Skip to main content

Full text of "Studies in Etymology"

See other formats


STOP 



Early Journal Content on JSTOR, Free to Anyone in the World 

This article is one of nearly 500,000 scholarly works digitized and made freely available to everyone in 
the world by JSTOR. 

Known as the Early Journal Content, this set of works include research articles, news, letters, and other 
writings published in more than 200 of the oldest leading academic journals. The works date from the 
mid-seventeenth to the early twentieth centuries. 

We encourage people to read and share the Early Journal Content openly and to tell others that this 
resource exists. People may post this content online or redistribute in any way for non-commercial 
purposes. 

Read more about Early Journal Content at http://about.jstor.org/participate-jstor/individuals/early- 
journal-content . 



JSTOR is a digital library of academic journals, books, and primary source objects. JSTOR helps people 
discover, use, and build upon a wide range of content through a powerful research and teaching 
platform, and preserves this content for future generations. JSTOR is part of ITHAKA, a not-for-profit 
organization that also includes Ithaka S+R and Portico. For more information about JSTOR, please 
contact support@jstor.org. 



IV.— STUDIES IN ETYMOLOGY. 
I. 

The Europeo-Armenian Treatment of tr\. 
i. 

otAAg> I \ TreXofiat '. r\a<o 

\ T«XXg> 

su-stul-it tollo, tulit, su<b>latus 

ab-stulit 

tdrati, turdti, iulaydmi 

trans 
tirds : clam : w\d-yms (?) 
clddes, clandestinus 

7rXet<!S«r 

'triones 



a-trrrjp 






Stella 






strbhis 




taras 


Armen. a-st\ 






ireXayoff 




taramga 


irXd^o/iai 




celer : tdras 




Lith. 


Mill, ke'ltas : celsus. 



Sk. tdrati 'cross over' (river or sky) has not been heretofore 
connected with, the above group of verbs. Collitz, BB. v. 101 fg., 
compares Tf\eda> with nikopai, from an I. E. hj qel-. Homer uses 
both these verbs as a copula, r 3 xAayyi) 7r<?X« olpavodi 'the noise 
rises to heaven' and H 282 vCg reXe'dn 'the night arises' connote, 
however, upward motion. Lat. colo 'till,' Sk. cdrati 'wander' 
(of leisurely motion), 'graze,' Collitz also compares. Neither colo 
nor cdratihzs, the sense of 'rise, cross over,' and, what will be of 
greater importance presently, no sense of rapid motion. 

Joh. Schmidt, in KZ. 25, 138, added Grk. re'XAu, ocarAXo) 'rise' 
to this group, making the striking equation TtipvreXhopivav iviavrav 

= irepniKofiivayv iviavTa>i>. 

Fick, however, in the latest edition of his Worterbuch, refers 
T«XAca to an I. E. stem tela-, which he defines by 'raise.' 



464 AMERICAN JOURNAL OF PHILOLOGY. 

But I cannot bring myself to separate the equivalent phrases 

irfpiTeWo/ievcov iviavrSiv and irtpmXoixevav k. t. X., though I separate 

reXXo) from colo, Sk. cdrati. Cdrati+ud is, to be sure, one of the 
regular phrases for sunrise; with this Schmidt compares avanWa. 
The words are, it seems to me, hardly on the same chronological 
footing. Ud+ si car is freely used of the heavenly bodies from 
the earliest Vedic times. <Wt«Xj 'caused to spring up (as food),' 
E 777, is an ana£ Xiyofievov, and avrokal 'rising places of the sun,' 
p. 4, is the same. For avareXXa of 'sunrise' L. and S. cite Hdt., 
Soph, and Aristoph., but Soph, uses re'XXa) in the same sense. It 
is a fair, if not a cogent, inference that in ud + V car the specific 
idea of 'rise' comes from the preposition, just as in the precisely 
parallel ud+ si i, ud+ sj yd; whereas t/XXo) has this force without 
the need of composition with such a preposition as ava. 

