v , _ not corresponding to any text. Here the position 
of the rueira is always that of a tag, usually after upajatis. 
The second hemistich occasionally ends in brevis, e. g., G. ii, 
68, 56 ; vii, 68, 25 ; R. vi, 62, 22 ; but, as in the Mahabharata, 
even this liberty is seldom taken. In R. v, 7, 15-17 (not in 
G.), of the twelve padas, all are regular save the first, which 
has an extra syllable : itl Va tad grham abhigamya gobhama- 
nam. 2 In G. vi, 39, 33, pada b has yagaskaram priyakaram 
banclhavapriyam, where R. 62, 22, is regular, yatha priyam 
priyarana bandhavapriya. R. omits the tag of G. vii, 68, 25 
(continuing with glokas). The case is interesting, because it 
is evidently an instance of breaking a chapter by means of 
a tag (perhaps as an aid in recitation), and because the rueira 

1 P.W. s. v. mahisa 2 c, gives a var. lec. I give the readings of the Calcutta 
and Bombay, 1895, Harivanja. 

2 It is easy to suggest fobhitam ; but this half-rucira half-praharsini pada 
really needs no emendation. See just below. 



tag thus employed is highly irregular (pada b: hanisyasi, 
Raghuvara, na ’tra samgayah) in making the crnsura answer 
for a long syllable; thus showing that there is a late (care- 
less) freedom as well as the freedom of early (undeveloped) 
forms. With one exception, no such substituted padas as 
equate upajatis hi the other epic occur in the Ramayana. 1 


Having thirteen syllables to the pada this metre is called 
an atijagatl, though its finale is that of a tristubh, w _ ii. As 
to the relation of the measure, it is clearly of the puspitagra 

class, in closing in ^ _ w _ w , as will be seen below ; and 

as clearly of rucira nature, both in its middle and even in its 

opening; for it begins with a mora-equivalent, , of the 

rucira’s diiambus, w _ w and continues with the rucira’s 
resolved tristubh form. In fact, as we have seen that a 
rucira pada may appear with the extra syllable of the pra- 
harsini, we can supply all the links from tristubh to puspi- 
tagra with actually extant measures (see also below, under 
matrachandas, p. 837) : 

tristubh-jagatl, w — \j , w w \j \j 

rucira type, w W ♦ W \J KJ W \J 

rucira freaky w — w , w w w w ^ 

praharsini, w w w w __ w w 

puspitagra, w w 'W' \^j ^ 

The secondaiy caesura sometimes makes the pada coincide 
very closely with the rucira, for example in R. ii, 79, 17 a-b: 

uctis te | vacanam idam | nieamya hrstah 
samatyah | saparisado ] viyataqokah, 

but in other cases this caesura causes a trochaic cadence to be 
struck with the beginning of a new word after the proceleus- 
maticus, as is clearly shown in R. ii, 107, 17 c-d: 

gacclia tvam | puravaram [ adya samprahrstah 
saiiihrstas j tv aham api | Dandakan praveksye 

1 For this exception in the Ramayana, see above, p. 309. 



Do thou now | to the city j fare with heart rejoicing 
while meantime j merrily I | will to Dandakas go 

It is rather striking that in both these examples the name 
of the metre seems to be implied in it, hrstah and samhrstas 
like praharsinl (or -anl) “ rejoicing,” but I do not know that 
this is more than an accident. 1 There is a parallel in the 
rucira-like pada cited above from the Ramayana, G. vii, 88, 22 : 

gato ravir suruciram astagailam 

The Ramayana has one more ease of this metre, G. vi, 
25, 41, sa krodhad vipulayaga mahiinubkavo, etc. The only 
short finals are at the end of the whole stanza. 

In the Mahabharata there are twelve cases, all regular (but 
the first hemistich as well as the second may end in brevis), 

with the same norm and varying caesura, _1 , w w w w, 

w w , or , w w w, w w w They are not 

all tags. For example, that at vii, 143, 48 and the group of 
four in xii, 322, 24-27 ; but that at i, 2, 396, is the tag of a 
tag, apparently merely a scholiastic addition : 

akhyanam tad idam anuttamam mahartliam, etc., 

as are those in i, 21, 18; 22, 12; 25, 17 (tag to a rucira tag, 
b ends in brevis), veclangany abhigamayanti sarvayatnaih, etc. 
The first two of these just mentioned are akin : in 21 c, vistlr- 
nam dadrgatur ambaraprakagam ; in 22 b, gambhlraih vikasitam 
ambaraprakagam. In the specimen at xiii, 7, 28, the praharsinl 
is by one gloka stanza 2 removed from the end of the section, 
and is a moral excrescence added to the tale : 

1 I may add that in the first example there is not only hrstah in I!, ii, 79, 
17, but in the rahfastha which precedes this tag we find : praharsajas tam 
prati baspabindavo, etc. See a case like this from the other epic cited in 
the next note. 

2 This final ?loka says : “ I hare repeated what the seer proclaimed in 
regard to the getting of good and evil fruit. Now what do you want to 
hear ? ” The floka before the praharsivi is : Bhismasyai ’tad vaeah <;rutva 
vismitah Kurupungavah, asan prahrstamanasah pritimanto ’bhavans tada. 
See the last note. 



yan mantre bhavati vrtho ’payujyamane 
yat some bhavati vrtha ’bhisuyamane 
yae ca ’gnau bhavati vrtha ’bhihuyamane 
tat sarvam bhavati vrtha ’bhidhlyamane 

The tendency to restrict the final syllaba anceps to the close 
of the stanza is observable in several of these cases. For in- 
stance, in the group cited from the twelfth book, the only 
final breves are at the end of whole stanzas, not at the end of 
the first hemistich. In i, 2, the first hemistich ends in a short 
vowel, but before two consonants (d ends in a) ; in i, 21 and 
22, no final is short. The only exception is the one noticed 
above, i, 25, 17 b. The two cases in H., C., 6248 and B., 3, 7, 
25 are tags, and have no final brevis. The former has hiatus 
in pada d (avoided in B., 2, 53, 57, manujenclra ca ’tmanistham) : 

yad y’uktaiii, kuru manujendra, atmanestam 

The latter, instead of C.’s amrtam, 11,303, has 

yat satyain yad anrtam adimaksaram vai, 

where (adima and ksara are karana and ktirya) 1 adima is a 
late adjective. 

On the verse gopta samlksya sukrtinam dadati lokan, see 
below under matrachandas. 


Another atijagatl, not found in the Mahabharata, but in one 
text of the Ramayana, is the mrgendramukha of R. vi, 101, 
55, which takes the place of a puspitagra tag in G. 85, 13. 
The posterior padas of the latter metre have regularly the 
form illustrated by G. at this place, muditamanah samud- 
Iksiturii tvarami, This form is 

simply quadrupled in order to make the mrgendramukha ; 
the cadence of which is often made trochaic through the 
caesura, as in this epic example, a and d : 

1 In the next pada, B. has yad bhutam bhavati mithag ca yad bhavisyam, 
where C. has yad bhutam bhavad amitam ca. 



yadi vadham iechasi j Ravanasya samkhye 
yadi ea krtam hi tave ’cchasi pratijnam 
yadi tava rajasutabkilasa, arya, 
kuru ca vaco mama | cighram adya vira 1 

The native division of the pada of course is w w w, w _ w, 
w _ w, _ w which fits pada b. The brevis at the end of 
either pada, as in this case, is probably due to the fact that 
the metre is a stereotyped posterior pada in repeated form. 


The remaining aksara tags are longer metres, the gakvarf, 
of syllables 4 X 14; the atigakvari, of 4 X 15: and the ati- 
dhrti, of 4 X 19. Of the first, there are two varieties, of which 
one is the asambadha, found only in the Bombay R. ii, 116, 

25, with the norm (according to the example, , _ 

w w w, w w w, ) violated as follows (prior hemistich) : 

Ramah saiiisadhya rsiganam anugamanad 
deqat tasmat kulapatim abhivadya rsim 
samyak pritais tair anumata upadistarthah 
punyam vasaya svanilayam upasampede 

To this is added a supplementary tag, a peculiar stanza 
(where G. iii, 1, 35, has a vahgastha tag), in which the last 
pada differs from the three preceding ; a, b, c, being alike in 
having each the fourteen mono of the even vaitaliya pada 
(explained below), and eleven syllables, but not in a fixed 
order; against seventeen morse and twelve syllables in d. 
Pada b is aparavaktra, but I do not know what to call the 
whole (R. ii, 116, 26) : 

acramam rsivirahitam prabhuh 
ksanam api na jahau sa Raghavah 
Raghavam hi satatam anugatas 
tapasaq ca ’rsacarite dlirtagunah 

1 The stress, but not the quantity, is Saturnian : kuru ca vaco mama | vfrum 
mfhi Casmena | cTgiiram adya vira | insece versutum. The name mrgendra- 
mukha comes from the mnemonic verse : ksudhitamrgendramukham mrga 
upetya (Brown). 




The second cakvaii called vasantatilaka (or °kam) is found 
twice in the Rfimayana, but only in the last and latest book, 
vii, 8, 28, where it is followed by a jagati upajati as a final tag, 
and vii, 96, 23, also a tag. In the latter case, all padas have 
heavy final syllables. In the former, pada c has final brevis, 
but this liberty is taken in the case of the vasantatilaka even 
by the classical writers. 1 The metre is clearly hypermetric 

tristubh : v_/ , w w w , — w \j — w _ — or w — \ j , w w — , 

w w w In the first example, three padas have caesura 

after the fifth, like other hypennetric tristubhs : 

esa maya tava naradhipa raksasaiiam 
utpattir adya | kathita sakala, yathavat 
hhuyo nibodha | Raghusattama, Ravanasyd 
janma prabhavam | atulam sasutasya sarvam 

The Mahabharata lias twelve occurrences of vasantatilakiis, 
but only eight separate stanzas, the others being vain repeti- 
tions of old material. The first three are in the tag-group at 
the end of i, 2, 391 ff., which ends in a praharsinl. The second 
of this group has short finals in b and d; the third (winch 
follows immediately after two ^-lokas) has final breeds in a. 
The stanzas are benedictive and are partially repeated at the 
end of xviii, 5, 67-68, where B. lias the third of this group 
(omitted here in C.), and this again is found at the beginning 
of the Harivan^a. In all these occurrences of the same stanza, 
dadati is left at the end of pada a ; but in c the reading varies 
between satatarii cmotT in xviii and crnuyac ca nityam or 
tadvat in i, 2, 395 and Harivanca, i, 1, 4. In xiii, 151, 80, 
the same stanza has kathayec ca nityam. I give it in full on 
account of its universal interest: 

1 Compare the note to Vamana’s Stilregeln by Professor Cappeller, p. 23. 
The final brevis in prior padas is found also in inseriptional poetry. Compare 
e. g., the third and tenth stanzas in Vatsabhatti’s poem, fifth century, given 
in Biihler’s essay on Indian inscriptions, p. 01, where padas a and c respec- 
tively close in brevis; or the fifth and twenty-fifth, where, in each, both the 
prior padas end in brevis. In fact, the tendency here is to close the hemistich 
in heavy syllables and the prior padas in light syllables (25, 27, 31, 32, 40). 



yo gocataiii kanakagrngamayam dadati 
vipraya vedaviduse subahugrutaya 1 
punyam ca bh&ratakatham satatam crnoti 
tulyam phalam bhavati tasya ca tasya cai ’va 

In the thirteenth book (as in the case of the Ramayana, 
this metre is found only in pseudo-epic or late books), there 
are two new cases of vasantatilaka. The first, 14, 189, is 
unique in not being a tag (only d has final brevis) ; the 
other (with a gardulavikrldita) being a tag, as usual. The 
latter is united with the beneclietive stanza above, and like 
it has final brevis in the first pada, 151, 80-81 (80 being the 
stanza quoted above). 

The Harivanga has a tag-group (followed by one cloka) of 
three more vasantatilaka stanzas at 3, 114, 39-41, the last of 
which also has final brevis in c : 

41, c, jyotis trilokajanakam tridacaikavandyam 

d, aksnor mama ’stu satatam hrdaye 'cyutakhyam 


This is an atigakvarl, 4 X 15 syllables, having syllaba anceps 
regularly only at the close of the hemistich, but in one in- 
stance at the end of a prior pada, a freedom found among 
classical works only in the Mrcchakatikam, according to 
Professor Cappeller. 2 The metre is found in both epics; 
but the Ramayana has only one case common to R. and G., 
and that is in the last book, vii, 59, 23 = G. 61, 21, the 
stanza only ending in brevis. It is a tag. In R. vi, 40, 
29-30, there are two cases, not in G., both regular, a tag 
couplet (in the former case both hemistichs end in brevis). 
G. ii, 106, 29-30, has two stanzas, not in R., a tag (final 
brevis only at the end of the first stanza). The natural 

division is often w ^ w , www, , _ w w , with 

Ccesura after the spondee. The Mahabharata has eleven cases, 

1 v. 1. bahim^rutaya in the Bombay H, Also ca for su-, and other vari- 
ants in Anugasana. 

2 Loc. cit. 



and (like the one case in both texts of the Ramayana) they 
are all in the later epic: vii, 73, 48; viii, 85, 1-4; 90, 24; 
xiii, 6, 45-47; H. 2, 105, 84; and 3, 132, 100. The one in 
Drona unites with a puspitagra, but, although both are alm ost 
at the end of a chapter, they are rather a tag to a speech than 
to the chapter itself. Those in Kama are at the beginning 
and in the middle of their respective chapters. Those in 
Anu§asana are a tag, except that two glokas follow. In the 
group of viii, 85, all the padas end long except the posterior 
padas of the third stanza, both of which have final brevis. 
The two cases in Hariv. are tags (one gloka following in the 
latter) with brevis only at the close of the stanza. An irreg- 
ularity appears in xiii, 6, 46 c-d : 

bahutarasusamrddhya manusanam grlianl 
pitrvanabhavanabhaiii dnjyate ea ’maranam 

In 47, the hemistichs end in brevis ; in 45, only the first 
hemistich. The plural grhani is remarked upon as Vedic by 
the scholiast, who thus supports it; but grharn (vai?) is 
probably right. 

A very common cadence, whereby the end of the pada 

assumes the fall _ w , w , rather than _ w w , 

is illustrated by H. 3, 132, 100 a-b (cited above) : 
ajaram amaram ekam dhyeyam adyantacunyaiii 
sagunam agunam adyarii sthulam atyantasuksmam 

Another kind of malinl, not found in the epics, begins with 

W , WWW ) showing that the epic form is a further 

resolution of an original tristubh, which may be represented by 

1 _ w w This is, of course, the vaigva- 

devl form of the hypermetric tristubh, 1 the close relation of 
which with the puspitagra is well shown in vii, 73, 48-49: 

48 a-b : asurasuramanusyah paksino vo ’rago va 

pitrrajanicara va brahmadevarsayo va 

49 a-b : yadi viqati rasatalam tad agryaiii 

viyad api devapuram Diteh purarn va 

1 Compare Professor Jacobi’s learned essay, Entwickelung der indischen 
Metrik in nachvedischer Zeit, ZDMG. vol. xxxviii, p. 609. 



The content of this malinl appears a little further on, 7T, 
26, in the form of a puspitagra : 

yadi ca manujapannagah pigaca 
rajanicarah patagah surasurac ca 

and in viii, 37, 36, in aparavaktra : asurasuramahoragan naran. 


The only remaining aksaraechandas in the epic is the ati- 
dhrti (4 x 19) gardulavikrldita, which occurs in the eighth 
and thirteenth books of the Mahabharata. The chapter of 
the former book graced with a malinl is also enlivened with 
the “ tiger’s play,” viii, 90, 42 (two lines in C., 4668-9). It 

is not a tag and is perfectly regular, four times , _ w w 

w_w, vy, w There are also one and a 

half stanzas at xiii, 14, 229, and a whole stanza ib. 234; 
neither of which is a tag. This position of a fancy-metre 
in a chapter instead of at its end always shows a late section 
(affected in the Harivaiiga). In xiii, 151, 79, the gardulavi- 
krldita joins with vasantatilakas to make a tag. All the speci- 
mens are regular. The metre may be a late development 
from the tristubh. The intermediate phases, however, are 
not very clear, though the genesis may tentatively be as- 
sumed as; , _ ^ w _ w _ w (as in the viiitallya, below), 

^ w , __ w or two stanzas to the strophe, as in 

the classical grouping of glokas, with shift of caesura. This 
metre is not found in the Ramayana. 

Ardhasamavrtta (Matrachandas). 

(A) Puspitagra aistd Aparav ak tra. 

These metres, as is indicated by their name “ semi-equal,” 
are uneven in their padas. They are not quite mora-metres, 
since the number and position of their syllables, heavy or 
light, are regularly fixed; but on the other hand they are 
not like aksara metres, for their padas are not identical. In 
the epic, however, the rule of fixed syllables is not strictly 
preserved. The cadence of the hemistich, with its unequal 



padas, has either wholly trochaic close or alternate trochaic 
and iambic. The first is illustrated by R. vi, 33, 36 c-b : 

tam iha qaranam 

abhyupaihi devl 
haya iva mandalam 

aqu yah karoti 

Rapid as a charger is, 

Hasten, hurry quickly. ' 

As already remarked, the second piida of this puspitagra, 
when quadrupled, makes the mrgendramukha (above, p. 331), 
which also has trochaic fall. The aparavaktra, which has one 
syllable (usually two mone) less than the puspitagra, shows 
more clearly the derivation from the tristubh, R. ii, 39, 41 : 

Dacaratha-veipna babhuva yat pura 

or, again, in M. viii, 37, 42 : 

bhavatu bhavatu, kim vikatthase, 
nanu mama tasya hi yuddham udyatam 

There is one form of tristubh which actually corresponds 
to the second verse of the puspitagra, when its breves are 
equated with heavy syllables, thus: 

tristubh I manai11 na kuryan na ’dadhlta rosam 

• • [ — . — W 

. , _ _ i r \j \_/ — w — 

puspi agra j svaparamatair gahanam pratarkayadbhih 

Professor Jacobi also sees in the jagati or tristubh the ori- 
gin of the puspitagra, though he is inclined to adopt a more 
complicated development (from a Vedic verse of 12 + 8 
syllables). 1 

The puspitagra and aparavaktra are used only as tag- 
metres; sometimes, as in R. v, 16, 30 (not in G.) inserted 

1 ZDMG. yoI. xxxviii, p. 591 ft. Professor Jacobi, p. 595, regards the puspi- 
tagra as a development from a pure matrachandas, which in turn he refers 
to the satobrhati (4 X 12 + 8). Compare also the same author, IS. vol. xvii, 
p. 449. 




among upajati tags common to both texts ; sometimes, as in 
G. iii, 54, 28 (not in R.) after a common tristubh-tag ; or in 
other similar situations. 1 

The puspitagra occurs much more frequently as a tag- 
metre than does the aparavaktra. For example, in the Ila- 
mayana, the puspitagra is found four times as often. There 
are, however, only thirteen eases common to the two texts, 
R. and G. Besides these, G. has fourteen, and R. has twenty- 
one cases not found in the alternate text. 

The mark of the posterior pada, as distinguished from the 
prior, is the apparent insertion of a heavy syllable (in terms 
of matra metre, two mora;), at a point which is usually fixed 
as after the initial four breves. This, however, is not always 
the case. Thus in G. v, 31, 62 b, corresponding to d, which 
latter, vacanam idam mama Maithili pratilii, is regular, ap- 
pears as posterior pada of a puspitagra: 

lavanajalanidkir gospadlkrto me, 

where the heavy syllable is put after all the breves, perhaps 
merely on account of the awkward phrase (in (jloka, ib. 33, 
23, gospadlkrtah). Later rule especially forbids this arrange- 
ment for all matrachandases : “ In the opening of prior padas, 

w w, and of posterior padas, w w and _ v w w w w ^ 

and w w w www-_, are forbidden." 2 

Further, for the prior pada may be substituted a different 
cadence, almost that of the vaitallya, ww_ ww_ , ww_, 
w This occurs in G. vi, 62, 44 a (where R. 83, 44, has 

the normal w w, w www, w w thus : 

G., ayam adya vibho tava ca priyartham 
R., ayam anagha tavo ’ditali priyartham 

Compare G. vi, 92, 83 b: svabala 'bhivrto rane vyarajata, 

1 In G. vi, 39, 32, where R. has only a rucira, there is a puspitagra inserted 
before the rucira. These two names, by the way, appear together as ordinary 
adjectives “blooming and shining” (trees), supuspitagran ruciran (vrksan), 
R. v, 14, 41. 

2 Weber, IS. vol. viii, p. 309. 



where R. 108, 34, has svajanabala ’bhivrto rane babhuvS. 1 * 

The prior pacla may be hypermetric. Thus R. vi, 107, 
68 a-b : 

Dacarathasutaraksasendrayos tayor 
jayam anaveksya rane sa Raghavasya 

A parallel ease or two occurs in the other epic (see below). 

Occasionally there is a quasi inversion, w_ w _ w _, of the 

ending _ w _ v_/ This occurs twice in R., but only in 

Adi and Uttara. The first ease presents varied readings. In 
G. i, 22, 20, there is simply the not unusual equivalence of a 
and c puspitagra and b and d (aparavaktra) catalectic. But in 
R. the same stanza, i, 19, 22, lias, besides, the irregular pada a : 

WWW WWW W W W WWW w w w w w 

WWW WWW W W | = b 

that is, instead of iti hrdayavidaranam tadanlm in G. a, R. has 
iti sahrdayamanovidaranam. Tliis can scarcely be a mere 
lapsus, as the finale occurs again in the Mahabharata and in 
R. vii, 29, 38 c-d : 

yad ayam atulabalas tvaya ’dya vai 
tridacjapatis tridaqacj ca nirjitah 

In the latter passage, 37 a has w w as close : 

atha saranavigatam uttamaujah 3 

While posterior pudas have syllaba anceps, as in G. vi, 92, 
83 b, cited above, a prior pada has this only in R. vi, 33, 36, 

1 Another case of variation, R. vi, 84, 22 d = G. 63, 22, where G. has asura- 

varo ’nmathanaya yatha mahendrah may be corrupt (for asuravaro ’nma- 
thane yatha mahendrah ? ). B. has divijaripumathane yatha mahendrah (for 
ripor 1). 

3 In b, compare G. v, 36, 77 b, Janakanrpatmajadhrtam ; but R. 38, 70, has 
Janakanrpatmajayadhrtam prabhavat, which is correct. In R. vii, 29, 37 and 
38 are puspitagras ; 39 and 40 are aparavaktras. In G. the only irregularity 
here is in (37) 38 c, svasutasya vacanam atipriyam tat. Here in 40 = R. 39, 
a is aparavaktra and b is puspitagra, though the latter may have added the 
unnecessary tvam that makes the change. The same is true of R. 38 a. 

I have noticed besides only the following puspitagra irregularities, which 
seem to me more grammatical than metrical, or mere errors : G. ii, 29, 

29 b, w w for w , read apratimarupa I G. iv, 34, 35 c, read 

anrtamadhura 0 ? Neither stanza is found in R. 



dev! (cited above), where, however, G. has Site (here, 9, 39, 
abhayamkaram is to be read). In posterior padas, final syllaba 
anceps is found about a dozen times in the forty-odd puspita- 
gras of the Ramayana text. 