At this point let us seek to fix the meaning of Sk. V ' tr- a little 
more definitely. In combination with ava ' off, down ' we find a 
meaning 'descend,' especially used of the descent of a divine being 
to the earth ; cf. the Anglo-Indian avatar. Does this sense derive 
from compounding ava with a general notion of motion, as in the 
combination ava+\/i 'come down,' or does ava have a force 
comparable with that of German ab- in absteigen, or English dis- 
in dismount! Delbriick, S. F. v, p. 449, seems to decide for the 
former derivation. But the compound is susceptible of another 
interpretation, as e. g. ava+ si sd 'bind' > 'unbind, take the yoke 
off horses,' which Delbriiek makes 'take the horses (out from) 
under the yoke,' and ava+ si tan 'string' > 'unstring (a bow)'; 
so ava+ si tr 'rise' might suffer reversal to 'descend.' 

Ud+ si tr means 'come up out of (the water).' Here again it is 
questionable whether the signification 'rise' is partly inherent in 
the verb or wholly acquired from the preposition. 

Even in the derivative taramga 'wave, billow' it is difficult to 
decide between the sense of 'rising' and 'going across.' 

In Lat. trans, e. g. trans Rhenum, trans mantes, the sense is 
'over' or 'across,' but just which sense is original it is impossible 
to decide. If we bear in mind the interchangeability of 'over' 
and 'across' in our own language, we can see why the primary 
signification of si tr is so hard to fix. 

If we compare Sk. V If w ' tn reXka and its congeners, one or two 
close coincidences of derived meaning present themselves. B6ht- 
lingk, in the new dictionary, s. v. tar, defines 3) 'lay behind one 
(a road),' i. e. ' complete a journey,' e. g. rtasya panthdm na 



STUDIES IAT ETYMOLOGY. 465 

taranti duskrtah 'evil-doers finish not the road of the rta.' Pind. 

• o 

Ol. 2, 126 ereiXav Mos 6&6v is an exact parallel. Out of this special 
usage comes the general meaning ' complete, fulfil,' which we have 
in TcXor 'fulfilment' and its verb reKetv. 

ore'AXo) in its causative sense 'raise' corresponds still more closely 
with V tr 'rise.' Hesiod, Scutum 288 eVtoroXaoV 8e xn&vas \ 
iaraKaro 'they have tucked up their tunics,' etc., is a capital 
example for this sense, and eVioroXdoV gives it a clinching force. 
y 1 1 ioTta . . . aT(l\av ' they took in sail ' shows also the sense ' lift, 
raise.' Etymology comes to our help in the difficult question 
whether the Homeric ship had the yard fastened to the mast so 
that the furling was accomplished by raising the sail to the yard, 
or whether the yard and sail were lowered together in furling. 
For the former explanation see Smith's New Diet, of Antiquities, 
p. 218, s. v. navis. The causative 1 sense of otAXg> ' raise' clinches 
the connection with Tt'XXo> 'rise,' already advanced by Meister, 
Gr. Dial. II, p. 215, on the basis of the common meanings 'set, 
place, despatch.' «V«t«XX<b 'enjoin, command,' Horn., and eWreXXoi, 
same meanings, Aesch., Soph., Eur. and Thuc, is proof enough 
of this equation. 

A very common use of ot«'AX<b is in the sense 'rig out a ship.' 
As a matter of definition, or«XXo> might be turned by 'launch,' i. e. 
' make a ship swim on the water,' causative, as it were, to Bohtl.'s 
definition of hj tar, 2) 'float on the surface, swim.' It is simpler, 
however, to recognize that in the act of launching a small boat 
there is as much lifting as dragging. The sense 'launch' fits well 
/3 287, where Athena says she is one or to« vija Boijv trriKia koi ap 
tyo/wi airos ' to launch you a fleet ship, and go with you myself.' 
Later, describing other details, she says, vs. 295, &k<i 8' «$07rXt'o-o-<n<T« 
ivrjiroptv (Ipe'i tt6vto> 'and we will fit her out, and launch her in the 
broad sea'.' Here ivrjaopnv, as well as i(f>oir\ia-a-avTn, may be 
regarded as an expansion of o-reXe<» in vs. 287. In | 247 Ulysses, 
speaking of his journey to Egypt, adds : vijas ev oTtlXavra <riv 
dvTidiots irdpouriv ' featly launching my ships with the help of,' etc., 
and in the next line : iwea vijas arftKa, 6oS>s 8' ('o-ayclpcTo Xaor ' nine 
ships I launched, and the folk came trooping together' — a long 
feast and sacrifice follows, and then the embarkation. If oretKa 

1 Greek and Latin have no real living causative conjugation ; any verb is 
liable to have intransitive and transitive, i. e. causative, force. In Sk. ytar 
has causative force without the causative sign -dya- ; cf. Boht. P. W., s. v. 
ytar, 9). 



466 AMERICAN JOURNAL OF PHILOLOGY. 

does not signify 'launch,' then there is no mention of launching.' 
Against the interpretation 'launch,' the nine days' duration of the 
festivities speaks. 

The comparison of rkaw 'lift, bear' (in a transferred sense also) 
with tAAoj, tollo is not new. See Fick's Wbrterbuch, s. v. tela-. 

The following are coincidences of meaning between »J tar and 
tollo: Bohtl. P. W., s. v. tar, 6) 'get possession of, overpower 
(enemies)': tollo 'make away with, destroy'; V tar, 9) 'carry one 
over or through': tollo 'take up a child (to save it alive).' 
Suscipio may be regarded as a translation of tollo 'save,' to suit 
the technicalities of Roman family life. The custom of saving or 
destroying infants (by exposing) was Indo-European ; cf. Zimmer, 
Altind. Leben, p. 319 ff. 

It is perhaps not going too far afield to compare tollere diem 
("Cic. Leg. 3. 18. 40) 'to consume the day (in speaking)' with 
RV. 5. 45. 11 tarema gatam himali 'may we complete a hundred 
winters.' 

By accepting the equation of cti-AX<» with tollo we are enabled 
to explain the form su-stulit. Here we have a reduplication as in 
7-o-Ti)fu, but the vowel has been affected by the root-vowel su-stulit 
<l*se-stiilit; cf. spo-pondi <C*spe-pondi. 1 On the other hand, 
Lat. su-slulit may reflect the original type, and we may regard 
*spe-spond-i as refreshed out of an original type *se-spondi or 
*pe-spondi. Ab-slulit, ab-scido, O. E. scidan. Germ, scheit 'cut,' 
ab-stergo 'wipe off': strigilis, <rr\eyyls 'scraper,' furnish a starting- 
point for abs- before other verbs that never possessed an s initial. 

Su<.b>ldlus is a contamination of *slldius and su-stulit, and a 
popular etymology made it sublalus. 

Regarding the semasiological similarity between Sk. tarati, Gr. 
(o-)re'XXw (in some of their senses) and Lat. tollo as sufficiently 
established, let us now examine the phonetic processes involved 
in comparing them. 

I begin With Te'XXw = 7r/Xey/ai, or rather nepiTcWoixivmv = VXo/jeVa)!'. 

The explanation of Collitz, out of an original velar, is entirely 
satisfactory from the phonetic point of view, but Sk. V car 
signified a leisurely, wandering, horizontal motion, whence its 
application to the grazing of cattle. In Gr. @ovko\os 'cow-herd,' 
m;roXof 'goat-herd,' Lat. o-pilio 'shepherd' < "ovi-pilio (with a p 
due to Oscan influence; cf. Brug. I, p. 321), we see a derivative 
noun meaning 'pasturer.' In agricola 'tiller of the soil' we must 

1 Cf., however, infra p. 479. 



STUDIES IN ETYMOLOGY. 467 

recognize a civilization advanced a trifle beyond the nomadic 
stage. Gr.woXeva 'turn up land with a plough, plough,' wo\e'a, 
ditto, and w6\os 'land turned with a plough' are fossils from the 
same age of civilization; nfkedpov 'a square measure of land' is 
doubtless to be referred here also. Out of such compounds as 
jSouxdW it was easy for Greek to impart the notion of speed into 
our root. The prehistoric *lnwoiroKos was doubtless a horse-rider 
and racer, whence Homeric k«'Xi/s- 'race-horse,' afterwards trans- 
ferred to the other racing sphere, Ke'Xi/r 'light, rapid boat.' Lat. 
callis 'path trodden by cattle, mountain-path'; 'mountain pastur- 
age' is the passing-note for still another possible derivation of 
meaning to celsus 'lofty,' etc. ko\o>v6s 'hili,' Lith. kdlnas, Lat. 
collis, have also, perhaps, reached their meaning by the callis- 
path, if I may be pardoned the pun. The hills were the grazing 
places of the cattle. Sophocles fames KoIWo'r as tvnvirov, tvnaKov, 
Oed. Col. 711. My friend, Dr. Kirby Smith, has called my 
attention to the following passages in Latin lyric. Cuique pecus 
denso pascebant agmine colles, Tib. Eleg. IV 1, 186; jungere et 
in solito pascere monte pecus, ibid. I 2, 72 ; Quid tibi cum speculo 
montana armenta petenti, Ov. A. A. 1, 305. Further passages 
are Ov. Met. 2, 841 ; 3, 408. Similarly, to use a modern instance, 
the -word pasture always implies 'hill, mountain' in Vermont. In 
point is also 'Lebt wohl, ihr Berge, ihr geliebte Triften.' As O. E. 
hyll shows, this sense was already reached in the I. E. time. Cf. 
Sk. cdrana ' pasturing.' 