The aparavaktra is a puspitagra shortened by one long 
syllable, two morse, in each pada ; or in other words, its pada is 
a catalectic puspitagra, pada. To native prosodians, as to Euro- 
pean scholars, the shorter is the type, and the puspitagra is an 
expanded aparavaktra ; a view that appears to me erroneous. 
The aparavaktra occurs in the Ramayana, as said above, not 
quite one-fourth so often as the puspitagra. 1 Like the latter, 
it is used alone, or with other metres to make tags. The final 
syllables are always long. Irregularities are rare ; a substitute 
like that in the puspitagra occurs in G. ii, 82, 15 a : 

\J KJ \J KJ \J \J W W 

\-J \J \J \J, \J W , \J W 

wwww, (ca satl omit ca ?) 

V W V, W W , W V 

Here R. has a regular aparavaktra, ii, 81, 16. In G. iv, 62, 25, 
the second pada is plavagapungavah paripurnamanasah, for 
R.’s (63, 15) plavagavarah pratilabdhapaurusah ; and in G. 
63, 29, plavangamah paripurnamanasah. 

There is only one passage in the Uttara, vii, 29, 37^40, 
where puspitagra and aparavaktra are found. Otherwise these 
metres are distributed pretty evenly over the Ramayana, 
except that the first book has no aparavaktra, 2 and only one 
puspitagra common to both texts, but R. here has four not in G. 
The reason is that the later epic prefers pure matrachandas. 

Interchange of aparavaktra and puspitagra padas occurs 
occasionally, as in G. ii, 15, 36 (R. has upendra here), where a 

1 There are only six cases common to both texts ; besides, two in R. not 
in G. ; three in G. not in R. ; twelve in all, as G. at iv, 62, 25 and 63, 29 has 
the one at R. 63, 15. In the last case, the first pada is the same in the three 
stanzas ; in R. all the other padas are normal, but in G. 62, 25 d is a puspi- 
tagra pada, as is c of 63, 29. The missing stanza in the alternate text is due 
merely to the latter having a puspitagra in G. iii, 7, 36; R. vi, 68, 24. 

2 The fifth book has no aparavaktra, but it has half a dozen puspitagras. 
The sixth book has the greatest number of puspitagras. 



and c are puspitagra padas and b and d are aparavaktra 
padas in regular interchange ; or as in G. v, 36, 77, where 
only the last pada of the stanza is catalectic (of aparavaktra 

It is clear that the puspitagra, a form of tristubh, and the 
aparavaktra, a catalectic puspitagra, are not regarded as separ- 
ate but as interchangeable in pada formation. As complete 
stanzas, the latter compared with the former, are rare. The 
pada type is not absolutely fixed. 

Before comparing the usage in the Mahabharata, I shall 
complete this description of the phenomena in the Ramayana 
with an account of the 


In the later part of the Ramayana — if one may dare sug- 
gest that' any epic poem in India was not all written at the 
same moment — the place of the puspitagra and aparavaktra, 
as tag-metres, is taken by pure matrachandases, namely, the 
aupacchandasika and vaitallya, which bear to each other the 
same relation as that held by the former pair; that is to say, 
the vaitallya pada is a catalectic aupacchandasika pada. 
These two pairs are essentially identical, as may be seen by 
comparing the posterior padas, which in each are increased 
by a long syllable. The posterior pada of the aupacchanda- 
sika is 

? W w , 

which, when catalectic, should have final syllaba anceps ; but 
this never happens at the end of the first hemistich, only at 
the end of the stanza, an indication that the vaitallya is the 
derived form. Again, the aupacchandasika is really the epic 
stanza metre. The vaitallya is used but once as a stanza, all 
the other cases being merely catalectic padas of an aupacehan- 
dasika stanza. The prior pada in aupacchandasika may 
also end in brevis, and, as the spondee is usually resolved 
into an anapaest in both padas, we get the norm (16 and 18 
morse) : 



(a) uu,_uu_,u_u_!i 

(b) W , \J w , w w Si 

or (b) , _ w w w _ w _ ^ 

This is evidently a variety of the puspitagra. 1 That is, it 
reverts to a tristubh origin. 

R. vii, 57, 21 — G. 59, 22, may be taken as the typical 

W, W , w ( ) 

w w , — w w , w w w 

W , ^ W , >-> \J 

O KJ , >JU , vy W 

G. adds te to R.’s pada a, iti sarvam aijesato maya (te). The 
final syllable of the stanza in vii, 61, 24 = G., 66, 24, vaitaliya, 
is short in R., long in G. Prior padas do not usually end 
in brevis, but they do occasionally, as in G. vii, 87, 18 (not 
in R.), where in b the spondaic type of opening is illustrated : 

iti karma sudarunam sa krtva 
Dando dandam avaptavan ugram 
qmu sarvam acesatas tad adya 
kathayisye tava rajasinhavrtta 

The close of b, however, shows an unusual phase of the type 
of the equivalent variant with spondee ; but it is not neces- 
sary to suppose that a brevis is lost before ugram. Both 
posterior padas may begin with a spondee (but end in 

— w — w ), as in R. vii, 55, 21 = G. 57, 22 (all padas end 

long), e. g., tulyavyadhigatau maliaprabliavau, apparently an 
older form than the usual resolved type. 

As in the case of the puspitagra and aparavaktra, the cata- 
lectic (vaitaliya) pada may take the place of the full measure. 
Thus in R. vii, 95, 17 (not in G.), the spondee type (b) is used 
as a catalectic pada : 

iti sampravicarya rajasihhah 
qvobhute qapathasya niijcayam 
visasarja munln nrpahg ca sarvan 
sa mahatma, mahato mahanubhavah 

* Compare the form cited above, ww w w w, as a variant 

of puspitagra (b). 



In R. vii, 86, 21 (G. 93, 21), a-b show a new form of this 
combination : 

iti Laksmanavakyam uttamaiii 
nrpatir atlvamanoharam mahatma 

that is, a vaitallya prior and puspitiigra posterior pada. Com- 
pare the only ease not in the Uttarakancla, where in G. ii, 
81, 33 (not in R.), a lame aparavaktra hemistich is followed 
by a lame matra hemistich (fifteen morae) : 


Vy KJ W, W , W __ 

\J \J KJ W W, 

\J V-/ , \J W , W W 

The patois metres show that the miitra-form was used early, 
but how much earlier than the third century b. c. it is impos- 
sible to say. The vaitallya itself is a common metre of the 
Dhammapada. 1 

Matrachandas in the Mahabharata. 

The many “ semi-equals ” in the great epic form a fair 
parallel to the state of tilings in the little epic. But there 
are no regular vaitallya or aupacchandasika stanzas at all. 
In a late passage of Vana and in (anti there is a sporadic 
approach to vaitallya form. On the other hand, there are 
over ninety-one puspitagras and aparavaktras. They are 
found chiefly in the later part of the epic and appear more in 
groups than they do in the Ramayana. The interchange of 
puspitagra and aparavaktra padas, of which I have spoken 
above, is met with in the very first example at the end 
of i, 30: 

dkrtamanasah pariraksane 'mrtasya 
asurapuravidaranah sura 

j valanasamiddhavapuhprakacinah 

1 The type here has in the posterior pada either anapaest, spondee or am- 
phimacer as an opening ; but both here and in the choriambs much greater 
freedom is allowed than in the epic, where, despite the occasional irregularities 
noticed above, the form is much more systematized than in Pali. 



iti samaravaram surah sthitas te 
parighasahasracataih samakulam 
vigalitam iva ca ’mbarantaram 
tapanamarlcivikaeitam babhase 

In the first stanza the padas are aparav., puspit., aparav., 
aparav. ; in the second, puspit., aparav., aparav., puspit. Al- 
most the same as the latter is the arrangement in a tag to a 
danakathana (followed by three tristubhs), at the end of iii, 
200, 126, where a puspit. pada is followed by an aparav. pada 
in the first couplet ; but the second begins with the posterior 
puspitagra pada, and is followed by the posterior pada of an 
aparavaktra : 

c-d : bhavati sahasragunarii dinasya rahor 
visuvati ca ’ksavam acnute phalam 

as if the posterior pada were used originally in either position 
as the norm ; which would agree with the identification with 
the tristubh ventured above. 

Of the eight puspitagriis in the seventh book, six (all tags) 
are perfectly regular (2 X 16 + 18) and require no notice 
(for C. 2731, rajam 0 , read rajani% as in B. 77, 26). Here 
only hemistichs end in brevis. Two cases deserve notice. In 
vii, 1622 = 37, 37 b, C. has pitrsuracarana-siddhasafighaih, in 
B., siddhayaksasanghaih. But B. is often less better than 
bettered, and here the net result of three corrections is to 
make a perfect puspitagra out of C.’s scheme, which is 

W W W W W W W W WWW w ™ W \J \J , 16 -f- 15 

W'-'W WWW W__W W .WWW WW W W , 17 + 17 

but this is attained by adding yaksa in b ; changing avanita^ 
lavigatai<j ca to avanitalagatai<j ca in c ; and inventing the 
word ativibabhau for abhibabhau in d (B, ativibabhau huta- 
bhug yatha ’jyasiktah). Mates to pada c were shown above 
from the Ramayana. Irregular too as is d, it is not lightly to 
be rejected, since it has its perfect parallel in the eighth book 
(below), as also in Hariv. C. 11,269 d (3, 6, 4 d) 

(iti sa nrpatir fltmavfnis tada ’sau) 

tad anu(vi)cintya babhtlva vitamanyuh 



where, for C.’s anucintya, ammcintya of B. may be a corrected 
reading, as above it is easy to propose abkivibabkau and refer 
to the Rig Veda for the form. 

The case at vii, 182, 27 = 8273, shows a better reading in 
B., where hi is required (accidentally omitted in C.). The 
padas here are regular, the stanza's end having brevis (in 77, 
26, the first hemistich ends in brevis). The chief peculiarity 
here is that the passage stands in the middle of the chapter, 
the other cases in Drona being tags. 

Once w w w — w takes the place of w w w w w w , producing 
in pikla a the choriambus equivalent to that in b and d. This 
happens in one of the two great groups of late aparavaktras 
in the eighth book, viii, 30, 3 (almost at the beginning of the 
chapter) 1 : 

WWW, W W , W W 

W W W W, W W , W W 

W W W W W W , W W 

W W W W, W W — , W W 

The rest of the twenty-five “ semi-equals ” in the eighth 
book are all grouped together in 37, 31 ff., where, after one 
puspitagra pada, follow, as in the last group after a stanza, 
aparavaktras only. In this group of twelve stanzas, breves 
occur but rarely at the end of the hemistich, in (31), 40, and 
42 at the stanza’s end, in 35 alone at the end of b. Only two 
of these stanzas require a word. In 37 c-d, where the first of 
the two padas has seventeen morse (for fourteen), 

dinakarasadrcaih qarottamair yudha 
Kurusu baliun vinihatya tan arln, 

it seems simple to drop the hypermetric and unnecessary 
yudha ; but it is in both texts (Xllakantha says that this par- 
ticular stanza is visamaiu chanclas) and has a parallel in 
Hariv. 11,269, where (C. only) a puspitagra begins : 

1 The first stanza of the chapter is a 9 loka ; the first stanza of the group is 
a puspitagra ; then follow aparavaktras to 0, where the first half is cataleetic 
(aparavaktra) and the second half is puspitagra (as in 13, b ends in brevis) ; 
10 is a regular aparavaktra; 12-14, regular aparavaktras; 11 is regular in 
B. a, but irregular in C. (finivrsabhayarapiditas for °(;arair nipiditam). Here 
d ends in brevis. 



vidhivihitam aqakyam anyatha hi kartum 1 

A similar case has been shown above in the Ramayana. 

The other stanza deserving notice is the first of the group, 
viii, 37, 31 = 1737. Both texts have a puspitagra pada in a ; 
an aparavaktra pada in b ; and in c-d 

jugupisava iha ’dya Pandavam 

kim bahuna | saha tair jayami tam 

that is, _ ^ w for w w _ of the resolution in vaitallya (but the 
csesura in d is after the choriambus : “ Though the gods may 
wish to guard the Pandu here to-day, what then ? I shall 
conquer him, gods and all 

In Canti, the puspitagras are generally too regular to be 
interesting. A big bunch of them in Moksa makes a tag at 
the end of adhy. 179, thirteen in all. They have an unusual 
number of final breves, but only because vratam idarn ajagaram 
9 uci§ carami is the final refrain of ten of them (only twice 
has b brevis). Of the twenty-one stanzas of this class in 
^anti (Moksa), sixteen are puspitagras ; five, aparavaktras. 
About the same proportion obtains in Harivanca, where there 
are twenty-two stanzas of ardhasamas, of which only three are 
aparavaktras. All those in (Snti are tags, either following 
tristubhs or followed by another supplementary tag (as in the 
case of a rathoddhata mentioned above). In xii, 250, 12 b = 
9035 (yad avidusam) mahadbhayam (paratra) in C. appears 
to be a lapsus ; in B. as sumahadbhayam, and in 10,530, yad 
avidusam sumahadbhayam bhavet ; but compare the parallel 
below in H. The following is a parallel to the case above 
in the Ramayana in its late form (w _ w _ w _) : xii, 319, 
112 = 11,836 (the order of morm is 17 + 18 + 16 or 17 + 16) ; 
where B. has : 

yad upauisadam upakarot tatha ’sau 
Janakanrpasya pura hi Yajfiavalkyah 

1 This is in the stanza referred to above. In this case, H. 3, 6, 4 a has only 
vidhivihitam acakyam anyatha, to which C. adds kartum. The fact that the 
Bame superfluity of syllables is found in the Ramayana must at least make 
doubtful an instant acceptance of the more usual form given in what is so 
often a clearly improved text. 



yad upaganitagagvatavyayam tac 
chubham amrtatvam agokam archati 

(here C. in c has °ganitam). Both texts have thus in a : 


and C. has in e : 

WWW W W , W W 

The last stanza in the book, 366, 9 = 13,943, has, as an 
aparavaktra tag, morse 14 + 18 + 14 + 18, alternate calalectic 
verses, of which I have spoken above. 

The remaining matrachandases in (Jiinti are discussed below. 
The thirteenth book has no aparavaktras but nine puspitagras, 
all of wliich are perfectly regular (the hemistich ends in brevis, 
e. g., 76, 31). All except those in the extraordinary (late) 
section, 14, 180, and 190, are tags, though 26, 101-2 are fol- 
lowed by four glokas. 1 

Apart from the padas already noticed, the Harivanga has 
little of interest. Interchange of the two forms (a, catalectic) 
occurs in 3, 6, 3. In the puspitagras at 12,705-6, the latter 
has in b, w ^ ^ ^ _ w, as in the lapsus above. 

Here sa has been dropped, (3, 42, 21) dititanayam (sa) 
mrgadhipo dadarga. As usual in the later books, several of 
the stanzas are not tags: 2, 123, 32 is followed by glokas and 
ruciras, but is near the end of the section; at the beginning 
are the three of 3, 6, 2 if. ; in the middle of the section are 
3, 49, 31 = 12,960, and 3, 50, 12 = 12,989; as are the four 
in 3, 51, vss. 18, 29, 42, 49 = 13,024-35-51-58. Many of the 
final stanzas are benedictive, as in 3, 6, 10, where puspitagras 
are interwoven in an upajati kavyastuti : 

vijayati vasudharh ca rajavrttir 

dhanam atulaiii labhate dvisajjayam ca 
vipulam api dhanam labhec ca vaigyah 
sugatim iyac chravanac ca gudrajatih 
puranam etac caritam mabatmanam 

adhltya buddhim labhate ca naistiklm, etc. 

1 Here C., 1800 b, has the meaningless words : filataraye tripathaganuyo- 
garupan, for °rataye . . . pathanuyoga 0 in B. 



It will be convenient here to put together the forms of 
ardhasamavrttas thus far exhibited in the two epics. In the 
Mahabharata and Rainayana the general types of aparavaktra 
and puspitagra are : 

(a 1 ) w w w w w w , w w (_), 14 (16) morse 

(b 1 ) w w w w w w, — w w ( ), 16 (18) morse 

These may be called the types, because the following vari- 
ations are proportionally insignificant. But, though few in 
number, they are important as showing that there was no 
absolute line between the fixed matrachandas and the free 
matrachandas, for these variations may just as well be re- 
garded as, e. g., vaitaliya padas as variants of aparavaktra 
padas. But it must be remembered that they do not repre- 
sent padas of, e. g., vaitaliya stanzas ; only equivalent padas 
of, e. g., aparavaktra stanzas, which I call variants on account 
of their position : 

In M. and R. both are found the following variants of (a 1 ) : 

(a 2 ) WWW WWW w w w ( ) 

In both texts of both epics, two cases in M. ; three in R. In 
M. both cases are in piida c ; in R., only in aparavaktra. 

(a 8 ) - w w www _ w _ w _ w _ (hypermeter) 

In M., in both texts and also in Harivanga ; in R., one case. 
In M. alone : 

(a 4 ) WWW ww w w 

In R. alone : 

(a 5 ) www www, ww w (B., vii) 

(a 6 ) ^ ^ - ww_ ww_ w (G., 17 morse) 

(a 7 ) www www — w — w w — (doubtful, pada c, 15 morse) 

(a 8 ) www ww — w — w (only in G., pada c, 15 morse) 

In M. and R. both is found the following variant of (b l ) : 

(b 2 ) wwww_ w w — w (only in C. and G., 15 morse) 

In M alone : 

(b 8 ) www — w w — w — w (only in C. and Harivanga, 

padas b and d, 17 morse) 

(b 4 ) w w w w — , w — w — w __ w (sic, bis in C.) 

(b 5 ) WW ww w w 



In R alone : 

(b 6 ) w w w www_, _w_w (only in G., forbidden by 


(b T ) w w w w , w w — w — w — (only in G.) 

(b s ) w w ww, — w — w — (only in G.) 

(b 9 ) w w w w — w — w — ww (only in G., a prior aupa- 


The complete vaitaliya and aupacchandasika stanzas, of 
perfect mora form, found only in the later Ramayana, have 
the scheme : 

(a) W W, W W , W W (m) 

(b , W W , W W (id) 

(b 2 ) W W , W W , W W (id) 

(b 3 ) , WW W 

Before taking up the odd eases remaining, I cannot refrain 
from departing somewhat from a purely metrical point of 
view, to express admiration for the art with which these 
metres are handled. The poets of the later epic play with 
them skilfully. They are not apprentices hut master work- 
men. I give two illustrations. In one, the metre is em- 
ployed to give a list of fighters and weapons, the names of 
which are cleverly moulded together to form half a perfect 
stanza. In the other the poet is indulging in satire at the 
expense of the philosophers: 

viii, 30, 5, parighamusalaqaktitomarair 

nakharabhuQundigadaqatair hatah 
dviradanarahayah sahasraco 

rudhiranadipravahas tada ’bliavan 

xii, 179, 35, bahukathitam idam hi buddhimadbhih 

kavibhir abhiprathayadbhir atmaklrtim 
idam idam iti tatra tatra tat tat 1 

svaparamatair gahanam pratarkayadbhih 

I have now given seriatim all the matrachandas cases in 
the great epic, with the exception of one case in Yana, to be 

1 t. 1 . lianta. 



mentioned immediately, and two or three peculiar groups in 
Canti, also to be discussed below. It will have been noticed 
that in the later books great heaps of stanzas of this metre 
are piled together. Thus all the twenty-five in Karna (a 
late book in its present shape) are in two sections, thirteen 
stanzas in one, twelve in another ; while in r.'anti another group 
of thirteen is found. This stupid massing of adornments — 
for these tag-metres were used originally only as fringe-work 
— the still later tliirteenth book exceeds by uniting together 
in one heap, first, a puspitagra, xiii, 14, 180, then four aryas, 
ib. 181—4, then two qlokas, ib. 185-6, then an arya, ib. 187, 
then an upajati, ib. 188, then a vasantatilaka, ib. 189, then 
a puspitagra, ib. 190, then an arya, ib. 191. 

Despite this profusion of puspitagras and aparavaktras, 
the Mahabharata has no such regular vaitaliyas and aupae- 
chandasikas as has the later Ramayana. But the following 
interesting verses occur in the popular story of Yudhistlura 
and the daemon, who required him to answer certain ques- 
tions. They are not tags, iii, 313, 112-113; they are late; 
and they are an approach to vaitaliyas: 

priyavaeanavadi kim labhate 
vimrqitakaryakarah kim labhate 
bahumitrakarah kim labhate 

dharme ratah kim labhate kathaya 

w w, \j kj \j — \j , 15 

WWW w, w w , w w , 16 

ww, — ww , , 14 

— , — w \y w, 16 

priyavaeanavadi priyo bhavati 
vim rej i tak&ryak aro 'dhikam jayati 
bahumitrakarah sukham vasate 

yaq ca dharmaratah sa gatim labhate 

W, W W , W Ul/ V| 15 

W w w \J, — uu , uuy, 16 

w, — w — ,\j — w — , 15 

W, W KJ , UU , 17 



In C., 17,397-98, the same text. This is the kind of stoiy 
which, because it appears Buddhistic, is often labelled as a 
matter of course ‘certainly old.’ But the tale, on general 
principles, is just as likely to be late as early ; perhaps more 
so, when one considers that kings interviewed by spirits who 
ask conundrums are merely stalking-horses, and must first be 
famous as kings before such stories are fastened upon them. 
This particular tale bears all the marks of a late inset. 1 

Although the great epic lacks the regular vaitaliya of the 
Bamayana’s Uttarakanda, yet Canti offers a type of metres 
which shows forms ending in the close of this measure. 
For besides the usual ending of the matra form, 

the close may also be _ w w (called apatiilika). Also 

the beginning of the verses given below is of matra-formation, 
but the matras are not regular. The group xii, 322, 28-32 = 
12,071-75, follows a group of praharsinls (4 X 13 syllables): 

28, raja sada dharmaparah ^ubhaQubhasya 

gopta samlksya sukrtinaiii dadhati lokan 
bahuvidham api carati pravicati 
sukham anupagataiii niravadyam 

W , > W W W 

W W W 



Morse 20 + 21 + 14 + 14, the first hemistich bridging the 

preceding praharsinls, , _ w _ w , and the 

apatalika (c-d scheme also in 30, below). 

29, cvano blilsanakaya ayomukhani 

vayansi balagrdhra[kula] paksinaiii ca safighah 
narakadane rudhirapa guruvaca — 
nanudam uparatam vicanty asantah 

, W W , W W w 

W W, W W W [w w] , W W 

W W W W WWW , w w w w 


19 + 19 + lo 4 - 16 

1 Compare Holtzmann, who rightly says that the story is a late addition 
to the third book to connect it with the fourth, Neunzehn Bucher, p. 95. 