Sk. cdrana 'path, road' is perhaps to be compared with callis 
< qsl-ni. It is to be remarked that Sk. hj car shows no forms 
with lingual vowel save clrnd (Upan.). Cultus may derive from 
*cdlitns > * coitus, but, in composition, adcu/tus, etc. It is not 
necessary to deny the kinship of Kt\(v8oc, and KfXda 'order' may 
be perhaps connected with 0ouKdW, etc. 

Sk. hj car, however, shows no trace of the meaning 'rise' which 
was claimed above to be the primary signification of tiWus, TtklBa 
and Sk. »/ ir, for carana- 'pillar' derives from the meaning 'goer, 
foot,' and so probably does Lat. columen. I repeat again that the 
characteristic notion of I. E. \j qel is that of leisurely, wandering 
motion. We find a Greek representative of this root in ir\aiaa> 
'wander, roam,' denominative to nXuvtj 'roaming.' 1 wXdi/ij : \J gel 
: : vttvos ; iJ svep. 

1 Tr\avii does not occur in literature till Aesch. and Hdt. irAavdu is an air. 
Xey. in Horn. ♦ 321 — a late book ; cf. J ebb, Horn., p. 124. 



468 AMERICAN JOURNAL OF PHILOLOGY. 

But how are we to mediate between Sk. j^/r and n-e'Xo/wu? A 
Greek tfreX- would in its reduced stage become tX-. But this is 
a group particularly difficult of utterance in initial position. 
Meyer, in his Organs of Speech, p. 326 (Appleton's Science 
Series), characterizes this group as follows: "These (i.e. the 
groups pi, 11, kl and ql) are all formed easily at the commencement 
of words if the mouth is adjusted for the / position before the 
formation of the explosive, so that the liberated air, in passing 
over the dorsum of the tongue, will produce the sound of /. The 
only difficulty lies in 11, for the apex of the tongue, which had 
been removed from the palate for /, has to be instantly replaced 
for the formation of /, and thus a small hiatus can scarcely be 
avoided."... 'This combination is mostly confined to names 
derived from the ancient language of Mexico.' 

Apart from theoretical phonetics, we have the actual practice 
of Latin and Lithuanian, where // is converted into kl in the 
interior of words, and, as I shall hope to show, initially also; cf. 
Brug. Gr. I, pp. 281, 288. 1 Greek r\ao> seems to vouch for the 
Hellenic mastery of tX°. Greek furnishes also a very limited 
numberof suffixes in -t\o-, e.g. avrXos 1) 'bilge-water,' 2) 'bucket' 2 
(cf. avrXlop 'bucket' in Aristophanes), x^Xor 'liquid' (< 'to be 
poured') and c'xctXt) 'handle.' avrXos contains, I believe, the ^ rtX- 
in its suffix, and derives from the meaning 'to be raised up' its 
sense of 'bilge-water,' whence 'hold,' the place of the 'bilge- 
water.' 3 xwXor seems to be a late epic analogon of an-Xos. Pos- 
sibly e'xeVXi) ' plough-handle ' is a combination of derivatives from 
e' x ° 'a handle to hold by' and t«X° 'a handle to lift by.' 4 

The permanence of the initial group in tXoo> is capable of 
explanation even on the theory that the Greek tongue did feel the 
difficulty of the group and avoided it. Beside rXd&> are the forms 
r<i\as 'suffering' > 'wretched' and rdXavrov 'lifting machine' > 
'scales,' which derive from rXX-. Further, erXij and T(rkay.(v were 
susceptible of the syllabication er-Xij, TeV-Xu/^v, thus allowing for 
the unavoidable hiatus of tX" (cf. supra Meyer, 1. a). In tX" forms. 
we might expect a difficulty of articulation. In Latin that diffi- 

1 The apparent exception lattts < tlalus will be discussed below, under 
'Splendidus and its Congeners.' 

"Is 7iE/Ure 'milking-pail, cup' a congener of avrXoc 'bucket' from the yreA 
'lift'? 