30, maryadaniyata svayambhuva ya jhe 'mah 
prabhinatti daqaguna manonugatvat 
nivasati bbrqarn asukham pitrvisa — 
ya-vipiuam avagahya sa papah 

, , \J W, \J * 1_ 

\j \j — kj vy 

\j \J 'U w w w, w w { 

\j \j \j (z 

22 + 18 + 14 + 14 

31, yo lubdhah subhrcam priyanrtac ca manusyah 
satatauikrtivaucana 1 -bbiratih syat 
upauidhibhir asukhakrt sa paramanirayago 
bhycam asukham anubhavati duskrtakarma 

a, , — uu — , w — \j — w\_/ (= 32 a) 

b, •u \j \j \j \j \j — \j> — ww 

C, WWV/VJ, wwww — , 

d, w \j \j \j w \j 

22 + 17 + 19 + 18. Here c lias the resolved equivalent of 
the v v _ close of a, b, d. The choriamb of a is all 

resolved in d, ^ ^ w w ^ w ; in c only the first 

syllable, ^ ^ _ (as if sa were interpolated). 

32, usnaih Yaitaranltn mahanadlm 2 avagadho 
paraquvanaqayo nipatito vasati (ca) 
ca mahaniraye bhrcartah 

, — v/ — > (v/ — ) vy — w (= 31 a) 


V W v (\^) 

V Ui W W , W 

22 (19) + 13 + 16 + 13 

= 28 c) 
= 28 d) 

1 C. vacana, but N. vaiicana cauryadi. 

2 C. omits maha°. 




In xii, 336, 11-12 = 12706-7 occur two lines, as printed in 
C., which seem to be rather rhythmical prose than poetry ; 
but in 347, 18-22 = 13444 there are five matra stanzas, of 
which I give the scheme alone (they are not arranged in the 
same way in both texts) : 


W W W _ (16 X 2) 

\J \J KJ \J KJ , \J KJ KJ KJ , KJ \J KJ \J , 

WWW wwww (16 + 17) 

B. adds w w w w w w which C. gives to the next stanza. 

19, WWW WWW , WW W — w, ww ww , w w (16 + 14) 

C. adds w w _ w w which B. gives to the next hemistich. 

w W WW , WW WW , WW WW , WWW w (16 X 2) 

20, WW WW ,WW WW , W — w,ww ww (16 X 2) 

WW W W,WW — ww w w w ,www w (16 + 17) 

21, w ww, w ww , (w ), w w w w , w w w 

(16 4- 22 or 17) 

WW W WWW W , WW w w, ww ww (15 + 16) 

Perhaps puranam in 21 is to be omitted. The text is : 

taiii lokasaksinam ajam purusam puranam ravivar- 
nam Icvararii gatim bahuqah 
pranamadhvam ekamanaso yatah salilodbhavo 'pi 
tam rsim pranatah 

22, WW W W, WW WW , w ww, ww w w (16 + 17) 

w W, W W W W , W ww, w w w (16 + 18) 

The arya form is clear in stanzas 18 and 20. On the other 
hand, the first stanza is an almost pure praharanakalita pada, 
w w wwww—, while the pramitaksara pada, w w — w — w 

U V \J \J > prevails in the following stanzas ; not, however, 

as pure gakvarl or jagatl stanzas, but with matra resolution. 
The stanzas, if they are treated as one group, may perhaps be 
considered as rather rough matrasamakas (four padas of six- 




teen morse each), partly of the vitjloka type ; 1 or as aryagiti 
(but with four morse in the sixth foot), mixed with matra- 
samakas. Nothing of this sort is found in the Ramayana. 


The statement that the arya metre occurs in Buddhistic 
writings (and earliest inscriptions) but not in epic poetry, was 
made so long ago that the learned author of Das Ramayana 
can scarcely at this date be held responsible for the slight over- 
sight. 2 Nor is the main argument, to which this statement 
served as a support, especially affected by the fact that the 
Mahabharata, besides the stanzas of aryaglti mentioned in the 
last paragraph, has eight arya stanzas ; since these are in parts 
of the epic so late that their presence, as affecting epic poetry 
in general, may be discounted ; at least for any one who takes 
a reasonably historical view of the growth of the great epic. 
Six occur in xiii, 14, 181-84, 187, 191 = 772-75, 778, 782 : 

181, _ » W W, ; W W. W W W W, W D, V W W, 

, — . w w, ; w w, w w , w, , w 

182, , w w — , w w ; w w, , w u, , w, 

w w w w, 3 w — w, w w ; — w w, , w, , 

183, w w — , , ; w — w, , w w, , , 

, WW, WWWW, WW, WW , W,WW , 

184, w w w w , w w w w , w w ; w w , w w » w w , 

J w W , WWW W : W W t W W J w, — 

187, W w , W W , w w ; w W , w w , w , , , 

W W W W , W W , W W W W • , W W , W I - 

191, ww — , w — w, ; wwww, , w. , , 

The last two stanzas are upaglti, that is, they have the 

1 Colebrooke, Essays, vol. ii, pp. 78, 142 ff. ; Weber, Ind. Stud., vol. Tiii, 
pp. 314-318. I am indebted to a query note in Professor Cappeller’s manu- 
script for the suggestion that these may be imperfect aksaracchandases of 
the types named. The pure matrasamaka has brevis in the pada’s ninth 

2 ZDMG., vol. xxxviii, p. 600 ; Das Ramayana, p. 93. 

* B., bhavati hi ; C. omits hi. 

4 Text : yesam na ksanam api rucito haraearanasmaranavicchedah ; ayagiti 
and neglected caesura; but if api (an easy intrusion) were extruded, the 

neglected caesura would be in its usual place, , w w, w w ; wwww, 

w w, w, , with the arya final foot of two morse. 



short verse in each hemistich. The full eighth foot, aryagiti, 
is found only in 183 b (if left uneorrected). There are no 
irregularities in the use of the amphibrach. Brevis may stand 
at the end of the first hemistich. 1 

Two cases occur in Harivah§a. I give the text: 

1, 1, 3, jayati Para^arasunuh satyavatihrdayanandano Vyasah 
yasya ’syakamalagalitam vaiimayam amrtam jagat 

1, 1, 7, yo Harivarujaiii lekhayati yatha vidhina mahatapah 

(in C.) sa yati Hareh padakamala[m] kamalam yatha madh- 
upo lubdhah 

(in B.) sa jayati Haripadakamalam madhupo hi yatha rasena 

The first stanza is regular. The second neglects the usual 
caesura after the third foot in the first hemistich in both texts ; 
while C.’s text is impossible in the second, though the metre 
may be set right by omitting the antecedent and reading 
(without sa) : 

— w w, yu — ; w w — kj , w, , _ 

The text of B. is regular, with w _ y/ as sixth foot, where (in 
the second hemistich) stands w in the cases above. 

On page 164, 1 cited in full a stanza beginning: ahuh sastim 
buddhigunan vai (the sixty Samkhya gunas) ; the scheme 
(unique in the epic) for the whole stanza being (xii, 256, 12) : 

WV/ w W , W W W W w 

Although this lacks the marked characteristics of the arya, 
both in its early and in its later forms, it is yet a gana metre 
which may be reckoned either as aryagiti, or as matrasamaka, 
but not pure. 

As to the origin of the ganacchandas, the metre seems to 
me to be rather a species than a genus. As seen in the speci- 

1 There is here no case of four breves in the sixth foot of the second hemi- 
stich, which occurs in classic writers and inscriptional ary as, e. g., Vatsa- 
bhatti, loc. cit., vs. 39. 



mens above under matrasamakas, they are interchangeable 
with the latter, of which they are only a more special type, 

with slv slv, w _ w, vy vtv, as the last four feet of the 

hemistich (compare 183 a, only this is not in the aryaglti 
form, but has the alternative one heavy syllable for two, or 
two morse for four). The matrasamakas in turn are the 
equivalent in morse of the §loka strophe (that is, a unit com- 
posed of two glokas, such as the classical uniters affect), the 
thirty-two syllables of the half strophe answering to the thirty- 
two morse of the hemistich in the matrasamaka and aryaglti 
(the one mora of the sixth foot and two morse of the eighth 
foot being special modifications). 1 

The Distribution of Fancy-Metres in the G-reat Epic. 

The relation of §loka and tristubh, 2 which in the whole 
Mahabharata stand numerically in the rough proportion of 
95,000 to 5000 (out of 101,900 stanzas or prose equiva- 
lents, the sum of the whole), varies enormously from book 
to book, one tristubh to three hundred and ten Qlokas in the 
eighteenth book, almost nine hundred tristubhs to four thou- 
sand glokas in the eighth book, the extremes in absolute 
number of tristubhs as well as in their proportion to Qlokas. 

From reasons quite apart from metre, I have elsewhere 
maintained that the first part of book i, and book xiii, with 
the Harivaruja were late, as compared with books vii, viii, xii, 
but that these in their turn contain very late additions to 

1 One may, indeed, take the ?loka hemistich in the form w , 

w — w — and reckon it in mone, 15 -f 14, as a hemistich of a 

matrasamaka, which is as nearly correct, that is as near to a real samaka, 
as are the cases above, where the pada may have 15, 16, or 17 mora. But I 
prefer to rest with the fact that the matrasamaka is a parallel in terms of 
mora to the floka-strophe in terms of syllables, without attempting a deriva- 
tion. For particular studies of the ganacchandas, see Professor Cappeller’s 
Die Ganacchandas, and Professor Jacobi, ZDMG. vol. xxxviii, p. 595 ff. The 
latter scholar believes the arya to have been a musical adaptation, and to 
have come into Sanskrt from Prakrt poetry. The metre can be traced back 
to the time of Afoka. 

2 That is tristubh and jagati. There are just about the same number of 
tristubh-jagati stanzas in the Mahabharata as in the Big Veda. 



the original epic, often palpable intrusions. 1 The use of the 
fancy-metres seems to illustrate the general correctness of my 
former analysis. Thus the rueira occurs in i, iii, vii, xii, xiii, 
Hariv. ; the vasantatilaka only in i, xiii, xviii, Hariv. ; the 
malinl only in vii, viii, xiii, Hariv.; the arya only in xiii, 
Hariv. The tag-metres of Adi are confined to the first quar- 
ter (two thousand) of the eight thousand in the whole book. 
They cease after Sarpasattra (almost after the beginning 
of Astlka), or, in other words, they occur almost entirely in 
the most modem part of the book. Books ii, v, and vi have 
no fancy metres at all ; book ix has but one, a bhujamgapra- 
yata. On the other hand, books iv, x, xi, xiv, xv, xvi, and 
xvii have none also, which however, need not surprise us 
much, as most of them are short supplementary books, and 
the fourteenth is mainly an imitation of the Gita. That the 
fourth book is not adorned with these metres indicates perhaps 
that it was -written between the time of the early epic and the 
whole pseudo-epic. The much interpolated eighth book would 
be comparatively free from these adornments were it not for 
its massed heaps of arclhasamavrttas, twenty-five in all (other- 
wise it has only one gardulavikiidita and five malinls). The 
seventh book, on the other hand, has two drutavilambitas, 
nine ruciras, one praharsinl, one malinl, and eight ardhasama- 
vrttas, — twenty-one in all. The first book, that is, its first 
quarter, has thirty-one, of which twenty-two are ruciras ; four, 
praharsinls ; three, vasantatilakas ; two, arclhasamavrttas. The 
pseudo-epic shows the greatest variety, as well as of course 
the greatest number, the books represented (with the ex- 
ception of one vasantatilaka in the eighteenth) being the 
twelfth, thirteenth, and HarivanQa, with 48 2 , 28), and 43, 

1 Compare the paper on the Bharata and Great Bharata, AJP., vol. xix, 
p. 10 fE. That there are antique parts in books generally late, no one I be- 
lieve, has ever denied. Nor has any competent critic ever denied that in 
books generally old late passages are found. Adi, Vana, and Anu^asana, 
and in a less degree Kama, are a hodge-podge of old and new, and the only 
question of moment is whether in each instance old or new prevails or is 



The number of occurrences of each metre, according to 
the books in which they are found, is given in the following 

Cases occurring in books. 











Rathoddhata . . . 







Drutavilambita . 



Vai£vadevl .... 


• • 











PraharsinI .... 


♦ • 




. • 



Vasantatilaka . . 


. • 

. • 











. • 


. . 


. . 


Puspitagra \ 


Aparavaktra > . 







. . 



Matrasamaka ) 














How are we to account for these fancy-metres ? Let us 
imagine for a moment — to indulge in rather a harmless 
fancy — that the whole epic was written by one individual, 
not of course by Vyasa the arranger, but by Krit the maker, 
even as the pseudo-epic says; though the latter sets reason- 
able bounds to the human imagination and very properly adds 
that the maker of such a poem must have been divine. 

This superhuman being, Krit (Bharatakrt or better, Maha- 
bharatakrt) must have had from the beginning a well-devel- 
oped ear for fancy-metres. When he writes them he writes 
them very carefully, seldom opposing the rules that later 



writers, say of 500 A. D. and later, impose upon themselves, 
except in the case of the ardhasamavrttas. These at one 
time he writes correctly and another loosely, as if he occa- 
sionally failed to grasp the distinction between this class of 
metres and that of the strict matracliandas ; which is rather 
peculiar, when one considers how correctly he writes at other 
times. But, passing this point, how are we to account for 
the distribution of these metres? Evidently there is only 
one way. Having started out with the statement that the 
poem was to glitter with various fancy-metres, the poet first 
gave an exhibition of what he could do, reserving, however, 
the more complicated styles for the end of the poem. Then, 
settling down into the story, he got so absorbed in it that he 
forgot all about the fancy-metres, till after several thousand 
stanzas he suddenly remembered them and turned off three 
ruciras and six ardhasamavrttas, e. g., as tags, lauding (fiva's 
gift and Arjuna’s glorious trip to heaven ; but then, becoming 
interested again, again dropped them, while he wrote to the 
end of the sixth book. With the seventh book, feeling that 
an interminable series of similar and repeated battle-scenes 
was getting a little dull, he sprinkled five different kinds of 
fancy metres over his last production, and in the eighth 
emptied a box of them in a heap, wdiich lasted till the first 
part of the poem was complete. On resuming his labors (we 
are expressly told that he rested before taking up the latter 
half of the poem) he decided that, as all interest in the story 
itself was over, the only way to liven up a philosophic en- 
cyclopedia would be to adorn it with a good many more 
fancy-metres, and toward the end he brought out the aryiis, 
which he had had concealed all the time, but kept as a final 
attraction. In this last part also he emptied whole boxes of 
metres together, just as he had done so desperately in the 
eighth book. 

This seems to me an entirely satisfactory explanation, 
granting the premiss. But in case one is dissatisfied with 
the (native) assumption of a homogeneous Homer, one might 
consider whether it were not equally probable that the present 



poem was a gradual accumulation and that fancy-metres were 
first used as tags 1 to chapters in the later part of the work, 
as an artistic improvement on the old-fashioned tristubh tag 
(to §loka sections) ; and so find the reason why the masses 
of fancy-metres are placed in the middle of sections in a 
later exaggeration, a vicious inclination to adorn the whole 
body with gewgaws, whereas at an earlier date it was deemed 
a sufficient beauty to tag them on to the end of a section. 
The only difficulty in this assumption is that it recognizes 
as valid the delirament of believing in the historical growth 
of the epic. 

As regards the arya, it makes no difference whether it was 
a Prakrit style known before the epic was begun or not. Just 
as in the case of the Rig Veda, the point is not whether such 
and such a form existed, but only whether (and if so, in how 
far) the poets admitted the form into hymns ; 2 3 so here, the 
question is simply as to when Sanskrit writers utilized Prakrt 
melodies. It is somewhat as if one should properly try to 
define the decade in which a piece of X’s music was com- 
posed by considering that it was in rag-time. One might 
object that rag-time melodies have been used for unnumbered 
decades by the negroes. The reply would be : True ; but it 
is only in the last decade of the nineteenth century that 
rag-time has been utilized by composers ; ergo, X must have 
published his composition in that decade or later. 

When then did the vulgar arya (i. e., melody used as a 

1 The expression tag-metres answers exactly to the function of the fancy- 
metres in the Ramayana, and pretty closely to their function in the Bliarata. 
I have indicated above the few cases where in the latter poem they have been 
inserted in other positions. There can be no serious doubt that such medial 
position simply shows how late is the passage where are found such stanzas 
thus located. The bhujamgaprayata appears in medial position in (Janti ; the 

drutavilambita, in Drona ; where also the rucira (usually only tag) ; the pra- 
harsini (medial), only in Drona and tjianti ; the vasantatilaka, generally a tag, 
medial only in Anuyasana ; the maiini, medial in lvarna ; the yardfilavikrldita, 
medial in both these last. 

3 The all-sufficient answer to the unsatisfactory contention that, because 
certain Vedic forms are pre-Vedic, therefore their employment by Vedic poets 
cannot be used in evidence of the age of certain hymns. 



frame for literature) appear in Sanskrit poetry ? Tlie author 
of the Ramayana, using freely the aksaracchandas and ardha- 
samavrttas as tag-poetry, either knew it not or ignored it. 
The later poets of the Mahabharata, doing the same, ignored 
it also. Only the poets of the latest tracts, the fourteenth 
section of Anugfisana and benedictions in Harivanga, used it, 
whether inventing or utilizing is a subsidiary question. The 
employment of this metre, if borrowed from the vulgar, stands 
parallel, therefore, to the adoption of Prakrit licence in 
prosody . 1 

Further, the sometime intrusion into the middle of a chap- 
ter of metres used originally only as tags, shows that parts of 
the Mahabharata reflect a later phase than that of the Rama- 
yana, which still confines them to their earlier function. In 
fact, the Mahabharata is here on a level with the poems of 
inscriptions where all metres are flung together , 2 and, like 
these poems, its later parts show a predilection for long com- 
pounds and for long sentences extending over many verses. 

The total result of a comparison of the various metres in 
the two epics shows in outline : 

In the Mahabharata In the Ramayana 

(a) early (Yedic) gloka 
early (Yedic) tristubh 

(b) almost classical gloka (b) almost classical gloka 

classical tristubh classical tristubh 

(c) late gloka stanzas (pure (c) early use of fancy metres 

late tristubh stanzas (ga- 

late use of fancy metres 

A review of the results obtained in regard to the chief 
metre of the epic makes it clear that the presence in the 

1 Only xiii, 14 is really affected. The benedietive Harivaiifa verses are 
an addition too late to affect dates. Even the native (Bombay) edition omits 
them from the text proper. 

2 See on this point, Bidder's essay, Das Alter der Indisclien Kunstpoesie, 
with examples at the end. 



Mahabharata of glokas of an older and also later type than are 
found in the Ramayana indicates not only that the style of 
the Mahabharata is more antique in one part than in another, 
but also that this difference is not due to conscious metrical 
variations on the part of one poet ; or, in other words, that 
the epic was not made all at once. For the general shape of 
cl okas might voluntarily be shifted, though even here it is 
not probable that a poet who wrote in the refined style com- 
mon to the Ramayana and to parts of the pseudo-epic Maha- 
bharata would shift back to diiambic close of the prior pada 
or a free use of the fourth vipula. But even granting this, 
there remain the subtle differences which are perceptible only 
with careful and patient study, elements of style not patent 
to the rough-and-ready critique which scorns analysis. The 
poet who had trained himself to eschew first vipulas after 
diiambs and renounce a syllaba anceps would not write first 
in this particular style and then in the careless old-fashioned 
manner. The very presence of the more refined art precludes 
the presumption that the same poet in the same poem on the 
same subject would have lapsed back into barbarism. For 
the distinction is not one that separates moral discourses from 
the epic story. Except in the case of a few obvious imita- 
tions or parodies of Cfruti texts, topics of the same sort are 
treated with a difference of style attributable only to different 
authors and in all reasonable probability to different ages. 



We have now reached a point where an intelligent opinion 
may be formed in regard to the general make-up of the Mar 
habharata. It is based, as was shown in the second chapter, 
on a more or less stereotyped diction, and contains adventitious 
matter common to both epics. It contains allusions to the 
latest pre-elassical works, as was shown in the first chapter ; 
while its didactic parts recapitulate the later Upanishads ; and 
it shows acquaintance with a much larger number of Vedic 
schools than were recognized even at a late date. Its philo- 
sophical sections, as was shown in the third chapter, reflect 
varied schools and contradictory systems, some of which are 
as late as our era. Its metres, as have just been explained, 
preclude the probability of its having been written by one 
poet, or even by several poets of the same era. It appears to 
be a heterogeneous collection of strings wound about a 
nucleus almost lost sight of. The nucleus, however, is a 

This story is in its details so abhorrent to the writers of the 
epic that they make every effort to whitewash the heroes, at 
one time explaining that what they did would have been 
wicked if it had not been done by divinely inspired heroes ; at 
another frankly stating that the heroes did wrong. It is not 
then probable that had the writers inteifded to write a moral 
tale they would have built on such material. Hence the tale 
existed as such before it became the nucleus of a sermon. 
There are then two elements in the epic, narrative and 

In its present didactic form the epic is recited. At its own 
close we learn that it was not given as a dramatic recitation, 
still less as a rhapsodic production. A priestly reciter, vaeaka, 



patkaka, “ speaks ” or “ reads ” the epic as “ he sits com- 
fortably and recites, carefully pronouncing the sixty-three 
letters (sounds) 1 according to their respective eight places of 
utterance ” (as gutturals, etc.). He reads from manuscripts, 
samhitapustakas, which, after the performance is over and the 
gentleman has been dismissed with a brahmasutra and a hand- 
some fee, are wrapped in cloth and piously revered. The 
recitation takes four months, and should be performed by 
Brahmans during vasso, the rainy season, xviii, 6, 21 £f. (i, 
62, 32). 

Such recited stories are recognized elsewhere. A knight 
leaves town to go into the woods accompanied with “ priests 
who know the Vedas and Vediingas,” and “ priests who recite 
divine tales,” divyakhyanani ye ca ’pi patlianti, but also, and 
distinguished from these, with sutah piiuranikah and kathakah 
(besides hermits, §ramana<j ca vanaukasah), i. 214, 2-3. 
The story-tellers here named may be represented again by 
knights who tell each other, as they sit and talk, “ the glorious 
deeds of old and many other tales,” or, as it is expressed else- 
where, “ tales of war and moil and genealogies of seers and 
gods.” 2 

But buried with the story-nucleus are elements also more or 
less concealed. The first of these is the genealogical verses, 
anuvan 9 a<;loka, or anuvantjya gatlia, which in the extract 

1 samskrtah sarvayastrajiiah . . asariisaktaksarapadam svarabhavasamanv- 
itam trisastivarnasarnyuktam astasthanasamiritam vacayed vacakah svasthah 
svasinah susamahitah, xviii, 6, 21, and H. loc. cit. in PW. s. varna. In the 
enumeration of parvans following, the Anuyasana is omitted, as it is in one 
of the lists in Adi, whereas the other list makes it a separate work : “ After 
this (i. e., after Canti as rajadharmanuyasana, apaddharma, and moksa) with 
329 or v. 1. 339 sections and 14,732 ylokas [our text has 13,943 stanzas of all 
kinds] must be reckoned the Anuyasana with 146 sections and 8000 ylokas ” 
[our text 7796] ; where atah urdhvam shows, with the figures, that the Anu- 
yasana is not included with Qanti (the former is also called anuyasanikam 
parva), i, 2, 76-78, 328-331. On the list i, 1, 88 ff. which omits the thirteenth, 
seventeenth, and eighteenth books, see AJP. xix, p. 5. 