3 Cf., however, Brug. Gr. II, p. 113. 

i XvT?Lo; and i^erA? may well proceed from *xv61o-,*ixef>lo. The latter 
almost certainly does. Cf. ykveffkav. 



STUDIES IN ETYMOLOGY. 469 

culty results in cl, but as in Greek X exerts a labializing influence 
on q, it would scarcely exert a palatalizing influence on t. Allit- 
eration would have been a further compelling motive in combina- 
tions of, say, *tXojmkos with eVi and nepi. An analogous change is 
shown by West Germanic and Norse fl <pl < tl, e. g. O. H. G. 
flehan 'fondle, flatter, beseech,' Goth. gapldihan 'fondle, comfort, 
exhort,' Brug. Gr. I, p. 287. 1 

The change of initial il- to cl- for Latin is on the same phonetic 
basis as the change of interior tl. 

Out of a Gr. *tX° > 77X a new series would be evolved : wX", 
■nik-, 7roX- beside *tX°, teX-, toX- ; out of a Latin 11°, in the same 
way, cl", eel-, col- ; e. g. VXopevor : n-Ao/uu : n-oXor. 

Leaving the ground of analogies in other languages, a very 
striking instance of this change in Greek is furnished by 6X" > j3X°, 
in Aeolic /3Xijp : 8i\eap 'bait.' I see no good reason for separating 
SAeap from 86\os 'trick,' but, p. 252, 'bait.' Osc. dolum, Lat. dolus, 
O. Norse tdl belong to the same group. Cf., however, Brug. Gr. I, 
p. 318, who claims a connection with /3dXXo> < I. E. nj gel. Joh. 
Schmidt's comparison with O. H. G. querdar 'bait' (KZ. XXV 
153, but accessible to me only in Kluge's synopsis, Wort., s. v. 
k'oder) involves a dissimilation out of *8epap, *8eperpov, and does 
not seem to me to be forceful enough to separate SeXeap from SdXor. 

I take the following equation to be certain at any rate. Horn. 
/3Xo>0po£ ' tall ' = Sk. dirgh-a ' long ' — reported also to be the name 
of several varieties of trees and grass. $\a>8-po- is < *dl$li-ro-. 
O. Big. dlugu 'long' is from the same stem. Lat. longus < 
*dlongus shows perhaps the stem of the compv. drdghiyahs (cf. 
longius), with infixed nasal. For 8 instead of <f> in [$\a0p6s cf. 
Brug.'s explanation of i\a0p6s (Hesych.), I, p. 320. Of course, 
the connection of SoXi^d? with dirgha is not affected by the further 
association of p\a>6-p6s. Brugmann, however, I, p. 245, compares 
/3Xo>0-pdr with Sk. murdhan 'head,' and Kluge, Wort., s. v. lang, 

'It seems to me possible to equate Goth, "pldihan with O. Bulg. tlFsti 'tap, 
knock' and Gr. -KXlaaopai 'knock with the feet, trot,' vgl. p 318 at <5' ev /lev 
Tpux<-y, ev ch wliaaovTo nodsootv 'And they ran well, and pattered merrily with 
their feet.' My command of Slavic lexical material does not enable me to 
learn whether this explanation is impossible for tleiti or not. The transfer of 
meaning from 'strike, pat, tap' to 'fondle' and 'beseech' is not difficult. Cf. 
'love-licks.' irfo,n6a, oir?.ei<6a 'of sexual intercourse' may be for nWrnoa, 
popularly interpreted in the light of av/zn-leKa ' have sexual intercourse with.' 
O. Bulg. tliika beside tlesti is probably for tllka; cf. Lesk., Hdbch. Abg. 
Sprach., §gu, 2, 5; 19. 



470 AMERICAN JOURNAL OF PHILOLOGY. 

denies the connection of longus with dirghd-. The Lat. lago 'a 
sort of clematis' has perhaps the weak stem = *dl%h-. Cf. above 
what is said of the definitions of dirghd-. 