2 tatra purvavyatitani vikrantani ’tarani ca bahuni kathayitva tau remate, 
i, 222, 29; praviyya tarn sabham ramyam vijahrate ca, Bharata; tatra yud- 
dhakathay citrah parikleyahy ca, I’arthiva, kathayoge kathayoge kathayam 
asatuh sada, rslnam devatanarii ca vanyans tiiv ahatuh sada, xiv, 15, 5-7. 


just referred to are recited ; as, again, in iii, 88, 5, is found : 
Markandeyo jagau gathiim (anuvantjyam). Suck memorial 
stanzas in honor of the family either are strictly genealogical, 
“ Devayanl bore Yadu and Turvasu,” or characterize a man, as 
in i, 95, 46 (the other, ib. 9), where (Smtanu’s name is de- 
rived, on the strength of such a stanza, from his having the 
healing touch (the careless compilers a little later, i, 97, 19, 
give another derivation). Such stanzas are sometimes inserted 
in prose narration 1 in honor of the family, though occasionally 
of very general content. On the other hand, really genealogi- 
cal stanzas may be introduced without any statement as to 
their character, though the poets usually quote them from 
rhapsodes, “ men who know the tales of old here sing (or 
recite) this gatha,” apy atra gathaih gayanti ye puranavido 
janah, i, 121, 13; vii, 67, 14. 

Though, as was shown in a previous chapter, the word for 
sing is scarcely more than recite, yet it reflects conditions 
where bards actually sang songs in honor of kings. The 
ancient age knew, in fact, just such a distinction as underlies 
the double character of the epic. On the one hand, it had its 
slowly repeated circle of tales (sometimes mistranslated by 
cycle of tales), 2 and on the other, impromptu bardic lays, not 
in inherited form but improvisations, where the rhapsode, as 
is especially provided for in the ritual, on a certain occasion 
was “ to sing an original song, the subject of which should be, 
This king fought, this king conquered in such a battle.” The 
song is here accompanied with the lute or lyre, which in the 
epic is called seven-stringed, saptatantrl vina (ewraTovos 
<pop/xiy£, see above, p. 172). Such song as “hero-praising 
verse,” naratjans! gatha, are recognized in the Grhya Sutras 

1 Compare the illustrations by Lassen and 'Weber and Holtzmann himself, 
summed up in the last writer’s work. loc. cit., p. 2. 

2 The tales of a (year’s) circle, pariplavam akhyanam, hare no cyclic ele- 
ment. Tor literature on the early rhapsodes and reciters, see Qat. Br. xiii, 4, 
3,3,5; Weber, IS. i, p. 186. Compare Par. G. S. i, 15, 17. The traditional 
legend in the epic is called (param) paryagatam akhyanam kathitam. xii, 340, 
125, 138. The early improvised lays are called STayamsambhrta gathah (loc. 
cit., £at. Br.). 



as traditional texts, differentiated from “ legends,” and were 
probably genealogical stanzas preserved in the family. Be- 
sides the single singer, there were also bands of singers who 
“ sang the (reigning) king with the old kings.” 1 

In these Brahmana stories, the rhapsode-lay took place at 
the very time of the priestly recited tale, which circled round 
a year. There is no difference of date between them. The 
rhapsode and the reciter were contemporary. So in the epic, 
although the recitation of tales is noticed, yet rhapsodes are 
constantly mentioned. In xiv, 70, 7, praise is rendered by 
dancers and luck-wishers and also by rhapsodes, granthikas, 
“ in congratulations that uttered the praise of the Kuru-race,” 
Kuruvaii§astavakhyabhir aglrbhih. In parallel scenes we find 
“ story-tellers,” who could praise only by performing their 
business, as in iv, 70, 20: “ Eight hundred bejewelled Sutas 
along with magadhas (singers) praised him, as the seers did 
(Jakra of old;” ib. 72, 29: “Singers, gayanas, those skilled 
in tales, akhyana§Uas, dancers, and reciters of genealogical 
verses, natavaitalikas, 2 stood praising him, as did Sutas with 
magadhas.” Again in vii, 82, 2-3: jagur gitani gayakah 
Kuruvan§astavarthani : “ Singers sang songs which lauded the 
Kuru-race,” where the rhapsode, granthika, above, appears as 
singer, gayana. For the history of the poem it is worth 
noticing that, though the Pandus are the present heroes, the 
stereotyped phrase is always of “praise of the Kuru race,” 
even where a Pandu is praised. 

We have in the epic the names of what are to-day the 
epic reciters, kugilava and kathaka, and the repeaters of 
genealogical verses (in distinction from the Sutas), 5 called 

1 Weber, loc. cit., and Episches ira Vedischen Ritual, p. 6. 

a So in ii, 4, 7, natas, Sutas, and vaitalikas wait on the king along with 
boxers and wrestlers. Such epic professionals are called (besides panisvani- 
kas) magadhas, n^ndivadyas, bandins, gayanas, saukhya<?ayikas, vaitalikas, 
kathakas, granthikas, gathins, ku^ilavas and pauranikas (Sutas). 

8 So xii, 37, 43, where a king is praised by Sutas, vaitalikas, and (subhasita) 
magadhas. Compare the distinction in R. vi, 427, 3, with Comm. : the Sutas 
“ know praise and Puranas ” the vaitalikas recite genealogical verses. Both 
epics have the group (phrase) siitamagadhabandinah. 


When the lyre is mentioned, it is to wake up sleepers by 
means of “sweet songs and the sound of the lyre,” glta, 
vlna§abda, i, 218, 14. Only Narada, a superhuman archetypi- 
cal bard, comes skilled in dance and song with his melodious 
mind-soothing tortoise-lyre, ix, 54, 19. 1 

There is then in the epic, though a musical accompaniment 
is unknown, a distinct recollection of the practice of reciting 
lays, gltani, the sole object of which was to “ praise the Kuru 
race,” as opposed to reading or reciting conversationally stories 
of ancient times. To neither of these elements can a judicious 
historian ascribe priority. The story and the lay are equally 
old. Their union was rendered possible as soon as the lay, 
formerly sung, was dissociated from music and repeated as a 
heroic tale of antiquity. This union was the foundation of 
the present epic. 

Traces of the epic quality of the early poem cannot be 
disregarded. The central tale and many another tale woven 
into the present narrative are thoroughly heroic. To this 
day, warped and twisted from its original purpose, it is the 
story, not the sermon, that holds enthralled the throng that 
listens to the recitation of the great epic. Be it either epic, 
its tale is still popular in India. But the people cannot 
understand it. Hence the poem is read by a priest, while 
a translator and interpreter, of no mean histrionic talent, 
takes up his words and renders them in forcible patois, ac- 
companying the dramatic recital by still more dramatic ges- 
tures and contortions. Such a recitation, without the inter- 
mediate interpreter (the modern dliaraka) was undoubtedly 
the performance given (not by the later pathaka, but) by the 
earlier epic gathin, gayaka, and granthika, just as they are 
depicted about the second century b. c. on the Sanchi Tope. 2 

1 The panisvanikas mentioned above may be pantomimists or simple 
“hand-clappers.” The latter is the meaning in the cognate panivadaka at 
R. ii, 65, 4 (compare Brahmajala Sutta, Rhys Davids’ note, p. 8). In the pas- 
sage above, ix, 54, 19, the prakarta kalahanam ca nityam ca kalahapriyah is 
represented as kacchapim sukhasabdantam grhya vinam, a late passage, 

2 Le'vi, Le the'atre indien, p. 309. 



But though it is a gross exaggeration of the facts, as well as 
a misapprehension of poetic values, to make the epic a poem 
that was from the start a moral and religious narrative, yet, 
inasmuch as in the hands of the priest the latter element 
was made predominant, there is no objection to the statement 
that from the point of view of the epic as a whole the M aha - 
bharata is to-day less tale than teaching. That this double 
character was recognized by those who contributed the in- 
troduction to the poem itself is indisputable (above, p. 53). 
The “tales” are counted as separate. The original Bharata 
was only a quarter of its present size. Then, as later, the 
different elements were still distinguished, and the poem was 
not regarded as wholly a Smrti or instruction-book, but as an 
artistic poem, Ivavya, per se. So the pseudo-epic vaunts its 
own literary finish: gahde cii ’rthe ca lietau ca esa prathama- 
sargaja (sarasvati), xii, 336, 36. 

The particular school of priests in whose hands the epic 
was transformed was probably that of the Yajurvedins. The 
Yajur Veda is “the birth-place of the warrior caste,” accord- 
ing to a well-known verse, and it has been shown by Weber 
that the Qatapatha, a Yajur Veda text, stands in peculiarly 
close relation to the didactic epic. 1 As has been shown 
in the first chapter, the yatapatha is the only Brahmana 
praised, perhaps even mentioned, in the epic ; while the 
Yajur Veda yatarudriya is exalted above all texts (except 
perhaps where Indra sings this, Vishnu sings the jyestha 
saman, and Brahrnd, the rathamtara, xiii, 14, 282, but even 
here the yatarudriya is not slighted). In dividing the Iti- 
hasa from the Purana, moreover, the epic groups the former 
with the Yajur Veda, as against the Purana with the other 
Vedas, viii, 34, 45. Here the Itihasa represents the epic, as 
it does in the similar antithesis of xii, 302, 109 : yac ca ’pi 
drstam vividham purane yac ce ’tihasesu mahatsu drstam, 

1 Valmiki too belonged to this school. Compare Weber, IS., xiii, p. 440, 
and as cited by Holtzmann, loc. cit , p. 18 ; Muir, OST., i, p. 17, citing TB. 
iii, 12, 9, 2, where the Vai?yas are derived from the Rig Veda, the Ksatriyas 
from the Yajur Veda, and the Brahmans from the Sama Veda. 


where, as already observed, the Great Itihasas point to sev- 
eral epic poems. Lastly, the Upanishads especially copied in 
the epic are those belonging to the Yajur Veda. 

But while this is true of the completed epic, there is noth- 
ing to show that the Bliarati Katha was the especial property 
of any school, and no preference is given to the Yajur Veda 
in the later epic, for in the Gita the Sama Veda stands as the 
best, “ I am the Sama V eda among V edas,” 10, 22, and this 
is cited with approval and enlarged upon in xiii, 14, 323 : 
“Thou art the Sama Veda among Vedas, the ^atarudria 
among Yajus hymns, the Eternal Youth among Yogas, Kapila 
among Sarhkhyas.” 

In the epic itself the Sutas called pauranikas are recognized 
as the re-writers and reciters of the epic. They probably took 
the epic legends and arranged them in order for the popular 
recitation, which is also recognized when “priests recite the 
Mahabharata at the assemblies of warriors,” v, 141, 56, a 
passage recently cited by Professor Jacobi, as evidence of a 
difference between the manner of handing down the heroic 
tales and the recitals of legends. 1 

The method of narrating the epic stories is that of the old 
priestly legend, where the verse-tale is knit together, as in the 
epic, by prose statements as to the speaker. So in the epic, 
a narrative, not a rhapsodic or dramatic, delivery is inchoated 
by such phrases. In the Ramayana, on the other hand, the 
verse is knit more closely together, and the speakers are 
indicated almost always in the verse. The one exception is a 
late addition (G. ii, 110, 4-5). 

The Mahabharata is not only a Veda, it is so important a 
Veda that to read it is to dispense with the need of reading 
other Veclas. 2 In the dynamic alteration consequent on the 
attaining of such an ideal, w'e may expect to find that the tale, 
as a tale, is full of the grossest incongruities ; for to fulfil its 

1 Gbttingische Gelehrte Anzeige, 1899, p. 877 ff. I fully agree with the 
author’s view in regard to the “ Puranic ” Sutas being the compilers of the 
epic mass. 

2 rijneyah sa ca vedanam parago bharatam pathan, i, 62, 32. 



encyclopedic character all is tisli that comes to the net, and 
scarcely an attempt is made to smooth away any save the 
most g lar ing inconsistencies. Tale is added to tale, doctrine 
to doctrine, without much regard to the effect produced by 
the juxtaposition. If we take these facts as they stand, 
which is the more probable interpretation, that they were 
originally composed hi this incongruous combination or that 
they are the result of such a genesis as has just been ex- 
plained ? As for the facts, I will illustrate them, though to 
any Bharatavid they are already patent. 

In i, 214, Arjuna protests that he is a brahmacarin for 
twelve years, in accordance with the agreement (chapter 212) 
that he has made with his brother, which is to the effect that 
he will be “ a bralimacarin in the woods for twelve years.” 
This can have only one meaning. A bralimacarin is not a 
man wandering about on love-adventures, but a chaste stu- 
dent. Above all, chastity is implied. Now the first thing 
the hero Arjuna does is to violate his agreement by having a 
connection with Ulupi, a beautiful water-witch, who easily 
persuades him to break his vow; after which he resides in 
a city, taking to himself a wife with whom he lives for three 
years. After this he has a new adventure with some en- 
chanted nymphs and then stays with Krishna ; when, in a new 
vikranta or derringdo (the hero’s rape of Subhadra, chapter 
220), all the talk of bralimacarin wandering in the woods stops 
ineonsequently. When he marries (in town) not a word is 
said of his vow ; but when he approaches Krishna on the sub- 
ject of Subhadra the poet makes the former say “ how can a 
wood-wanderer fall in love ? ” This is the only allusion, and 
one entirely ignored, to the matter of the vow ; which in the 
earlier Manipur scene is absolutely unnoticed. Each of these 
feats is a separate heroic tale and they are all contradictory to 
the setting in which they have been placed by the diadochoi 
and later epic manipulators. As heroic tales they are per- 
fectly intelligible. Certain feats in separate stories were 
attributed to the hero. They had to be combined and they 
were combined by letting him go off by himself under a vow 


of wandering in the woods. The wood-wanderer was usually 
a chaste ascetic, so he was given this character, but this role 
is kept for only one of the noble deeds. For after he has 
protested once at the outset, all pretence of his being a brah- 
macarin vanishes and the next we know he is comfortably 
mated and living in town, while still supposed by the poets 
to be a brahmacarin in the woods. The independent origin 
of these stories is seen at the beginning in the formula “ Hear 
now a wonder-tale of him,” tatra tasya ’dbhutam karma grnu 
tvam, 214, 7. Such formulae of special tales are found fre- 
quently, idarii yah ymuyad vrttam is another, used for the 
UrvagI episode, iii, 46, 62. Another is like our “ once upon 
a time,” pura krtayuge rajan, e. g., ix, 40, 3. 

The fact that Arjuna is here banished for twelve years is 
not without significance. The epic has been completed on 
rather formal lines. Agni is satiated for twelve years at 
Ivhandava. Arjuna’s banishment is for the same length of 
time as that of the brothers as a family. So the epic is 
divided into eighteen books, as there are eighteen Puranas (p. 
49) ; and there are eighteen armies battling for just eighteen 
days, and eighteen branches of younger Yadavas; 1 while 
finally there are eighteen islands of earth. The number of 
islands deserves particular notice, as it is one of the innumer- 
able small indications that the poem has been retouched. 
Earth has four, seven, or at most thirteen islands in all litera- 
ture of respectable antiquity. Seven is the usual number in 
the epic as it is in the older Puranas, but in the hymn to the 
sun at iii, 3, 52, “ earth with its thirteen islands ” is men- 
tioned. 2 The mention of eighteen is found, of course, in one 
of the books where one who distinguished between the early 
and late elements would be apt to look for it, in the much 
inflated and rewritten seventh book, where (above, p. 229), 
with customary inconsistency, it stands beside another refer- 
ence to the usual seven islands, sarvan astada§a dvipan, vii,~ 
70, 15; sapta dvipan, 21. 

1 ii, 14, 40, 55 ; also 18,000 brothers and cousins. 56. 

2 The same passage calls the sun, 9I. 61, viyasvan mihirah pusa mitrah. 



Another tale which bears evidence of having been rewritten 
and still shows its inconsistencies is found in iii, 12, 91 if. 
Here Bhima and his brothers and mother are surrounded by 
fire, and he rescues them by taking them on his back and 
leaping clear over the fire. No suggestion is given of any 
other means of escape. On the contrary it is emphasized that 
he can fly like the wind or Garuda, and the escape is due 
entirely to his divine power and strength. But in i, 2, 104 ; 
61, 22; and 148, 12, 20 ff., the same story is told with an 
added element which quite does away with the old solution. 
Here (in the later first book) the party escape through an 
underground tunnel, suranga (9I. 12) or surunga, and after 
they are well off in the woods far from the fire, Bhima is 
made to pick them up and carry them. The old feat was too 
attractive to lose, so it was kept postponed, but the later ver- 
sion with the Greek word to mark its lateness takes the place 
of the older jump. No one can read the account in Vana and 
fail to see that it is not a mere hasty resume omitting the 
surunga, but that the original escape is a feat of the wind-god’s 
son. But the first part of this same section in Yana contains 
a laudation to Krishna-Vishnu which is as palpable a late 
addition as one could find in any work. 

The surunga, “ syrinx,” is not the only Greek word added 
in the later epic. As such must certainly be reckoned trikona 
= t piyaivos. There are in fact two kona. One is Sanskrit or 
dialectic for kvana, the “ sounder,” or drumstick of the Rama- 
yana, vi, 32, 43 ; 42, 34, and elsewhere (not in the Mahabha- 
rata). The other is found in the pseudo-epic xiv, 88, 32: 
catugcityah . . . astada§ akaratm: ikah sarukmapakso nicitas tri- 
kono garudakrtih, of an altar (the corresponding passage in R. 
i, 14, 29, has trigunah), where the word must mean angle and 
be the equivalent of rplywvos. 

The question of the character of the epic is so intertwined 
with its date that I will not apologize for pausing here a mo- 
ment to speak of another geographical and ethnographical 
feature. The apologia published under the title Genesis des 
Mahabharata omits to reply to the rather startling conclusion 


drawn by Weber in a recent monograph on the name Bahllka, 
or Balhlka, as it appears in the epic. In the Sitzungsberickt 
of the Berlin Academy, 1892, pp. 987 ff., Weber claims that 
any work containing this name or that of Pahlava must be as 
late as the first to the fourth century a. d. I cannot but think 
that the escape from this conclusion, in part suggested by 
Weber himself, is correct. In the rewriting of foreign names it 
is perfectly possible that later copyists should have incorpo- 
rated a form current hi their own day rather than conserved a 
form no longer current, which it was easy to do when not for- 
bidden by the metre. Again, that there was actual confusion 
between the forms Vahlka and Balhlka, the former being a Pun- 
jab clan, the latter the Bactrians, it is not difficult to show. 
According to tradition, a drink especially beloved by the Balhl- 
kas is sauvira, or sauvlraka. This can scarcely be anything 
else than the drink suvlraka, said to be lauded in the epic by 
degraded foreigners. But here the foreigners are not Bahlikas 
but Vahlkas, whose Madrika (woman) sings, viii, 40, 39-40, 
“ I will give up my family rather than my beloved suvlraka,” 

ma maiir suvlrakam kaccid yacatam dayitam mama 

putrarh dadyam patim dadyam na tu dadyam suvi- 

It is possible that the epic arose further to the north-west, 
and in its south-eastern journey, for it ends in being revised in 
the south-east, 1 has transferred the attributes of one people to 
another, as it has transferred geographical statements, and 
made seven Sarasvatls out of the Seven Rivers of antiquity, 
ix, 38, 3. As an indication of the earlier habitat may be men- 
tioned the very puzzling remark made in iii, 34, 11. Here 
there is an apparent allusion to the agreement in ii, 76, which 
agreement is that on being recognized before the expiration of 
the thirteenth year, either party shall give up his kingdom 
(svarajyam, §1. 14) ; and it is assumed throughout that the 
two kingdoms are those of Hastinapur on the Ganges and 

1 See on this point the evidence presented in my paper on the Bharata and 
the Great Bharata. Am. Journ. Phil. vol. xix, p. 21 ff. 



Indraprastha on the Jumna. But in the passage of Vana just 
referred to there is an (old) tristubh resume of the situation, 
which makes the Kuru say : 

bravlmi satyam KurusamsadI ’ha 
tavai ’va ta, Bharata, pafica nadyah 

Here we get an account where the Pandus are lost in the older 
Bharatas, and to them the Kuru king says, “ If we break this 
agreement, yours shall be all this Punjab.” But what has the 
Punjab to do with the epic in its present form ? It is a land 
of Yahikas and generally despised peoples (who morally are 
not much better than barbarians), and also a holy land (an- 
other little inconsistency disregarded in the synthetic method) ; 
but, whatever it is morally, it has nothing to do politically with 
the present epic heroes, except to provide them with some of 
their best allies, a fact, however, tha’t in itself may be signifi- 
cant of earlier Western relations. 1 

To return to the evidence of remaking in the epic. Passing 
over the passage ix, 33 to 55, a long interpolation thrust mid- 
way into a dramatic scene, we find that chapter 61 begins with 
the repetition of the precedent beginning of chapter 59, which 
latter, after 15 glokas, together with chapter 60, is taken up 
with a moral discourse of Yudhisthira, who reproaches Bhima 
for insulting the fallen foe. Then Rama joins in and is about 
to slay Bhima, when Krishna defends the latter, saying that his 
ignoble insult was entirely proper. This argument of Krishna 
is characterized by Sanjaya as dharmacchalam, or, in other 
words, Krishna is said to be a pious hypocrite (60, 26) ; Rama 
departs in disgust, and the virtuous heroes “became very 
joyless ” (31). Then Krishna, who has all along been approv- 
ing the act, turns to Yudhisthira who reproved it, and says, 

1 Jacobi touches on the significance of these Western allies in the review 
mentioned above. The “ land of the Bharatas ” extends northwest of the 
Punjab even to the foot of the Himalayas, for in coming from Hemakuta to 
Mithila one traverses first the Haimavata Varsa, then “passing beyond this 
arrives at the Bharata Varsa, and (so) reaches Aryavarta” (seeing on the 
journey “ different districts inhabited by Chinese and Huns,” cinahunanise- 
vitan), xii, 326, 14-15. But this is the Varsa or country in general. 



“Why do you approve of this sin?” Yudhisthira answers, 
“ I am not pleased with it, but (because we were so badly 
treated by this man therefore) I overlook it. Let Pandu’s 
son take his pleasure whether he does right or wrong ” (38). 
And when Yudhisthira had said this, Krishna answered “as 
you will,” and Yudhisthira then “expressed gratification at 
what Bhima had done in the fight.” In the next chapter, 
Krishna is openly charged with violating all rules of honor 
and noble conduct (61, 38) ; to which the god at first replies 
by specious reasoning (tit for tat), and then, throwing off all 
disguise, says : “ This man could not be killed by righteous 
means, nor could your other enemies have been slain, if I had 
not acted thus sinfully,” yadi nai ’ vamvidham jatu kuryam 
jihmam aharn rane (64). 

Here there is something more than dramatic incongruities 
to notice. For is it conceivable that any priests, setting out to 
write a moral tale which should inculcate virtue, would first 
make one of the heroes do an ignoble thing, and then have 
both their great god and their chief human exponent of mo- 
rality combine in applauding what was openly acknowledged 
even by the gods to be dishonorable conduct ? Even if the act 
was dramatically permitted for the purpose of setting its con- 
demnation in a stronger light and thus purging in the end, 
can we imagine that the only vindicator of virtue should be 
Rama, and that Krishna and Yudhisthira of all others should 
cut so contemptible a figure? On the other hand, is not the 
whole scene explicable without any far-fetched hypothesis, if 
we assume that we have here the mingling of older incident, 
inseparable from the heroic narrative, and the later teaching 
administered by a moral deus ex machina? As the scene 
stands it is grotesque. Krishna’s sudden attack on Y udhist- 
hira is entirely uncalled-for ; and the Latter, who has first de- 
nounced the deed, then joins with the former in approving the 
very thing of which Krishna himself half way through the 
scene disapproves. 

But to those who think that the epic was built on a moral 
didactic plan this is only one of many cases where a satisfactory 



explanation in accordance with the theory will prove difficult. 
They must explain why polyandry, in wliich the heroes in- 
dulge, while it is condemned, is permitted. 1 Ludwig explains 
this “ sharing of the jewel ” (i, 195, 25) as a “ Mythisches 
Element ; ” others hark back to the old-fashioned allegorical 
treatment. But why is allegory with a bad moral seriously 
defended if the heroes are merely to be represented as models ? 
On the other hand, it is known that polyandry was no un- 
common thing on the borders of Brahmanic civilization, and 
Biihler recognized the custom within its pale; while the 
Pandus have no Brahmanic standing, and are evidently a new 
people from without the pale. 2 As a simple historic element 
it is perfectly natural, explained otherwise it remains an in- 
explicable mystery. So too with all the violations of the 
ethical code which are enumerated in the chapter referred to 
above. As characters in an historical epic, the heroes’ acts are 
easily understood; as priestly models, dummies for sermons, 
their doings are beyond explanation. 

Apart from the ignoble conduct of heroes, there are other 
items. Getting drunk at a picnic, for instance, is not proper 
conduct for an exemplary Hindu lady. But in the later epic 
the most virtuous ladies get so drunk that they cannot walk 
straight, madaskhalitagaminyah, i, 222, 21, madotkate, 23. 
Such shocking behavior belongs to the revelry of the Harivahga 
and the probably contemporaneous tale here jovially recorded. 
It is not a moral episode of the fifth century B. c. Elsewhere 
ladies are supposed to be “ unseen by the sun and wind,” not 
only before they are married, but afterwards. 3 Drinking sura 

1 i, 158, 36; 195,27,28. 

2 This follows from the sharp contrast presented by the Kurus and Pandus 
in Brahmanic literature. While the Kurus are a famous folk in ancient 
records, the Pandus are there utterly unknown. 

3 ii, 69, 4 ff. ; iii, 62, 21. The formal phrase here is noticeable. Draupadi 
says : yam na vayur na ca ’dityo drstavantau pura grhe, sd ’ham adtja sabha- 
madhye dr./yumi janasamsadi (she was one of the ladies who got drunl; at 
the outdoor picnic). So DamayantI, of whom Nala says: yam na vayur na 
ca ’dityah pura pa9yati me priyam, se ’yam adya sabhamadhye jete bhuvar 


is especially forbidden by the codes, but it is drunk without 
compunction by the heroes. 1 

The subject of meat-eating is not a trivial one to the Hindu. 
I need not cite the numerous passages describing the slaugh- 
ter and eating of animals by the epic heroes, more especially as 
I have elsewhere illustrated the fact very fully. 2 What I wish 
to point out particularly at the present time is the impossi- 
bility of supposing that the same plan of moral teaching is 
carried out not only in the tales of meat-eating, but in the 
orthodox teaching that meat may be eaten at a sacrifice, and 
in the strict vegetarian diet even at sacrifices, which is in- 
sisted upon in the ahinsa doctrine of the later epic. 3 * * * * 8 Here, 
not only is the substitution of a deer for a horse a new 
feature in the Agvamedha sacrifice, xii, 343, 52 ; but a king 
is held up as a model because there was no killing of animals 
at an agvamedha. For this model king was ahinsrah gucir 
aksudrah, that is “ he did no harm to any living thing, he was 
pure and not cruel ” (aksudra = akrura), xii, 337, 10. The 
parts of the sacrifice were all wood-growth, for there is a 
vaidikl grutih which says bljilir yajuesu yastavyam; ajasam- 

1 The codes are early Sutras as well as gastras, e. g., Gaut. xxi, 1-7. In iv, 
72, 28, at a wedding, suramaireyapanani and meat of all kinds, mrgas and 
medhyah pajavali. Karna’s asurarratam (surarahitam, It.) indicates his 
habitual use of sura, iii, 257, 17. Both Krishna and Arjuna are drunk when 
they receive an ambassador, v, 59, 5. 

2 Ruling Caste, p. 119. Further illustrations also are here given of the 

other vices mentioned. My position in regard to these points I find it neces- 
sary to restate, owing to the misrepresentation of them in the so-called 

Genesis des Mahabharata. The author simply parodies when, on p. 55, he 

says, “these passages cannot belong to a time” (etc.). In the presentation 
thus caricatured I separated no parts of the epic ; hut simply pointed out 

that the statements of the moral code are not in harmony with the action of 

the heroes. 

8 To this, perhaps, is due the intrusion into epic sacrifices (among n cva- 
medha, rajasuya, and other ancient rites) of the so-called pundarika sacrifice, 
or sacrifice of lotus(-roots), which is frequently mentioned, but appears to 
be unknown before the epic. The graciousness of the Vishnu cult is illus- 
trated by its insistence on vegetal and not animal offerings. The orthodox 
Brahman (also the Qakta) demands blood-sacrifices; Krishna prohibits tliem. 
The difference, still marked, appears in the epic and no “ synthesis ” can 
explain it otherwise. 



jnani bijani cchagan no hantum arhatba (you must not kill 
goats at a sacrifice; sacrifice with vegetables and call them 
goats) nai ’sa dharmah satam deva yatra vadhyeta vai paguh 
(it is not the rule among good men to kill animals), xii, 338, 
4. Now this whole teaching is opposed not only to the for- 
mal codes and to the practice of the epic heroes, but also 
to the formal teaching of the epic itself, which says ex- 
pressly : “No man does wrong in eating food prepared with 
the sacrificial verses,” yajusa sariiskrtam maiisam upabhunjan 
na dusyati, xiii, 163, 43. 1 Animal sacrifices are inveighed 
against in one part of the epic and praised in another (iii, 30, 
etc.)- Even human sacrifices are not only mentioned but also 
enjoined on the model heroes : “ Sacrifices are the chief means 
of success. Do thou therefore institute a Rajasuya, a horse- 
sacrifice, an all-sacrifice and a human sacrifice,” xiv, 3, 6-8. 2 

As to hunting, all epic heroes hunt and eat the meat of 
their victims ; but since tins practice is opposed to the ahinsa 
doctrine the casuist has a good deal of difficulty in reconcil- 
ing the practice of the model heroes with that doctrine. It 
is said to be permissible, because sacrificial animals may be 
eaten, and deer are brought under this head by a reference 
to Agastya who “ sanctified them.” But while Rama is quite 
content to say that hunting even with traps is permissible, 
because the saints of royal blood practised it of old; the 
teacher in the Mahabharata is still uneasy, even after con- 
tending that the quarry is “ sacrificial ; ” so he says that really 
the hunter is contending for Ins fife and it is a matter of 
fighting, which takes it out of the category of “ injury,” since 
the hunter himself is as likely to be killed as to kill. 3 All 

1 The chine is excepted, prsthamansam, 43. This and vrtliamansam is 
the same as putramahsam, that is, it is as had to eat meat not used for 
sacrificial purposes as it is to be a cannibal, for amrtam brahmana gava ity 
etat trayam ekatah, cows are as holy as Brahmans, 42. Compare also xiii, 
115 and 116 (below). 

2 Compare xiii, 103, 32 ££., “ arkayanas, turayanas, human sacrifices ” (and 

’ The passages of the two epics are related. Compare : ato rajarsayah 
serve mrgayam yanti, Bharata, with yanti rajarsayaj ca ’tra mrgayam dhar- 
makovidah, xiii, 116, 18, and R. iv, 18, 40, respectively. The law is laid down 


of this is good sense, but it does not save the teacher from 
the weakness of advancing two excuses, and thus betraying 
the fact that the whole ahinsa received from Buddhism and 
half accepted, is a late modification of the practice of the 
model heroes, who disregard the real ahinsa. Yudhisthira 
says frankly that he likes meat, and Bhisma agrees that it 
is a most pleasant and strengthening food ; but he says that 
those who indulge in it go to hell, and then explains that 
warriors may practise hunting for the reasons given above. 
It is no wonder that the model meat-eating hero says “my 
mind is befuddled on tliis point.” Formal Brahmanic law 
accounts hunting one of the four worst vices a king may 

Such contradictions are not those of a “ great-hearted poet ” 
who scorns the narrowness of accuracy. Of this latter class 
of contradictions the poem is full. The Hindu Homer nods 
continually. He forgets that his puppet is addressing Bhisma 
and makes him use the customary vocative, Yudhisthira, be- 
cause the latter is his ordinary dummy, iii, 82, 64; 85, 111. 
He says that even a wise man wiio sells soma goes to hell, 
and that the sale of soma by one who is wise is no fault, xiii, 
101, 12 ff., xii, 34, 31. His gods have no shadows in a well- 
known passage of Nala, but elsewhere “ the gods’ vast shad- 

in JIanu, vii, 50. The whole of xiii, 115 and 11G is an awkward attempt to 
unite hunting-morality with non-injury, na ca doso 'tra vidyate (Rama), 
bhunjan na dusyati (M.). Rama goes so far as to say that to kill a monkey 
is no crime, for the reasons given above, a peculiarly unbrahmanic argu- 
ment. Due to the influence of Buddhism sporadically represented is also 
the passage so similar to the Dhammapada (Dh. P. 385, tam aham brumi 
brahmanam, and 393, yamhi saccaii ca dhammo ca, so sukhi so ca brahmano) 
in iii, 216, 14-15, yas tu ijudro dame satye dharme ca satatotthitah tam 
brahmanam aham manye vrttena hi bhaved dvijah, and the parallel passage 
in xiii, 143, 4G ff., which declares that a tundra not only may become a sarns- 
krto dvijah hereafter, but that he should be revered, sevyah, like a regen- 
erate person, if he is “ pure of heart and of subdued senses,” since “ not birth, 
nor sacrament, nor learning, nor stock" (santatih) make one regenerate, but 
only conduct” is the cause of regeneracv (dvijatvasya vrttam eva tu kara- 
nam). "We have from Brahmanas and Sutras a pretty clear idea of what 
Brahmanism taught in regard to the Qudra. But it never taught this even 
in the Upanishads. It is pure Buddhism, taught as Brahmanism. 



ows ” are seen, ix, 37, 9. His saints are stars, but again only 
“like stars,” and finally “not stars,” iii, 25, 14; 261, 13; 

xii, 245, 22 ; 271, 25, etc. I lay as little weight on such 
contradictions as would any one familiar with the history of 
literature, and it is a mere travesty to say that to this class 
belong such fundamental differences as those which are char- 
acteristic of the precept and practice of the epic. No poem 
composed to teach certain doctrines would admit as its most 
virtuous characters those who disregarded these doctrines 

Whether the fact that only the pseudo-epic puts the Ath- 
arva-Yeda first in the list of Y edas be worthy of consideration 
or not, it has an interesting parallel in the fact that only the 
pseudo-epic places the Atliarvan priest before the others. 
In early works the Acarya, who taught gratis all the Vedas, 
is declared to be worth ten Upadhyayas, Vas. xiii, 48 ; iii, 
21-22 ; Maim, ii, 140-145. This Upadhyaya is the direct 
etymological ancestor of the modem ojha, vizard. In ancient 
times he was a sub-teacher, who taught for a livelihood one 
part of the Veda and Vediinga, and he is identified in the 
epic with the Purohita, who, as Professor Weber has shown, 
is essentially an Atharva-Yeda priest, 1 or magic-monger, whom 
seers regard as contemptible. 2 The pseudo-epic inverts the 
ancient ratio and makes the Upadhyaya worth ten Acaryas, 

xiii, 105, 14-45. 

1 One example of magic recorded in the epic is particularly interesting, 
as it is referred to the Kaulika-^astra, or left-hand cult, and is a parallel to 
the practice recorded in Theocritus’ second idyll. It is called chayopasevana 
or shadow-cult, and consists in making an image of an enemy and sticking 
pins into it to cause his death, iii, 32, 4. 

“ The Jatakas, too, regard the Purohita as a mere magic-monger, though 
they call him also acariya, Fick, Sociale Gliederung, p. 110. On the Purohita 
Upadhyaya, see the story of Marutta, xiv, 6, 7 ff. Here (and in xiii, 10, 30) 
the office is hereditary. The king in the former passage insists that his 
family Purohita shall serve him with an incantation, hut the priest tells 
him he is engaged elsewhere, and slfys “ Go and choose some one else as 
your Upadhyaya.” So in i, 3, 11 ff., where a proper Purohita is sought “ to 
kill bad magic” and is installed as Upadhyaya. On his practical importance 
and honors, compare i, 183, 1, 9 ; 6-7 ; v, 126, 2 ; 127, 26 ; ix, 41, 12. On the 
Contempt with which he is regarded, xiii, 10, 36; 94, 33; 135, 11. 


The epic in its present form is swollen with many additions, 
but they are all cast into the shade by the enormous mass 
added bodily to the epic as didactic books, containing more 
than twenty thousand stanzas. I have elsewhere fully ex- 
plained 1 the machinery by which this great appendix was 
added to the original work through suspending the death of 
the narrator, and shown that there are many indications left in 
the epic pointing to the fact that the narrator in the original 
version was actually killed before he uttered a word of the 
appendix. As this one fact disposes of the chief feature of 
that theory of the epic which holds that the work was origi- 
nally what it is to-day, and as no sufficient answer has been 
given to the facts adduced, there can be no further question 
in regard to the correctness of the term pseudo-epic as applied 
to these parts of the present poem . 2 There has been, so far 
as I know, no voice heard in favor of the so-called synthetic 
theory in regard to the nature of these late books, except 
certain utterances based apparently on a misconception. Thus 
it has been said, I think, by Professor Oldenburg, that the dis- 
covery of the lotus-stalk tale among the early Buddhistic 
legends tends to show that the epic book where it occurs is 
antique . 3 On this point this is to be said: No one has ever 
denied that there are early legends found in the late parts of 
the epic ; but the fact that this or that legend repeated in the 
pseudo-epic is found in other literature, no matter how old, 
does nothing toward proving either the antiquity of the book 
as a whole, which is just what the “ synthetic ” method con- 
tends for, or the antiquity of the epic form of the legend. 
The story of the Deluge, for example, is older than any 
Buddhistic monument ; but this does not prove that the epic 
version in the third book is old. The same is true of the 

1 Am. Journ. Phil., xix, p. 7 ff. 

2 In this view I am glad to see that Professor Jacobi, in the review cited 
above, fully agrees. So also M. Barth, Journal des Savants, 1897, p. 448. 

3 I am not sure that I have here cited the well-known Russian savant 
correctly, as I have seen only a notice of his paper ; but I believe the essential 
point is as given above. The Lotus-Theft, however, perhaps the same story, 
is alluded to as early as Ait. Br. v. 30. 



first book, where the pauranl katha of Khandava, for ex- 
ample, is a justifiable and instructive title, set as it is in a 
late book. A special “ ancient tale ” is just what it is ; in- 
truded awkwardly into the continuous later narration, 223, 
14-16, but still bearing traces of its heterogeneous character, 
as I have shown elsewhere. 1 Knowing, as Ave do, the loose 
and careless way in which epic texts have been handed down 
(compare the way in which appear the same passages given 
in different editions of the same epic or in both epics), and 
the freedom with which additions were made to the text, 
we are in such cases historically justified in saying only that 
certain matter of the epic stands parallel to certain Bhasya 
matter or Buddhistic matter. A tale is found in the epic. 
Its content is pictured on a stone or found in different form 
in a Jataka. What possible guarantee have we that the epic 
form of the tale is as old as the Jataka, still less that it is as 
old as the stone, least of all that the book in which the epic 
tale appears must as a whole be antique ? Only paucity of 
solid data could make eminent scholars build structures on 
such a morass. 

Having already given an example or two of late feat- 
ures in the pseudo-epic, I would now point to some of the 
characteristic marks of the later poem in other regards. Mid- 
way in the development of the epic stands the intrusion of 
the fourth book, where to fill out an extra year, not recog- 
nized in the early epic, the heroes live at court in various dis- 
guises. Here the worship of Durga is prominent, who is 
known by her Puranic title, mahisasuranaginl, iv, 6, 15, whose 
“ grace gives victory,” ib. 30 (though after the intrusion of 
the hymn nothing further is heard of her). The Durga here 
depicted bears a khetaka (as she does when the same hymn is 
repeated in vi, 23, 7), iv, 6, 4. This word for shield amid in- 
numerable passages describing arms, is unknown in the epic 
except in connection with Durga, but it is found in post- 
epical literature. It stands in the same historical position as 
does the epithet just mentioned. In these cases we have 
1 Bharata and Great Bharata, p. 15. 


general evidence of the lateness of the book as well as of the 
hymn to Durga. Matter and metre go hand in hand. 

A very striking example is given further in the show of 
arms which are described in this book. Although Arjuna is 
still a young man, yet, when the exhibitor comes to show his 
bow, Gandiva, he says “ And this is the world-renowned bow 
of the son of Prtha, which he carried for five-and-sixty years ” 

iv, 43, 1-6. Nothing could be plainer than this passage. The 
exhibition of arms was composed when the later poet had in 
mind the actual number of years the hero carried the bow 
according to the epic story. He forgot that he was composing 
a scene which was to fit into the hero's young manhood and 
not into the end of his life. In iv, 71, 15 Arjuna is recog- 
nized as still a “ dark-featured youth,” 1 and some time after 
this scene it is expressly stated that it was even then only 
thirty-three years since the time when Arjuna got the bow, 

v, 52, 10 (referring to the Khandava episode, i, 225). 2 

While it is obvious to one who is willing to examine the 

1 Here there is another inconsistency. In iv, 44, 20, instead of being a 
fyamo yuva as in 71, 15, he is called Arjuna because of his white steeds 
and complexion, “which is rare on earth,” where the “white” complexion 
matches steeds and deeds, “pure (white).” In v, 59, 10, Arjuna is also dark. 

1 According to v, 82, 40, and 90, 47 and 70, respectively, the time from the 
exile to the battle is thirteen years past (“this is the fourteenth"). Ignor- 
ing the discrepancy between twelve and thirteen years of exile, we must 
allow at least twenty-nine years for Arjuna to live before the Khandava 
incident, which, added to thirty-three, makes sixty-four, which would be 
Arjuna’s age when “a youth,” before the war begins! If, however, we over- 
look the statement of v, 52, 10, and add the years of exile to twenty-nine, 
we still get forty-odd years as his life-limit when he has carried the bow 
sixty-five years. It must be remembered that Arjuna was twenty-four years 
in exile, twelve years before the dicing and twelve or thirteen after it, and 
that Abhimanyu was sixteen when the war broke out (forty-four years for 
Arjuna if he won Draupadi when he was sixteen, and he could not have 
been twenty years older at that time). The synthesist may say “ How nar- 
row ! Poets do not regard such discrepancies,” but even poets are generally 
aware that a hero less than fifty cannot have carried a bow for more than 
sixty years, especially when he got it at the age of forty or thereabouts ! 
Krishna dies in the thirty-sixth year after the war (xi, 25, 44), which should 
make Arjuna about thirty at the beginning of tire war. This throws a side- 
light on the intrusion of the twelve-years exile as a brahmacarin, spoken of 



epic with careful analysis that the Gita and the thirteenth book, 
for example, are purely priestly products, and that one of them 
is on the whole as early as the other on the whole is late , 1 
it is not easy to decide what is the relation between these 
great groups of verses and the heroic epic, with which neither 
has any inner connection. Nevertheless, although there can 
be as a result of the inquiry only the liistorical probability 
usual in answering the problems of ancient literature, and not 
such a mathematical quod erat demonstrandum as the synthe- 
sist demands, we are not wholly at a loss to reply to this 
question. In the first place we have a very instructive anal- 
ogy in the intrusion into both epic texts of an incongruous 
didactic chapter found both in the Ramayana and the Malui- 
bharata, which bears on its face evidence of its gradual expan- 
sion. But even without this evidence it will, I think, be 
clear even to the synthesist that the same chapter cannot 
have arisen independently in both epics; so that in tins in- 
stance we have a plain case of the dynamic intrusion into 
an epic text of foreign didactic material . 2 

Again, the presence of a huge volume of extraneous addi- 
tions, containing both legends and didactic stuff, now tagged 
on to the epic as its nineteenth book and recognized in the 
last part of the epic itself, is an object-lesson in dynamic 
expansion which in itself shows how the pseudo-epic may 
with perfect regard to historic probability be supposed to have 
been added to the epic proper. The Ramayana too is instruc- 
tive, as it shows that whole chapters have been interpolated, 
as admitted by its commentator. The great epic itself admits 
that there is a difference between the main epic and the epi- 
sodes, in saying that the former is only one-fourth of the 
whole, and relegating seventy-six of its hundred thousand 
stanzas to the domain of the episodic epic . 3 

1 Compare the chapter on metres. 

2 This chapter is the Kaccit section ii, 5 and R. ii, 100, previously referred 
to, discussed in detail in AJP. xix, 147 ff. 

3 As an interesting example of the growth of Sanskrit popular poems, 
Mr. Grierson informs me that there is extant a vrddha or brhad Vishnu 
Purana, which contains large additions to the received text. 


That the priests developed the epic for their own interests, 
goes without saying; hence the long chapters of priestly 
origin on the duty of charity — to priests. That they added 
legends has already been shown, and the metre still attests 
the approximate age of a Nala or a Sulabha episode. But 
besides didactic and legendary masses, it was necessary, in 
order to popularize the poem, to keep some sort of proportion 
between the tale and its tumors. Hence the fighting episodes 
were increased, enlarged, rewritten, and inserted doubly, the 
same scene and description occurring in two different places. 
For this reason, while there is an appreciable difference in the 
metre of the different episodes which were inserted whole, the 
fighting scenes are chiefly of one §loka-type, — a type later 
than that of some of the episodes, but on a par with that of 
the later didactic and narrative insertions. 

Whether the original tale was occupied with the Pandus or 
not, the oldest heroes are not of this family, and the old Vedic 
tradition, while it recognizes Bharatas and Kurus, knows 
nothing about Pandus. The Kuru form of epic may perhaps 
be preserved in the verse (restored) of one of the oldest 
Upanishads, Chand. Up. iv, 17, 9: 

yato yata avartate tad tad gacchati manavah 

Kurun aqva ’bhiraksati, 

a gatha restored by omitting an evident interpolation. 1 The 
style is like the usual epic turn, e. g., Ii. vi, 106, 22, 

yena yena ratho yati tena tena pradhavati. 

Nevertheless, a Pandu epic of some sort existed as early 
as the third century b. c., as is shown by the testimony of 
Panini and the Jatakas (which may indeed give testimony for 
an era even later than the third century), though in the latter 
literature the epic story is not presented as it is in our epic. 
This takes us from the form to the date of the Mahabharata. 

1 Compare Miiller, SBE. i, p. 71. See also the Sutra verse on the Kurus’ 
defeat, cited by Professor Ludwig, Abh. Bohm. Ges. 1884, p. 5. 




First, to define the epic. If we mean by this word the 
beginnings of epic stoiy, as they may be imagined in the 
“circling narration,” in the original Bharat! Katha, or in 
the early mention of tales of heroes who are also epic char- 
acters, the time of this epic poetry may lie as far back as 
700 b. c. or 1700 b. c., for aught we know. There are no 
further data to go upon than the facts that a Bharata is men- 
tioned in the later Sutra, that the later part of the Qatapatha 
Brahmana mentions the “ circling narration,” and that akhy- 
ana, stories, some in regard to epic personages, told in prose 
and verse, go back to the early Vedic period. 1 We must be 
content with Weber’s conservative summary: “The Malia- 
bharata-saga (not the epic) in its fundamental parts extends 
to the Brahmana period.” 2 

If, on the other hand, we mean the epic as we now have it, 
a truly synthetical view must determine the date, and we shall 
fix the time of the present Mahabharata as one when the 
sixty-four kalas were known, when continuous iambic padas 
were written, when the latest systems of philosophy were 
recognized, when the trimurti was acknowledged, when there 
were one hundred and one Yajur Veda schools, when the 
sun was called Mihira, when Greek words had become familiar, 

1 On the early prose-poetic akbyana of the Vedic and Brahmanic age, com- 
pare the essays by von Bradke, Journal of the German Oriental Society, 
xxxvi, p. 474 ff. ; and Oldenherg, ib. X X X V i i , p. 54 ff., and xsxix, p. 52 ff. Ballad 
recitations, akkhana, are mentioned in early Buddhistic works, which we may 
doubtfully assign, as Professor Rhys Davids does undoubtingly, to the fifth 
century b. c. 

2 Episches im Vedischen Ritual, p. 8 : Die Mbharata-Sage reicht somit ihrer 
Grundlage nach in die Brahmana Periode hinein. 



and the Greeks were known as wise men, when the eighteen 
islands and eighteen Puranas were known, when was known 
the whole literature down to grammars, commentaries, Dharma- 
gastras, granthas, pustakas, written Yedas, and complete MSS. 
of the Mahabharata including the Harivanga. But this is a 
little too much, and even the inconsistent synthesist, who 
draws on a large vituperative thesaurus whenever another 
hints at intrusions into the epic, may well be pardoned for 
momentarily ceasing to be synthetic and exclaiming with 
reason Da liegt doch die Interpolation vor Augen ! 1 

That the complete Mahabharata, for the most part as we 
have it to-day, cannot be later than the fourth or fifth century 
of our era, follows from the fact, brought out first by Pro- 
fessor Bhandarkar and then by Professor Biihler, that it is 
referred to as a Smrti in inscriptions dated not much later 
than this, while by the fifth century at least it was about as 
long as it is now. 2 But we may go further back and say with 
comparative certainty that, with the exception of the parts 
latest added, the introduction to the first book and the last 
book, even the pseudo-epic was completed as early as 200 A. D. 
For the Roman denarius is known to the llarivaiica and the 
Harivanga is known to the first part of the first book and to 
the last book (implied also in the twelfth book) ; hence such 
parts of these books as recognize the Harivanga must be 
later than the introduction of Roman coins into the country 
(100-200 A. D.); but though coins are mentioned over and 
over, 3 nowhere, even in the twelfth and thirteenth books, is 
the denarius alluded to. 

1 Genesis des Mahabharata, p. 129. 

2 Quite important, on the other hand, is the fact recently emphasized by 
Dr. Cartellieri, WZ. xiii, p. 69, 1899 : “ Fiir Subandhu und Bana war das Maha- 
bharata . . . kein dharmagastra, sondern ein Ivavya,” which the poem itself 
proclaims itself to be, i, 1, 61. 

3 The money recognized is gold and silver “ made and unmade ” and niskas, 
though chests of precious metal are mentioned and a great deal of money is 
found when excavating for treasure (perhaps near Taxila). When the realm 
is prosperous the soldier’s pay is “not copper.” For references to money, 
coins, etc., see ii, 61, 2, 8, 20-30 ; iii, 15, 22 ; 255, 17 ; iv, 18, 18 ; 22, 10 ; 38, 43 ; 
xii, 328, 46 (threefold test of gold) ; xiv, 65, 20 (amount of treasure). On the 



Another interesting item is contributed by the further 
negative evidence afforded in the matter of copper-plate 
grants. Gifts to priests are especially urged in the Anugasana, 
and the gift of land above all is praised in the most extrava- 
gant terms. We know that by the second century of our era, 
and perhaps earlier, such gifts to priests were safeguarded by 
copper-plate grants, bearing the technical name of patta (pata) 
or tamrapatta,, and elaborate instructions for their making are 
given in the law-book of Narada and Vishnu, while they are 
mentioned in the code of Yajnavalkya, but not before; for 
Manu, though he mentions the boundary-line being “ re- 
corded,” nibaddha, has no suggestion of plate-grants. The 
epic, however, at least the pseudo-epic, speaks of writing 
down even the Vedas, and recognizes rock-inscriptions, but 
in the matter of recorded grants to priests says nothing at all ; 
much less does it recognize such a thing as a tamrapatta- 
The only terms used are parigraha and agrahara, but the 
latter, which is very rare, is never used in the sense of a land- 
grant, though gramagrahara occurs once in the later epic, xv, 
14, 14. Even the general gasana is never so employed. 1 It 
is true that this negative evidence does not prove the epic to 
have been completed before the tamrapatta was known ; but 
on the other hand, it is unlikely, were the tamrapatta the 
usual means of clinching a bhumidana when the Anugasana 
was composed, that this mode would have passed unnoticed, 

conquest of Taksagila, see i, 3, 20. According to ii, 61, 20, the soldier's pay is 
“ a thousand a month,” here presumably copper. 

1 Legal documents appear first in Vas. I)h. S., xvi, 10, 15, under the name 
lekhita. Probably the first deeds were written on cloth or boards, phalaka, as 
a board-copy precedes the rock-inscription, ASW1., iv, p. 102. The epic 
has pieture-pata, as in xv, 32, 20, dad rye citram patagatam yatha (aycarya- 
bhutam) and often. Hock-inscriptions are mentioned only in xiii, 139, 43, 
ciram tisthati medinyam yaile lekhyam iva ’rpitam. Written Vedas are 
alluded to only ib. 23, 72. Seals are used as passports, iii, 15, 19. Compare 
also ii, 55, 10, na lekhyam na ca matrka; v, 148, 23, citrakara iva 'lekhyam 
krtva ; ib. 189, 1, “ lekliya and other arts ; ” vii, 99, 7, namankitah (compare 
above, p. 205), of arrows. The conjunct ganaka lekhakah occurs only in xv, 
14, 8, and in the verse of the Kaccit section, ii, 5, 72, which is a subsequent 
addition even to this late chapter ; AJP., xix, p. 149. 



and we may conclude that the gift-sections of this book were 
at least as old as the oldest copper-plate grants to priests. 1 

The time of the whole Mahabharata generally speaking 
may then be from 200-400 A. d. This, however, takes into 
account neither subsequent additions, such as we know to 
have been made in later times, nor the various recastings in 
verbal form, which may safely be assumed to have occurred 
at the hands of successive copyists. 

For the terminus a quo, the external 2 evidence in regard to 
the Pandu epic, Mahabharata, though scanty, is valuable. It 
shows us first that the Mahabharata is not recognized in any 
Sanskrit literary work till after the end of the Brahmana 
period, and only in the latest Sutras, where it is an evident 
intrusion into the text. For the Grhya Sutras belong to the 
close of the Sutra period, and here the words Bharata and 
Mahabharata occur in a list of authors and works as substi- 
tutes for the earlier mention of Itihasa and Purana in the same 

1 The verse xii, 56, 62, which the author of Das Mahabharata als Epos und 
Rechtsbuch, p. 187, adduces to prove that written deeds were known, is given 
by him without the context. When this is examined it is found that the verse 
refers not to land but to a king’s realm. Neither does the text nor the com- 
mentator necessarily (as asserted, loc. cit.) make it refer to land-grants. The 
word used is visaya, a king’s realm or country (as in xiv, 32, 8) and the poet 
says that ministers who are given too much liberty “ rend the king’s realm by 
counterfeits ” (or falsifications). The situation and the analogy of 59, 49, and 
69, 22, and 100, 6, where general deceit and dissension are the means employed 
to destroy a realm, make it most probable that the word pratirupaka is used 
here to distinguish the forged laws and edicts of the usurping ministers from 
the true laws which the helpless king would enact. Such suppression of the 
king and substitution of false edicts arc thoroughly Oriental, and may easily 
be illustrated by the use of this very word, pratirupaka, in the Lotus of True 
Law, where pratirupaka means just such “ false laws ” substituted for the 
real king's true laws (iii, 22 ; SBE., xxi, p. 68, note, with Iranian parallel). 
The commentator says “ corrupt the country by false edict-documents,” that 
is, he gives a general application to the words, which may be interpreted as 
referring to land-grants, but this is not necessary. Possible would be the 
later law-meaning of frauds of any kind, perhaps counterfeit money. Certain 
it is that the passage is not “ a direct proof for forged documents,” still less 
for “false documents by means of which any one gets land.” 

2 Cis-indic evidence is negative and without weight. Megasthenes, c. 300 
b. c., has left no fragment on Hindu epics, and the source of Dio Chrvsos- 
tomos (100 A. d.), who mentions a Hindu Homer, is unknown. 



place, so recent a substitution in fact that some even of the 
latest of these Sutras still retain Itihasa and Purana. But 
when the words do actually occur they are plainly additions 
to the earlier list. Thus in (yahkhayana iv, 10, 13, the list 
is Sumantu, Jaimini, Vaigampayana, Paila, the Sutras, the 
Bhasya, Gargya, etc., noth no mention of the epic. But the 
Agvalayana text, iii, 4, 4, inserts the epic thus: Sumantu, 
Jaimini, Vaigampayana, Paila, the Sutras, the Bhasya, the 
Bharata, the Mahabharata , dharmdcari/as, Jananti , Btihavi , 
Gargya, etc. The next step is taken by the yambavya text, 
which does not notice the Bharata and recognizes only the 
Mahabharata (whereas some texts make even the Agvalayana 
Sutra omit Mahabharata altogether, reading Bliarata-dhar- 
macaryah). When it is remembered tliat these and other lists 
of literature are not uncommon in the Sutras, and that nowhere 
do we find any other reference to the Mahabharata, it becomes 
evident that we have important negative testimony for the 
lateness of the epic in such omission, which is strengthened 
by the evidently interpolated mention of the poem, withal in 
one of the latest Sutras. 1 

Patanjali, it may be admitted, recognizes a Pandu epic in 
the verse, asidvitlyo ’nusasara Pandavam, and in his account 
of the dramatic representation of the sacred legend, indis- 
solubly connected with the tale. 2 This takes us at farthest 
back to the second century ; but this date (p. 56) is doubtful. 

Panini knows the names of the epic heroes, and recognizes 
the Arjuna-Krislma cult in giving a derivative meaning 
“ worshipper of Arjuna ” (Krishna). He also, which is more 
important, recognizes the name Mahabharata. It cannot rea- 
sonably be claimed, I think, that this name does not refer to 
the epic. It stands, indeed, beside maha-Jabala, and might (as 
masculine) be supposed from this circumstance to mean “ the 

1 That these lists, anyway, are not of cogent historical value, has lately 
been emphasized by Dr. Wintemitz in his last review of Dahlmann. They 
certainly cannot help in dating the epic before the fourth century. The 
intrusion of the genus itihasa-purana into such lists is illustrated even in the 
Upanishads. Compare Mund. Up. i, 5, with the note at SBE., xv, p. 27. 

2 Compare Weber, IS., i, pp. 147-149 ; xiii, pp. 356-357. 



great descendant of Bharata,” yet not only do other words in 
the list show that this is not necessary, but further, there is 
no instance, either in the epic itself or in outside literature, 
where Mahabharata means a man, or where it does not mean 
the epic. In this particular, therefore, as it gives me pleasure 
to state, I believe that the Rev. Mr. Dahlmann is right, and 
that Panini knew an epic called the Mahabharata. That he 
knew it as a Pandu epic may reasonably be inferred from his 
mentioning, e. g., Yudhisthira, the chief hero of the epic. 1 

But no evidence has yet been brought forward to show con- 
clusively that Panini lived before the third century b. c. 

Again, it is one thing to say that Panini knew a Pandu 
Mahabharata, but quite another to say that his epic was our 
present epic. The Pandu epic as we have it represents a 
period subsequent not only to Buddhism 500 B. c., but to the 
Greek invasion 300 b. c. Buddliistic supremacy already de- 
cadent is implied by the passages (no synthesist may logically 
disregard them) which allude contemptuously to the edukas 
or Buddhistic monuments as having ousted the temples of 
the gods. Thus in iii, 190, 65, “ They will revere edukas, 
they will neglect the gods ; ” ib. 67, “ the earth shall be 
piled with edukas, 2 not adorned with god-houses.” With 
such expressions may be compared the thoroughly Buddliis- 
tic epithet, eaturmaharajika, in xii, 339, 40, and Buddliistic 
philosophy as expounded in the same boob. .More impor- 
tant than this evidence, however, which from the places 
where it is found may all belong to the recasting of the 
epic, is the architecture, 3 which is of stone and metal and 

1 He mentions him not as a Pandu but only as a name, like Gavisthira ; 
to distinguish the name from the expression (e. g. R. vi, 41, 65) yudhi sthirah, 
I presume. 

2 Lassen, loc. cit., p. 490. So, iii, 188, 56, vihara ; 49, pasanda; 67, seven 
suns ; all found in one place (p. 88). See final notes. 

3 Buddhistic buildings with wooden fences and walls of brick and stone 
are alluded to in Cull, vi, 3, 8. In connection with this subject it must be 
remembered that even the late Grhya Sutras in giving directions for house- 
building know only wooden thatched houses. The Greek account states 
that the Hindus used only mud, wood, and brick. This makes it improbable 
that wood architecture had almost disappeared in the third century. 



is attributed in all the more important building operations 
to the demon Asura or Danava Maya, who, by his magic 
power, 1 builds such huge buildings as are described, im- 
mense moated palaces with arches and a roof supported by 
a thousand pillars. There is in India no real architecture 
that goes back of the Buddhistic period, and of both Bud- 
dhistic and Jain architecture the remains are distinctly in- 
fluenced by Greek models. 2 

The Greeks are described as a western people (northwests 
ern, with Kambojas), famous as fighters, wearing especially 
fine metal armor, and their overthrow is alluded to. The 
allies engaged in the epic battles are not only native princes 
but also Greek kings and Persians, who come out of the West 
to the war. In one passage the Greeks are described as 
“ all-knowing,” though I tlfink this to be a late interpolated 
chapter. 3 * * * * 8 But rft^i, iii, 190, 90, surely implies the zodiac. 

But even if the passage mentioning all-knowing Greeks be 
an interpolation, the fact that the “ Greeks,” who must here 
be the real Greeks, bear the name Yavanas, shows that the 

1 So the great walls and palaces of Patna, which are especially mentioned 
in the Mahabhasya, are attributed by tradition to demoniac power ( Fa- 
Hien), and the great architecture of Mathura is also ascribed to superhuman 
power. On Maya’s maya, to which is attributed the most extensive building, 
compare ii, 1 ; v, 100, 1-2; viii, 33, 17 (Asura cities) ; R. iv, 51, 10. It is pos- 
sible that the Benares ghats are referred to in vii, 00, 1 (Ganga) cayanaili 
kancanaig cita. “ Golden” buildings may be only gilded wood (as they are 
to-day). Plated stone is mentioned in ii, 3, 32. Old Patna’s noble “walls and 
palaces ” are now unfortunately under the Ganges, in all probability. 

2 The caitva and stupa mounds (only K. has a caityaprasada, v, 43, 3), like 
the caves, are not to be compared with roofed palaces of stone and marble. 
A statue of iron is mentioned, ayaso Bhlmah, xi, 12, 15 ; iron bells in temples, 

xii, 141, 32. In ii, 4, 21-22, the Greeks are compared to Ivalakeya Asuras. 

Here, along with the king of Kamboja, is mentioned one king, (the) Kam- 

pana, “who was the only man that ever frightened, kamp, the Yavanas, (men) 
strong, heroic, and skilled in weapons. Like as Indra frightened the Kala- 

keya Asuras, so” (K, frightened the Greeks). Compare also Kalayavana 

who had the Garga-glory (p. 15) in xii, 340, 05, Weber, loc. cit. 

8 Compare ii, 14, 14 ; iii, 254, 18 ; xii, 101, 1 ff. ; Ruling Caste, p. 305 ; viii, 
45, 36, sarvajna Yavanah, in the expansion of the preceding vituperative sec- 
tion, where from hanta bhuyo bravlmi te, in 45, 1, Karna bursts out again in 
new virulence, which looks almost too much like a later adornment. 



Yavanas elsewhere mentioned 1 are also Greeks and not some 
other people exclusively. It is a desperate resort to imagine 
that, in all these cases, well-known names refer to other 
peoples, as the synthesist must assume in the case of the 
Greeks, Bactrians, Persians, Huns, and other foreigners men- 
tioned frequently throughout the poem. A further well- 
known indication of Greek influence is given by the fact 
that the Ivsudrakas and Malavas were united into one nation 
for the first time by the invasion of Alexander, 2 and that 
they appear thus united under the combined name ksudra- 
kamalavas in the epic, ii, 52, 15. The Romans, Romakas, 
are mentioned but once, in a formal list of all possible 
peoples, ii, 51, 17 (cannibals, Chinese, Greeks, Persians, 
Scythians, and other barbarians), and stand thus in marked 
contrast to the Greeks and Persians, Pahlavas, who are 
mentioned very often; though in the account of Krishna 
killing the Yavana whose name was Ivaserumat, iii, 12, 32, it 
lias been suggested by Weber that the name was really of 
Latin origin. It is clear from this that, while the Greeks 
were familiar, the Romans were as yet but a name. Further, 
the distinct prophecy that “ Scythians, Greeks, and Bactrians 
will rule unrighteously in the evil age to come ” (kali-age), 
wliich occurs in iii, 188, 35, is too clear a statement to be 
ignored or explained away. When this was written the 
peoples mentioned had already ruled Hindustan. If this 
were the only place where the names occurred, the Markiin- 
deya episode, it might be regarded as part of an interpolation 
in mass. But the people here described as foreign oppres- 
sors are all mentioned repeatedly as barbarians and warriors, 
associated generally, as in the passage just mentioned, with 
other peoples of the West, such as Abhlras and Kambojas. 
Thus in iii, 51, 23, “ Singhalese, Barbaras and barbarians, 3 

1 Yavanas or Yaunas (xii, 207, 42-3), i. e., Ionians. So Jacobi, loc. cit. 

2 Lassen, Ind. Alt. ii, pp. 169-171 ; Weber, Ind. Stud, xiii, p. 375. 

3 That is both the Hindu and native name for Ceylon, and the Greek and 
Hindu name for barbarian ! Sinhalan Barbaran Mlecchan ye ca Lankaniva- 
sinah. The word barbaras (= ol 0dp0apoi) occurs in both epics but not in 
literature of an earlier date. Weber, Ind. Lit., p. 237, note, calls attention 



and the inhabitants of Lanka” are grouped together, in con- 
trast to the “Western realms, those of the Persians, Greeks, 
and Scythians” (with the folk of Kashmeer, Daradas, Kira- 
tas, Huns, Chinese, Tusaras, Indus-dwellers, etc.). So in xii, 
207, 43, opposed to sinners of the South, are the Northern 
sinners, Greeks (Yaunas), Kambojans, Kandahar-people (Gan- 
dharas), Iviratas and Barbaras, who are here said to be wander- 
ing over this earth from the time of the Treta age, having 
customs like those of wild animals or of the lowest castes. 

Such allusions as these can mean only this: the Pandu- 
Epic, in its present form, was composed after the Greek inva- 
sion. 1 I have suggested above that the form of the name 
Bactrian does not compel us to accept Professor Weber’s 
conclusions in regard to the date of passages now containing 
this form. If this seems inconclusive, there is nothing for it 
but to refer the epic in its present form to a post-Christian 
era. But even otherwise, the presence of the Greeks and 
Bactrians as warriors and rulers in India cannot be explained 
out of the poem by a loose reference to the fact that India 
had heard of Yavanas before Alexander. 

This brings us to another point of view. A stanza fol- 
lowing the one last cited proclaims that “even Narada recog- 
nizes Krishna’s supremacy,” an utterance 2 which points clearly 
to a comparatively recent belief in Krishna as All-god, a point 
long recognized. On the basis of the Arjuna cult implied 
by Panini, the synthesist urges that the whole epic, in its 
present Smrti form and with its belief in the all-godhead of 
the Krishna-Arjuna pair, is as old as the fifth century B. C. 
But even if an Arjuna cult were traced back to this date, 

to this constant union of Greek with other Western peoples in other literature 
as well. The name was extended to Indo-Scythians and later even to Persians 
and Arabians. Weber, loc. cit. 

1 As has long ago been suggested, of the Greeks mentioned in the epic among 
the allied forces, Bhagadatta may be Apollodotus the founder of the Graeco- 
Indian kingdom (160 b. c.). Weber, Ind. Lit., p. 204 ff. This Greek is espe- 
cially mentioned not only as “ruler of the Yavanas,” but as the friend of the 
epic hero’s father, that is, as known to an older generation, ii, 14, 15; von 
Schroeder, Lit. und Cultur, p. 463 (with other references). 

2 Narado 'py atha Krsnasya param mene . . . ^aijvatattvam, xii, 207, 48. 



there would still be no evidence in regard to the cult of the 
twain as All-god. And this is the claim of the present epic, 
except where, as in the case just cited, incredulity is involun- 
tarily manifested or plainly stated (as in the reviling scene 
in Sabha). The Gita itself admits that those who worship 
Krishna as the All-god, or recognize him, are few in number : 
vasudevah 1 sarvam iti sa mahatma sudurlabhah, 7, 19; “Me 
(as All-god) in human form, not recognizing my godhead, 
fools despise,” 9, 11. The Mahabliasya does not recognize 
Krishna as All-god, but as hero and demigod. The cult is 
growing even in the epic itself. So, too, no Srnrti 2 can be 
implied by Panini’s words. 3 

I come now to the testimony of Buddhistic literature. As 
said above, the oldest literature knows only ballad tales. It 
may be assumed that the Jatakas are older than Agvaghosa, 
who knows epic tales, but not always in epic form, and does 
not refer to the epic either by name or by implication, his 
general agama being, as I have shown, a term used of any 
traditional literature, sacred or profaned The Jatakas may 

1 Mathura in the whole epic is the birthplace of Vasudeva, who seems to 
herd his cattle there ; while in the Mahabliasya it is bahu-Kurucara Mathura 
and the chief city of the Pancalas, clearly the older view. See ii, 14, 34, 
45 ff. ; xii, 340, 90; i, 221,46 (cows, mathuradefyali) ; IS. xiii, p. 379 ff. ; on 
Krishna as nut Vishnu in the Bhasya, ib., pp. 349, 353. In ii, 14, Krishna (as 
All-god 1) “could not injure his foe even in three hundred years,” 36 and 67. 

2 The state of mind that in the face of the “ evidence ” of Panini can lead 
one to say Panini was acquainted with a Pandu-Muhabharata peculiarly didactic 
(Das Mbh. als Rechtsbuch, p. 155) is inconceivable. The whole “ evidence ” at 
its most evincing is that Panini knew a Maliabharata in which the heroes 
were objects of such worship as is accorded to most Hindu heroes after death. 

3 So the later Ramayana is turning into just such a moral and didactic 
work as the other epic. I have already instanced the intrusion of the Kaccit 
section. So Kama, in vii, 55, 3, sets himself to telling homilies, with a familiar 
sound, kathain paramadharmisthdin vyahartum upacakrame (just as in xv, 29, 
14, katha divya dharmistliay ca ’bhavan, nrpa); and R. ib. 37, 24, kathah 
kathyante dharmasamyuktah puranajnair mahatmabhih. In the same way, 
the late (gradual) identification of Rama with Vishnu stands parallel to the 
change of the demigod Krishna to the All-god Vishnu, for Krishna is never 
mortal — there is no such antithesis — but he nevertheless is often not 
supreme god but only demigod in the epic. 

* So of law-rules in epic language, e. g., fistah yastresv anagatam vyava- 
syanty anu rajanam dharmam, R. iii, 50, 9 (G. has nayajastresu). 



go back to the third or fourth century, or they may not, so 
far as their present form is concerned. At any rate, they show 
no knowledge of the epic as such. What they show (the 
material has been sufficiently collected by the Rev. Mr. Dahl- 
mann) is that the epic characters were familiar and the story 
of the Pandus was known, although the characters do not 
occupy the position they do in the epic . 1 But no date of an 
epic, still less of our epic, can be established on casual refer- 
ences to the heroes of the epic found in literature the date of 
which is entirely uncertain. Perhaps it is negatively quite as 
significant that the Jatakas do not refer to the epic at all, but 
only to people mentioned in it. 

The present epic, if it records anything historical, records 
the growth of a great power in Hindustan, a power that could 
not have arisen before Buddhistic supremacy without leaving 
a trace of the mighty name of Pandu in the early literature. 
There is no such trace. Moreover, even the idea of such a 
power as our epic depicts was unknown before the great 
empire that arose under Buddhism. For this reason it is 
impossible to explain the Pandu realm described in the epic 
as an allegory of the fifth century, for we cannot have an 
allegory in unknown terms. The Pandus, be it remembered, 
rule all India, and the limits of their empire, as geographically 
defined in the epic, far surpass the pre-Aqokan imagination, 
as it is reflected in the literature. Even Manu has no idea of 
an empire. His king is a petty raj . 2 

Before the Mahabharata there were tales of Kurus and 
Bharats known to antiquity. Incongruous as the name 
appears to be, Bharata yet designates the Pandu epic. How 

1 The latter point proves nothing, for even in Sanskrit literature, as I 
pointed out long ago, the heroes of the two epics are mixed up confusedly, 
and we cannot suppose a Buddhist would be more careful than a Brahman 
in verifying references to Brahmanic literature. 

3 “ Great kings ” and “ emperors ” are indeed known even in pre-Buddhistic 
times, hut what was the “empire” of any king before Agokal Certainly 
not that of the Pandus. It is significant, in view of the great importance 
laid by some scholars on the cakravartin idea, that this word does not occur 
before the later Upanishads, although “ great kings ” are mentioned ; nor is 
it an early epic phrase. 



the Pandus succeeded in attaching themselves to the tales 
which told of the old national heroes is unknown. All 
theories and hypotheses of development are pure guesswork. 
What we know is that the tales which told of Kurus and 
Bharatas became the depository of the Pandus, who appear to 
have substituted themselves for Bharatas 1 and may in fact 
have been a branch of the tribe, which from a second-rate 
position raised itself to leadership. There is a theory that 
the epic story has been inverted, in favor of the Pandus; 
there is another that it is what it pretends to be, the strife of 
Pandus, calling themselves Bharatas, with the scions of the old 
Kurus. With the former, that so persuasively advanced by 
Professor Holtzmann, I have never been able to agree ; but 
my own theory I have from the beginning put forward merely 
as one of probable epic growth. 2 

While, however, it is necessary to recognize the doubtful 
character of speculation in regard to the exact course of epic 
development, it is not desirable to blink the truths that are 
made clear in view of the facts we actually possess, the evi- 
dence of remaking, the base of the poem resting on old Kurus 
and Bharatas, the present structure of Pandu material; the 
age of the Pandu poem as a whole (synthetically considered), 
evinced inter alia by its recognition of late philosophical 
writers such as Pancaciklia (c. 100 A. D.), by a growing 
modernness of metre, by acquaintance with Greeks and Greek 
art, etc. 

Putting these facts together with those gleaned from other 
works than the epic itself, we may tentatively assume as 
approximate dates of the whole work in its different stages : 
Bharata (Kuru) lays, perhaps combined into one, but with 
no evidence of an epic before 400 B. c. A Mahabharata tale 

1 The Bharat! Katha (never “ Pandu-tale ”), as the received name of the 
epic, certainly favors this view. 

2 This I was careful to point out at its first presentation in my Euling 
Caste (now nearly fifteen years ago) with mays and mights and seems, and 
other useful words. As a theory I still consider this the best yet offered, 
but I have never held it to be demonstrable, only more or less probable, in 
outline and detail respectively. 



with Pandu heroes, lays and legends combined by the Puranic 
diaskeuasts, Krishna as a demigod (no evidence of didactic 
form or of Krishna’s divine supremacy), 400-200 b. c. Re- 
making of the epic with Krishna as all-god, intrusion of 
masses of didactic matter, addition of Puranic material old 
and new' ; multiplication of exploits, 200 B. c. to 100—200 A. I). 
The last books added with the introduction to the first book, 
the swollen Anugasana separated from (atnti and recognized 
as a separate book, 200 to 400 A. D. ; and finally 400 A. D. + : 
occasional amplifications, the existence of which no one 
acquainted with Hindu literature w'ould be disposed antece- 
dently to doubt, such as the well known addition mentioned 
by Professor Weber, Lectures on Literature, p. 205 ; and per- 
haps the episode omitted by Ksemendra, 1 Indian Studies, No. 
ii, p. 52. 

In the case of these more precise dates there is only reason- 
able probability. They are and must be provisional till we 
know more than we know now. But certain are these four 
facts : 

1 ; That the Pandu epic as we have it, or even without the 
masses of didactic material, was composed or compiled after 
the Greek invasion; 2, That this epic only secondarily de- 
veloped its present masses of didactic material ; 3, That it did 
not become a specially religious propaganda of Krishnaism 
(in the accepted sense of that sect of Vaisnavas) till the first 
century B. c. ; 4, That the epic was practically completed by 
200 A. D. ; 5, That there is no “ date of the epic ” which will 
cover all its parts (though handbook makers may safely 
assign it in general to the second century B. c.). 

The question whether the epic is in any degree historical 

1 We cannot, however, be too cautious in accepting the negative evidence 
of one maiijari, or precis, as proof that the original work lacked a certain 
passage. I dissent altogether from the sweeping statement, made loc. cit., 
p. 27 : “ The importance of the condensations lies in the fact that by means 
of them we are enabled to determine the state of these works (epics, etc.) 
in his (Ksemendra’s) time.” Two or three compendia agreeing on one point 
of omission might “ determine,” hut one resume alone can only create a 
possibility, as in this case (p. 53 note). 



seems to me answerable, though not without doubt, and I 
cannot refrain from expressing an opinion on a point so im- 
portant. As I have remarked above, there is no reflex of 
Pandu glory in Brahmanic literature before the third or fourth 
century. It is, further, impossible to suppose that during the 
triumph of Buddhism such a poem could have been composed 
for the general public for which it was intended. The metre 
of the poem shows that its present form is Later than the epic 
form of Patafijali’s epic verses, but this indicates simply re- 
casting ; so that a Pandu Mahabharata may have existed pre- 
viously, as implied by Panini. But while a Buddhist emperor 
was alive no such Brahmanic emperor as that of the epic 
could have existed, no such attacks on Buddhism as are in the 
epic could have been made, and the epic of to-day could not 
have existed before the Greeks were personally familiar. In 
other words, granted a history, that history must have been 
composed at least as late as the history was possible. Panini's 
allusions and those of Buddhistic writers show that the Pandus 
were known as heroes. It is, further, most improbable that 
the compilers, who made the poem represent Pandu virtues 
and victories, would have chosen them for this position had 
they been mythical. In their reassertion of Brahmanism they 
would have chosen rather the well-known ancient Brahmanic 
heroes of the older tale, Bharatl Katha ; yet to appeal to the 
people something real and near was necessary. But while 
before the second century the conditions were lacking which 
could have produced the poem, with the second century they 
became possible ; 1 and there was already the Pandu tribe 

1 As this book goes to press I receive Kirste’s essay Zur Mahabharata- 
frage, who says, p. 224, “ It is incredible that the work could have been 
undertaken so long as a royal family favoring that sect (of Buddhists) 
reigned. This (state of affairs) suddenly changed when the Maurya dynasty 
(of Brhadratha) was overthrown by Pusyamitra in 178 b. c., for the new 
ruler opposed the Buddhists.” Professor Kirste thinks, indeed, that the 
polyandry of the heroes is not an historical trait, and gives a very ingenious 
explanation of it as a myth of divided divinity, which, however, scarcely 
seems to me probable. But I am glad to find my own suggestion, of the im- 
probability of the anti-Buddhistic epic being cast in its present shape before 
the second century n. c., supported by this independent reference to actual 
historical data. 



with its perhaps justified claim to be considered a branch of 
the Bharatas, its own later heroes, its cult of anti-Buddhistic 
type. In so far, then, as we may discern a historical germ in 
the midst of poetic extravagance, it would seem that the poem 
represents an actual legend of a real tribe, and in so far as 
that legend persists in its adherence to polyandry as an es- 
sential part of the legend, a tribe which, like so many others 
in India, had been brahmanized and perhaps become allied by 
marriage to the old Bharata tribe, whose legends were thus 
united with its own. 

Finally, I would speak shortly of the poem as a literary 
product of India. In what shape has epic poetry come down 
to us ? A text that is no text, enlarged and altered in every 
recension, chapter after chapter recognized even by native 
commentaries as praksipta, in a land without historical sense 
or care for the preservation of popular monuments, where no 
check was put on any reciter or copyist who might add what 
beauties or polish what parts he would, where it was a merit 
to add a glory to the pet god, where every popular poem was 
handled freely and is so to this day. Let us think ourselves 
back into the time when the reciter recited publicly and dra- 
matically; let us look at the battle scenes, where the same 
thing is repeated over and over, the same event recorded in 
different parts of the poem in slightly varying language. 
The Oriental, in his half-contemptuous admission of epic 
poetry into the realm of literature, knows no such thing as a 
definitive epic text. The Vedas and the classics are his only 
real care. A Bharatavid in India is even now more scorned 
than honored. 

If the epic as a whole belongs to no one era, and this re- 
mains an incontrovertible fact, it is then in the highest degree 
probable also that no one part of the whole can be assigned 
to a certain period. I mean, not only must we admit that 
old books contain more recent insets, as for example chapters 
five and eleven of book ii, and that late books contain old 
passages, as for example the rape of Subhadra and the burn- 
ing of Khandava in book i, or the lotus-theft in book 



xiii, but we must admit further that the smaller divisions, 
these special scenes themselves, have in all probability not 
remained untouched, but that the tale, the language, and the 
verse of the epic have been subjected to an evening process 
irregularly applied since first the poem was put together as a 
Mahabharata; great liberty being taken with the poem both 
by reciters and copyists, the establishment of the text by com- 
mentaries (noticed as early as the introductory chapter of the 
poem itself) proving no bar to occasional alterations and ad- 
ditions. Such changes were not introduced of set purpose 
(or the metre would have been made more uniform), but 
incidentally and illogically. The same tale was told not 
in identical language but with slight variations ; intrusions 
were not shunned; grammatical and metrical forms were 
handled freely, but with no thorough revision of form or sus- 
tained attempt at harmonizing incongruities of statement. It 
is for this reason that there is not a still sharper metrical line 
between old and new in the epic itself, and it is for tins rea- 
son that the epic verses of the Mahabhasya are freer than 
those of the Mahabharata. The former were fixed by their 
function as examples in a grammar ; the latter were exposed 
to constant though sporadic modification, and appear to-day 
as they survive after having endured the fret and friction of 
innumerable reciters and pedantic purists. One by one, and 
here and there, the transmitters, working neither in concert 
nor continuously, but at haphazard and at pleasure, have 
trimmed this mighty pile into a shape more uniform, though 
they have not altogether hid its growth, except from eyes 
that, seeing the whole as a thing of power and beauty, are per- 
haps less apt to mark the signs of varying age. 

But if this be so, it may be asked, and I think it will be 
asked, perhaps triumphantly, by those lacking in sobriety of 
judgment, what becomes of the results of the analysis of 
metres, of the discovery of late elements in this or that sec- 
tion ? What do they signify ? 

They signify and proclaim that the Great Epic was com- 
pleted in just the way the synthesist proclaims it was not 




completed. Pitched together and patched together, by the 
diaskeuasts and priests respectively, the older parts, though 
not free from rehandling, bear a general stamp of antiquity 
lacking in later parts. For this reason, the Gita and Gam- 
bling scene are, as wholes, metrically and stylistically more 
antique than are the Anugita and the extravaganzas in the 
battle-books ; and for this reason, the pseudo-epic comes 
nearest in syntax and forms to the hybrid language that is 
preserved in literary monuments immediately preceding and 
following the Christian era. But it is true that no one can 
prove the relative antiquity of the Gita and Gambling scene 
so absolutely as to prevent one devoid of historical sense 
from clinging to the notion that these parts of the epic are 
in origin synchronous with the pseudo-epic. Fortunately, 
however, the judgment of scholars is in general sane, and 
the determination of values may safely be left in their care. 



[M. is prefixed to Mbh. references only where confusion with R. is possible.] 

1, acirenai J va kalena, ix, 2, 5S ; R. v, 26, 23 ; vi, 61, 20 ; acirena 

tu, R. ii, 80, 11. 
atitayam, No. 94. 

2, atha dirghasya kalasya, iii, 70, 1; v, 160, 20; R. iv, 9, 17; 

vii, 99, 14 ; atha dirghena kalena, G. vi, 24, 3 ; R. vii, 24, 
5, 72; tato dirghena kalena, M. ix, 1, 50; sa tu dirghena 
k., ib. 48 ; 36, 10 ; atha kalena mahata, G. i, 40, 16 = R. 
38, 19, v. 1., atha dirghena kalena; atha k. m., also G. i, 
40, 22 = R., 38, 23, tatah kalena mahata. See above, p. 

atha ratryam, No. 94. 

atha ’nyad dhanur, No. 56, and No. 80. 

3, anayad Yamasadanam, vi, 54, 81 ; vii, 19, 15; G. iii, 34, 31 ; 

75, 28. See No. 225. 

4, anastamgata aditye, vii, 145, 19 ; acc., G. v, 3, 41 (in R. iv, 

67, 15, anastamitam) . 
anyat karmukam, No. 80. 
anyonyavadha 0 , No. 157. 

5, abhidudrava vegena, vi, 100, 49 ; 104, 34-35, etc. ; R. vi. 69, 

99 ; 76, 46. See No. 97. 

6, abhivadaye tva(m) bhagavan, iii, 207, 13 ; R. iii, 11, 72. 

7, amrsyamanas tam ghosam (tat karma), etc., H. iii, 60, 3 ; R. 

vi, 67, 142 ; 69, 141, etc. 

8, alatacakrapratima(m), iv, 61, 9; R. iv, 46, 13; vi, 93, 28. 

The first and last refer to weapons, R. iv, 46, 13 to earth, 
prthivl, alatacakrapratima drsta gospadavat krta. 

9, alatacakravat sainyam tada ’bhramata, viii, 81, 40 ; alatacakra- 

vac cakram bhramato 'rinirvahanam (sic!) G. iv, 5, 25. 
Compare, of persons, vi, 59, 22; vii, 7, 53; xiv, 77, 30. 



10, avaplutya rathat turnam, vi, 94, 22 ; 96, 39 ; G. vi, 18, 47 ; 

avatlrya, G. vi, 36, 87 ; rathad avaplutya tatah, M. vi, 59, 
99, etc. For other forms, see AJP. xix., p. 143. 

11, avasidanti, me pranah, iv, 61, 12; parisldanti me pranah, 

G. vi, 82, 6 = E. 101, 6, avasidanti gatrani. 

11 b, aqokah cokanaqanah, iii, 64, 107 ; aeokah qokavardhanah, 
E., iv, 1, 59. 

aQvanam khura° No. 247. 

12, astrani vividhani ca, vii, 7, 1 ; castrani, E. vi, 103, 29. The 

terminal is fixed, vasuni, vastrani, bhandani, etc., preced- 
ing, e. g., ix, 47, 24; 
asmin hate, No. 328. 
akarna, No. 170. 

13, Skrlda(m) iva Eudrasya ghnatah kalatyaye paQun, vii, 19, 

35 ; akrlda iva Eudrasya kruddhasya nighnatah paqun, G. 
vi, 73, 38 ; akridabhumih kruddhasya Eudrasye ’va ma- 
hatmanah, E. vi, 93, 35. Compare ix, 14, 18, Eudrasya 
’kridanam yatha. 

14, akhyatum upacakrame, xviii, 5, 7 ; E. iii, 11, 10 ; iv, 8, 46 ; 

52, 3 ; G. v, 66, 2, where E. 65, 2 has pravaktum upaca- 
krame. Compare vaktum samupacakrame, xiii, 87, 2. 
The phrase is common in E. ; rarer in M., owing to the 
use in the latter of the dramatic uvaca, extra metrum. 
Both epics have also the similar phrase vyahartum upa- 
cakrame, e. g., xii, 350, 15; E. vi, 115, 1; vii, 51, 1. See 
No. 57. 

15, ajaghano ’rasi kruddhah, vi, 61, 36; E. vi, 69, 152; 76, 29; 

passim in M. See 1. c., No. 10, p. 142, and note to No. 35. 

16, aditya iva tejasa, iii, 53, 2 ; E. vi, 55, 9 ; aditya iva tejasvl, 

E. v, 34, 28, metrical. See No. 176. 

17, alikhantam iva ’kaqam, iv, 38, 3; E. vi, 99, 12. 

18, avarta iva samjajne balasya mahato mahan, H. iii, 60, 4; 

G. vi, 32, 21 ; avarta iva gaiigasya toyasya, G. v, 50, 16 ; 
asld ganga iva ’vartah, M. vii, 36, 13. 

19, avista iva yudhyante, vi, 46, 3; avista iva kruddhas te (cakrus 

tumulam uttamam), G. vi, 54, 64. 

20, aqlvisa iva kruddhah, vii, 10, 31 ; E. v, 67, 7. 
aslt kila°, aslc catacata, etc., No. 334. 

21, asld raja Nalo nama, iii, 53, 1 ; asld raja Nimir nama, E. vii, 

55, 4. With Ylrasena-suto ball at the end of the first 


verse, compare Dyumatsenasuto ball, M. iii, 294, 18; suto 
ball, R. iii, 12, 2; Ayodhyayam pura raja Yuvaua^vasuto 
ball, R. vii, 67, 5; Prajapatisuto ball, R. vii, 90, 23 (in 
G., ‘bbavat). 

22, iti me nigcita matih, iii. 78, 6 ; G. v, 8, 25 (R. v. 1.) ; 68, 36 

(K- v'.l.). 

23, ity aslt tumulah Qabdah, vi, 119, 19; ity evam t. c., G. vi, 

19, 4 (R., evam sutumulah gabdah). Compare babhuva t. 
q., M. vi, 56, 22, etc.; R. vi, 58, 17, etc.; samjajne t. q., 
M. vi, 46, 17, and 1. c. No. 10, p. 144, ff. Compare Nos. 

24, idarn vacanam abravlt, iii, 69, 17, etc. ; R. i, 26, 33 ; iv, 8, 1, 

etc. Sometimes tato for idam, ix, 3, 51 (= C. 176, idam). 
About forty times in Ram., unnumbered in Mbh. See 
No. 237. 

25, Indradhvaja ivo ’cchritah (tato nipatito bhurnau), ix, 17, 53 

and often; Indraketum ivo ’cchritam, ix, 4, 16; Qakra- 
dhvaja ivo ’cchritah, R. v, i, 59. Compare utthapyamanah 
Qakrasya yantradhvaja ivo ’cchritah, R. ii, 77, 9 ; maha- 
merum ivo ’cchritam, ix, 37, 20 ; ubhav Indradhvajav iva 
(petatuh), ix, 12,24; dhvajav iva mahendrasya (nipetatuh), 
R. vi, 45, 17-18 ; jagama vasudham ksipram Cakrasye ’va 
mahadhvajah, G. iii, 34, 25 ; apatad devarajasya muktara- 
gmir iva dhvajah, R. iv, 17, 2; Indradhvaja ivo ’tsrsto 
yantranirmuktabandhavah (papata), M. vii, 93, 70 ; yan- 
tramukta iva dhvajah (papata), M. vii, 92, 72 ; yantracyuta 
iva dhvajah (papata), G. ii, 84, 8. 

Indracani, No. 275. 

25 b, ihai ’va prayam asisye, x, 11, 15; R. iv, 53, 19. 

26, uttistha rajan kirn cese, xi, 2, 2 ; G. vi, 95, 37 ; rajann uttistlia 

kirn cese, G. ii, 81, 10 ; uttistho ’ttistha, Gandhari, xi, 26, 
1 ; uttistho ’ttistha, kim Qese, R. vi, 111, 81 (preceded by 
No. 45) ; uttistho ’ttistha, bhadram te, M. i, 172, 4 ; R. i, 
35, 2; preceded in Mbh. by uvaca madkuram vakyam, 
•with which compare ix, 36, 50, uvaca parusam vakyam; 
ucub sumadhuraria vanlm, R. vii, 70, 1 ; bhadram te being 
current ad nauseam in both epics, 
uvaca . . . vakyam, No. 26. 

27, ekantabhavopagatah, xii, 337, 28; ekantabhavanugatah, R. 

vii, 38, 5. In both, of the men in £ v etadvipa, preceded 



in M. by tatra Narayanapara manavaQ canaravarcasah ; in 
E., by ananyamanaso nityam Narayanaparayanah tada ra- 
dbanasaktac ca taccittas tatparayanah (ananyamanasah is 
a Gita phrase, 9, 13, bbajanty ananyamanasah ; 8, 14, ana- 
nyacetah satatam). 

28, etac chrutva tu vacanam, vi, 48, 98 ; G. iv, 56, 19, and passim. 

29, etat te kathitam sarvam and (in prior pada) etat te sarvam 

akhyatam ; ix, 46, 108 ; G. vi, 82, 167. In M. preceded 
by yan mam tvam pariprccbasi, as in xii, 334, 40 ; xiii, 14, 
139, etc. 

30, etasminn an tare vlrah, vi, 48, 96, and often; E. iii, 30, 37; 

vi, 50, 7 ; vii, 28, 19 ; G. vi, 36, 99. The phrase here is 
etasminn antare, which is filled out with various words, as 
Eamah, E. vi, 111, 91; tatra or tasya (v. 1.), E. vi, 92, 58 ; 
kruddhah, E. vi, 100, 13 ; krodhat, 102, 47. Compare also 
etasminn antare $unye, M. vii, 17, 7 ; xii, 330, 1 ; cai ’va, 

vii, 19, 38; qurah, ix, 28, 17; G. vi, 32, 15, etc. A com- 
bination of this and the next (No. 31) is found in etasminn 
antare kale, “ in the meantime,” R. vi, 20, 33. 

31, etasminn eva kale tu, like the last, a standing phrase, e. g., 

i, 149, 1 ; iii, 54, 13 ; 168, 13 ; 298, 1 ; v, 121, 9 ; vi, 74, 36 ; 
ix, 51, 25 ; xii, 328, 3, etc. ; E. i, 9, 7 ; 33, 11 ; G. 21, 1, 

32, evam uktah pratyuvaea, or tathe ’ty uktva, vi, 59, 47 ; vii, 202 

70 ; ix, 35, 68 ; G. vi, 36, 102. Compare evam astv iti 
(with pratyuvaea), ix, 48, 52 ; G. vi, 109, 18 (co ’vaca) ; 
(krtva sa), ib. 82, 56. 

33, kaksam agnir iva jvalan, ix, 24, 62 ; kaksesv agnir iva jvalan 

4, 36 (C., kakse 'gnir iva samjvalan) ; vanany agnis ivo 
’tthitah, E. vi, 66, 12; kaksam agnir ivo ’tthitali, G. v, 
85, 24 ; kaksesv iva hutacanam, G. ii, 106, 25. Compare 
also (dahantam) kruddham agnirii yatha vanam, M. vii, 21, 
30 ; vanam agniri vai ’dhitah, E. ii, 63, 44, where G. 65, 
39, has quskarii kastham iva ’nalah, like E. v, 41, 11, <prskam 
vanam iva ’nalah. The iva ’nalah ending is common to 
both epics, e. g., dahan kaksam iva ’nalah, M. vii, 14, 1 
(followed in 2 by s&ksad agnim ivo ’tthitam, C. vrksam) ; 
tan me dahati gatrani (juskavrksam iva ’nalah, M. vi, 95, 7, 
etc. See also Eos. 75, 99, 117, 196, 226, 256, 291. 

34, kankapatrair ajihmag&ih, vi, 103, 11 and often; E. vi, 52, 4. 


Frequently close together with s varna, rukma, or herna 
punkhair ajihmagaih, vi, 114, 11 ; vii, 18, 18, hema ; G. 
vi, 19, 68. In G. vi, 20, 26, rukma° ajihmagraih, metrical 
(v. 1. in R.). The common terminal qarair ajihmagaih is 
sometimes inverted in jagatls, as in G. iv, 30, 22, though 
the regular qloka order is also found in this jagatl metre, 
ib. 34, 34. See No. 234. 

35, Kandarpa iva nipena, murtiman, iii, 53, 15; rupavan . . . 

kandarpa iva murtiman, R. v, 34, 30. This with aditya iva 
tejasvl, is a description of Rama, 28, as the two phrases, 
and also satyavadl (R. 29), here describe Nala. 

36, kampayann iva medinlm, ii, 29, 7 ; viii, 34, 58; ix, 18, 26, etc. ; 

kampayaiiQ ca ’pi, ix, 30, 60 ; sa kampayann iva mahlm, iii, 
78, 3 ; kampayann iva medinlm, G. vi, 37, 101 ; R. vi, 56, 
13 ; 67, 115 ; kampayantl ’va, G. iii, 62, 31 ; kampayantl 
’va parvatan, M. vii, 181, 11 ; calayann iva medinlm, R. 
iii, 67, 13; darayann iva, R. iv, 15, 5 (G. kampayann); 
darayann iva parvatan, M. iv, 46, 21; nadayann iva medi- 
nlm, G. vi, 46, 91. purayann iva medinlm, M. iii, 73, 8 
(purayanto dieyj daqa, ix, 46, 77), etc., etc. For dico daca, 
see No. 114. 
karam karena, No. 163. 
karnayata, No. 170. 

37, karmana manasa vaca, iii, 65, 32, 41 ; ix, 50, 2 ; xii, 327, 34 ; 

manasa karmana vaca caksusa ca, R. vii, 59, 1, 24. Com- 
pare Spriiche, 1,559 ff., 2,222 ; Dhammap. 391. 
kalarii na ’rhati, No. 196. 
kalpyatam me rathah, No. 230. 

38, kasaylkrtalocandh, °am, i, 102, 23; 131, 3; G. vi, 33, 17; 37, 

68. In M., sakrodhamarsajihmabhruh precedes in each 
instance. Compare Nos. 50, 51. 
kasya ’si. See above, p. 268. 

39, kancanosnlsinas tatra vetrajharjharapanayah, vi, 97, 33; 

kancukosnlsinas tatra vetrajharjharapanayah, R. vi, 114, 
21. Compare G. vi, 33, 10 and 13, vetrajharjharapanibbih. 

40, kamabanapraplditah, i, 220, 7 ; G. iii, 61, 2 (R. 55, 2, banaih) ; 

kamabanabhisamtaptah, iii, 280, 3; kamabanavacamgatah, 
R. vii, 88, 12. 

41, Kalacakram ivo ’dyatam, vii, 7, 31 ; iva ’param, jG. vi, 73, 33 

(R., 93, 30, iva prajah) ; kaladandam iva ’param (R., iva 



’ntakah) ; G. vi, 51, 89 = R. 71, 85. For the var. lec., 
compare s. dandahasta, No. 104, and kalaratrim iva ’ntakah, 
K. vi, 69, 134. Compare kalaratrim ivo ’dyatam, ix, 11, 
50; °suryam, xiii, 14, 270. 

Kaladandopama and Kalapacopama, No. 220. 

Kalananam, No. 272. 

Kalaratrim, No. 41. 

42, Kalantakayamopamah, iii, 22, 31 ; 27, 25 ; iv, 33, 25 ; vi, 54, 

47 ; G. iii, 32, 5 ; vi, 49, 36; R. vi, 57, 32 ; 60, 94; 82, 7 ; 
95, 41. See No. 220; and for Kalantakopama, see Nos. 

104, 105. 

43, Kalo hi duratikramah. While not generally including in this 

list the proverbs common to the two epics, I enter this 
particular proverb because of the similar environment in 
imam avastham prapto 'smi, Kalo hi duratikramah, ix, 64, 
9 (C. vai) ; so ‘yam adya hatah Qete, Kalo hi duratikramah, 
R. iii, 68, 21. For the rest, compare Am. Journ. Phil., 
vol. xx, p. 26, and add (besides the above) Kalo hi durati- 
kramah in M. ii, 46, 16 ; also H. iii, 2, 30, and 5, 36 ; dai- 
varii hi duratikramam, R. vii, 50, 18 ; daivaih tu, ix, 65, 
31 ; and the later version, lekha hi kalalikhitah sarvatha 
duratikrama, H. iii, 2, 27. 

44, kinkinljalasamvrta, ix, 23, 13, °aih rathaih; R. vii, 23, 1, 2, 

°am nagaram. Ordinarily in M., kinkinljalamalinam, etc., 
i, 221, 45 ; ii, 24, 18 ; viii, 86, 4 ; in R., kinkinlqatabhusita, 
vi, 102, 9 ; but I cannot say whether or not inalin appears 
in R. in this combination. See No. 113. 

45, kim marii na pratibhasase, part of a lament (see uttistha, 

above, No. 26), iii, 63, 9; 64, 19 ff.; xi, 20, 13-14; R. iii, 
60, 26; vi, 111, 80 (doubled in G. 95, 36, and v. 1. 37). 
In R. vi, 115, 15 (= G. 98, 12) kim ca mam na ’bhibha- 
sase, v. 1. as in G. 95, 37. 

46, kuqalam paryaprcchata, ix, 34, 17 ; R. i, 52, 4. 

47, krtakautukamangalah, i, 129, 24 ; viii, 1, 11 ; R. i, 73, 9. 
krtapurvahnikakriyah, No. 49. 

48, krtva karma suduskaram, vi, 14, 14 ; vii, 8, 32 ; R. ii, 101, 

5 ; vi, 76, 70 ; G. vi, 21, 11 ; 30, 37 ; 55, 36. Variations 
are naturally many, e. g., karma kurvanam duskaram, vi, 

105, 6; krtam karma suduskaram, R. vi, 67, 55; 127, 47 ; 
G. vi, 88, 17 ; karma kurvanti duskaram, R, vi, 65, 4 ; tat 


krtva duskaram karma, E. vi, 126, 14 ; karisyan karma 
duskaram, G. iv, 15, 20. Similar in R. are mahat karma 
krtam tvaya and krtam tvaya karma mahat suduskaram, 
G. vi, 112, 100 and G. vi, 36, 118, respectively ; aho mahat 
karma krtam nirartham, E. v, 48, 50; sadhu, Laksmana, 
tusto 'smi, karma te sukrtam krtam, G. vi, 70, 80; sudus- 
karam. tu tat karma, G. iv, 11, 7. Somewhere in M. ix 
(verse lost) occur together the two phrases, krtva na su- 
skaram karma, gato Vaivasvataksayam (No. 55). 

49, krtva paurvahniklh kriyah, iii, 168, 2 ; 296, 10 ; °kam karma, 

E. iii, 17,2; °klru kriyam, E. vii, 59, 1, 1; krtapurvah- 
nikakriyah, viii, 1, 13; E. i, 35, 3 (with the phrase, tac 
chrutva vacanam tasya). 

50, krodhasaiiiraktanayamih, i, 78, 35; vii, 1, 19; E. i, 62, 15; 

G. v, 89, 1 ; vi, 76, 11. In M. v, 9, 45, united with idam 
vacanam abravlt. See note to No. 51. 

51, krodhasamraktalocanah, v, 178, 40; vi, 100, 52; ix, 42, 13; 

E. v, 44, 19 ; vi, 95, 3 ; krodhat sam°, E. iv, 9, 22 ; vi, 98, 1. 
Both forms, No. 50, No. 51, are common in both epics. 
They are the same phrase differentiated according to 
metrical requirements, and interchange with the similar 
kopa- and rosa-forms, which it is unnecessary to give 
in detail. Variants are common, e. g., krodhaparyakuleks- 
anah, v, 178, 94; G. iv, 15, 17; often united with another 
iterate, e. g., rosasamraktanayana idam vacanam abravlt, 
G. iii, 57, 15 ; samraktanayanah krodhad (G. kopad) idam 
vacanam abravlt, E. vi, 59, 56 = G. 36, 33. Compare tarn 
krodharaktanayanam kurvantahi bhrukutimukham, G. iv, 
33, 40 ; sa krtva bhrukutlm vaktre rosasamraktalocanah, 
G. vi, 86, 46, where E. 102, 38, has sa krtva bhrukutlm 
kruddhah kimcit saihraktalocanah. See Nos. 106, 123, 
190, 198, and s. v. PW., where they are illustrated suffi- 

52, kroqantim kurarlm iva, i, 6, 12 ; G. ii, 68, 43 ; E. iv, 19, 29 ; 

yatha, vi, 32, 3; plural, xi, 12, 10; 16, 18; variauts, G. ii, 
67, 16 ; iv, 19, 4 ; v, 18, 12 ; E. vi, 49, 9, etc ; kurarlm iva 
vaQatim, M. iii, 63, 20. That in G. ii, 67, 16, the unusual 
form kuraryas trasita iva follows the exclamation ha natha 
h5 mrto 'si 'ti in 12, just as ha natha in N. 11, 23 follows 
kurarlm iva vaqatlm in 20 (above), is perhaps worth 



noticing, especially as this chapter of R. G. is not in the 
Bombay text and may be supposed to be late. The corre- 
spondence is not remarkable enough to prove copying, 
though it may be due to the influence of the Nala passage, 
as this episode is well known to the later Ramayana. 

53, ksitikampe yatha cailah, vii, 174, 23; yatha ’calah, vii, 36> 

29 ; ksitikampe yatha nagah, G. vi, 30, 30, where R. has 
ksitikampa iva drumah, 56, 31. See No. 248. 
khuranemisvanena ca, No. 247. 

54, gatapratyagatani ca, term, tech., vii, 19, 6; R. vi, 107, 32. 

See mandalani, No. 201. 

55, gato Yaivasvataksayam, or ninye, vii, 26, 53, and s. krtva 

karma, No. 48, above ; R. vi, 82, 183. 

56, gadam adaya vlryavan, ix, 11, 49 ; 32, 37 ; 55, 24 ; 56, 27, etc. ; 

R. vi, 69, 33. In G. vi, 49, 18, vipulam. See 1. c. No. 10, p. 
142, and No. 80, for parallel variants. 

57, gamanayo ’pacakrame, i, 151, 14 ; R. vii, 25, 51 ; gamanaya 

’bhicakrama, R. i, 77, 18 (G. 79, 4, upa°). See No. 14. 

58, Garudah pannagam yatha, viii, 87, 96 ; R. vi, 69, 6, °gan iva, 

where G. 48, 6, has °gaiii yatha; G. vi, 46, 3 has °gan iva. 
Many var. lec., e. g., Garutman iva. 

59, garjantau iva toyadau, ix, 55, 38 ; °tam, G. vi, 3, 19 ; garjanti 

na vrtha (jura nirjala iva toyadah, R. vi, 65, 3. See Nos. 

60, girih prasravanair iva, iii, 279, 5, with cakara rudhiram bhuvi 

preceding ; R. vi, 67, 89, with raraja qonitotsiktah preced- 
ing. G. vi, 46, 75 has giripra, an error. Compare G. ib. 
109, girih prasravanam yatha ; R. vi, 67, 121, girih prasra- 
vanair iva. In R. vi, 58, 55, gireh prasravano yatha, where 
G. 32, 43 has jalaiii prasravanad iva, as in R. vi, 45, 21, 
jalaiii prasravanav iva, and R. vi, 88, 61. 
gairikam, No. 318. 

61, cakara kadamam mahat, vii, 21, 37 ; R. vi, 86, 24; 95, 50; 

G. vi, 46, 108 ; karomi, M. iv, 21, 2 ; kurvanah, ix, 61, 30 ; 
akari, G. vi, 49, 43 ; krtva ca, G. vi, 110, 50 ; akarot, M. vii, 
32, 41 ; ix, 44, 3 ; cakara kadanam ghoram (metre), R. vi, 
58, 24; H. iii, 60, 3; kadanam sumahat cakruh, R. vi, 

62, caksurvisavam agatah : In vii, 17, 14, sa no distya ’strasam- 

pannatj caksurvisayam agatah; R. vi, 103, 19, distya ’si 


mama mandatmanq caksurvisayam agatah (G. 88, 24, mama 

63, candrasuryiiv iyo ’ditau, ix, 55, 22 ; G. v, 53, 25 = 69, 23 ; 
stiry acarid ramasav iva, M. iii, 2S8, 26. See Nos. 33, 189. 
cayattalaka, No. 186. 

caled dhi Himavan sthanat, ii, 77, 35; cailah, v, 82, 48; caled 
api ca Mandarah, G. v. 58, 9 (R. 59, 14, Mandarah pracaled 
api). See No. 153. 

65, camikaravibhusitlm, gadam, x, 9, 11 ; capam, R. iii, 20, 6. 

66, cittapramathini (bala devanam api) sundarl, iii, 53, 14; trai- 

\dkya.-sundari (kanta, sarva-) eittapramathinl, R. vii, 37, 1, 
29 (compare R. ii, 10, 30, mama cittapramathini). As said 
above, the Uttara recognizes the Nala, and this (praksipta) 
may be imitation. At any rate it may support pramatkinl 
against the Mbk. Bomb, and Calc, reading here, cittaprasa- 
danl, which, however, is found in xii, 133, 13, janacittapra- 
sadini ; compare naracittapramathibhih, R. i, 10, 4. 

67, citram laghu ca susthu ca, vii, 145, 77 ; laghu citraiii ca susthu 

ca, R. vi, 88, 65. 

68, cinta me vardhate 'tlva mumursa ca ’pi jayate, Karnasya 

nidhanam'crutva, viii, 9, 6; cinta me vartate tivra mumtirsa 
’pi ca jayate, bhrataram nihatam drstva, R. vi, 101, 7. See 
No. 213. 

69, c i n t ac] o k a p a r ay a i la h, vii, 1, 6; xv, 16, 18; G. iii, 52, 17 ; vari- 

ants, viii, 96, 58; xv, 21, 7. See Nos. 27, 116, 161, 293. 

70, chaye ’va ’nugata pathi, iii, 65, 57 ; chaye ’vft ’nugata Ramam, 

R. vii, 37, 3, 24, after rupena ’pratima loke (No. 236), also 
a Nala phrase. Compare No. 66. 
chinnamula iva drumah, No. 248. 

71, chinne ’va kadall vane, xi, 17, 1, nyapatad bhumau ; G. vi, 8, 

6, papata bhumau (both of grief-stunned woman) = R. vi, 
32, 6, but here jagama jagatim bala chinna tu kadall yatba. 
See Nos. 135, 136, ISO, 248. 
jarjarlkrta, Nos. 184, 235. 

72, jalarn surya iva ’mpibhih, vi, 109, 33 ; rnegham surya, G. vi, 

18, 40 (R. 43, 29, karair megham iva ’inhuman) ; tamah 
surya iva ’iipubhih, M. vii, 18, 24. 
jalam prasravanad iva, No. 60. 
jajvalyamana, No. 176. 
jatarupapariskrta, No. 335. 



73, jlmuta iva bliaskaram, vi, 64, 44 ; °tam iva °ah, G. vi, 21, 43 ; 

nlharam, E. i, 55, 25 ; toyadad iva bhaskarah, G. iv, 12, 24 
(papata). See No. 326. 

74, jlrnam tvacam ivo ’ragah, xiii, 62, 69 ; E. iii. 5, 37 ; sarpo 

jlrnam iva tvacam, xii, 265, 15 ; G. vi, 21, 40 ; tvacam sarpa 
iva ’mucya, M. v, 40, 2. See Nos. 106, 139 ; PraQ. v, 5. 
jvalantam iva tejasa, No. 176. 

75, jvalantam iva pavakam, j valanta iva pavakah (and jvalita iva), 

vi, 16, 12; 18, 6; xi, 25, 16, etc.; E. iii, 32, 5; vi, 50, 36; 
70, 19 ; 95, 33 ; G. 68, 36. Compare prajvalitam ivo ’lkam, 
M. v, 181, 5 ; praj valantam iva ’nalam, G. iii, 18, 23 ; 
jvalantam iva pannagam, M. vi, 82, 36 ; ix, 13, 21 ; G. iii, 
18, 39, pannagaih (but E. 12, 34, pavakaih) : also parvatam, 
M. vii, 80, 37, apacyata (on fire as it were). See Nos. Ill, 
176, 226, 255. For iva ’nalah, see Nos. 33, 99, 196, 291. 

76, jhillikagananaditam, iii, 64, i ; E. iii, 2, 3. The two descrip- 

tions (of a fearful forest) are similar also in the adjacent 
verses, e. g., nanapaksiganakirnam, in M. ; nanamrgagana- 
klrnam, in E. I have not entered others. 

77, ta enarn caradharabhir, dharabhir iva toyadah, vii, 26, 54; 

athai ’naiii caradharabhir, dharabhir iva toyadah, E. vi, 71, 
92 (in M., sisicuh ; in E. abhyavarsata) ; abhyavarsat tada 
Eamam dharabhir iva toyadah, E. vi, 100, 59; vavarsa 
(jaravarsena dh. i. t., M. vi, 58, 26. Compare mahendra iva 
dharabhih cjarair abhivavarsa ha, E. vi, 56, 11. See Nos. 
59, 158, 217, 244. 
tatah kilakila, No. 334. 

78, tatah prajavitacvena rathena rathinam varah. This hemi- 

stich H. 3, 59, 5 and also G. vi, 30, 6 (= E. 56, 6, but here 
pracalitacvena). The prior pada in M, vii, 116, 30 ; G. iii, 
33, 27 ; E. vi, 95, 42 (with rathena). See No. 287. 

79, tatah prabhate vimale, viii, 1, 9; xiv, 64, 16; E. vii, 59, 1, 1, 

with krtva paurvahnikim kriyam (No. 49) ; 68, 2. Com- 
pare prabhate vimale surye, E. ii, 86, 24. The first phrase 
is in tristubh as well as in qloka, loc. cit. 

80, tato ’nyad dhanur adaya, vi, 48, 67; G. iii, 34, 16, and 22. In 

the former of G., followed by pradlpta iva manyunfi (as in 
M. iii, 63, 13, pradlpta ’va ca manyuna). The usual phrase 
in M. begins with atha ’nvad, e. g., vi, 45, 33 ; 77, 68 ; 114, 
28; vii, 21, 17; ix, 10, 34; 15,21. Compare anyat karmu- 


kam adaya, and so 'nyat karmukam adaya, vi, 45, 29 ; 110, 
40 ; ix, 10, 45, etc. ; E. as cited loc. cit., No. 56. 
tato inuhurtam, No. 214. 

81, tato halahalaqabdah prltidah samajayata, i, 58, 9 ; tato hala- 

halaqabdas tumulah samajayata, E. ii, 1C, 33; the prior pada, 
M. vii, 21, 2 ; xiv, 74, 26 ; E. ii, 81, 14 ; vii, 21, 24 ; 32, 33 ; 
96, 12 ; G. iii, 31, 41 followed by the late trait, punah kola- 
halo mahan (not thus in M. or E.) ; G. ii, 82, 13, followed 
by sumahan samajayata. Compare No. 334. 

82, tatra ’sit sumahad yuddharii tumulaiii lomaharsanam, vi, 58, 

13 ; E. vi, 43, 16. For other forms, see 1. c. No. 10, p. 
144 if. In E., roma for loma, but according to Winternitz, 
loc. cit., these forms interchange also in MSS. of M. See 
Nos. 23, 83, 84. 

83, tad adbhutam iva ’bliavat, iii, 167, 17 and 31 ; v, 131, 25; vi, 

47, 28 ; 54, 82; vii, 7, 53 (with alatacakravad rajan) ; 14, 
27 and 38; 21, 14; ix, 12, 13; xii, 334, 2 and 4 and 11, 
etc., etc. G. i, 75, 28. Compare G. iii, 33, 22, tad abhud 
adbhutam yuddham tumulaiii lomaharsanam; E. iii, 51, 3, 
tad babhuva ’dbhutaih yuddham ; E. vi, 102, 18, tad babhau 
ca ’dbhutam. yuddham . . . romaharsanam ; M. xi, 16, 4, 
ranajiraih nrvlranam adbhutarii lomaharsanam ; ix, 15, 28, 
tatra ’dbhutam apaqyama, and 15, 41, tatra ’dbhutam paraiii 
cakre. In M. iii, 76, 41, tad adbhutatamarii drstva; E. vii, 
79, 1, tad adbhutatamarii vakyarii (jrutva. See also Nos. 
82, 84, 110. 

84, tad yuddham abhavad ghoram, vii, 16, 12 (sumahal loma- 

harsanam) ; G. vi, 58, 34 (in E., 79, 23, tatra for ghoram). 
M. adds devanam iva danavaih, wherewith compare E. vi, 
79, 2,