Under the phonetic conditions above urged, the following 
comparisons are submitted : ni\ayos 'wave' > 'sea' : Sk. taramga 

'wave'; cf. Horn, e 335 vvv 8' akbs iv irehayetiai 8ea>v e£ en/j-ope Tiprjs 

'and now, 'mid the waves o' the sea, the gods yield her honor.' 
Sk. taramga does not occur in any of the accented texts. We 
may, however, infer its accent from patdmga 'bird' ; sj pat 'fly,' 
variously reported as paroxytone and oxytone, paroxytone 
patdmga coinciding with Brugmann's latterly much attacked rule 
of n under the accent (Gr. I, p. 195). In terms of Brugmann's rule, 
■niXayos was an original oxytone, shifted, like neXeKvs and eXvrpov, to 
proparoxytone ; cf. Wheel., Gr. Nom. Ace, p. no. The suffix 
'Vigo- is also preserved in Lith., e. g. vargingas 'miserable' : 
vargas 'misery,' where the accent is paroxytone. Whether my 
comparison of n-Aoyot with taramga should be accepted or not, 
the I. E. character of the suffix -°^go- has, I think, been demon- 
strated (cf. Brug. Gr. II, p. 261). In Lat. prop-inquus the nasal 
vowel is perhaps to be recognized before the suffix -go-. Brug. 
Gr. II, p. 261, rem., suggests a relationship between -%o- and -qo-. 

With this explanation of neXayos is combined a possible one of 
n-Xa'fopu 'wander' < 'to be wave-tost.' Eng. waver : wave shows 
another facet of the same signification. The connection with 
irXdyws will come later into discussion. 

Sk. tirds 'over, past, beyond, through' = Zend taro are 
undoubted congeners of the *j tr. Eng. beyond means 'more 
than, except.' Murray, New Eng. Diet., s. v. beyond, /3 9) puts it 
as follows: "in negative and interrogative sentences almost = 
'except,' e. g. Shaks. Hen. VIII, III i 135 'Bring me a constant 
woman to her husband, One that ne'er dream'd a joy, beyond his 
pleasure.' Carlyle, Sart. Res. II vi ' No prospect of breakfast 
beyond elemental fluid.' " Lat. praeter as adverb and as prepo- 
sition means 'beyond, more than, except.' The various adverbial 
derivatives of the V per, to which praeter belongs, illustrate very 
fully the semasiological developments of the root of which tr is 
the Aryan representative. The V per I define as 'pass by,' in 
which action there are three stages — the motion towards one, the 
motion past or by (before) one, and the motion from or beyond 
one; all the various ramifications of meaning reduce to one of 
these three : Sk. pards, adv. ' beyond, afterwards ' ; prep. ' beyond, 



STUDIES IN ETYMOLOGY. 47 1 

more than, except ' ; purds, adv. ' in front, forwards ' ; prep. ' before ' ; 
pura, adv. 'before, hitherto'; prep, 'before (temporal), before (in 
defence of), except'; pdri, adv. 'round about'; prep, 'opposite, 
beyond (past), more than,' and the grammarians report the 
meaning 'except,' and a distributive force as in %<rk§am vrksani 
pari siiicati ' he sprinkles tree after tree,' cf. Gr. ripipav nap' fipipav 
'day after day' ; pra 'before, forwards, on past'; nipav- ' across, 
over, over against'; nipa 'beyond, across, more than'; napa 
'beside, by, at (e. g. napa Bvpyaiv 'before the door,' nap' oivw 'over 
wine'), along (rptyas nap norapov ' turning along the river'), beyond 

(nap 8ii/aptv 'past his Strength'), 'except' (ovk ?oti napa ravr' a\Xa 

'there is nothing except this'); nepl 'roundabout, around, beyond 
(= more than) (e. g. n*p\ n-oXXoC noulo-dai) ; npcs so nearly covers 
napa that no examples need be given : it may be remarked in 
passing that per vim and npbs fiiav 'by force' are etymological as 
well as syntactical parallels, and so are per Joiiem and npbs Auir, 
in asseveration and ascription of agency ; a by-form of npns is napos 
< °nr p 6s : Lat. por in por-rigo 'put before one,' porrecius 'laid 
out' > 'dead'; porro 'forward, onward'; per 'exceedingly,' e.g. 
permultus, permagnus, perceler; cf. Horn, n 186 nept /«V 8eieiv 
raxyv ' passing swift at running.' Eng. passing is a perfect parallel, 
e. g. "O passing traitor; perjured and unjust," Shak. 3 Hen. VI, 
V i 106; "This Ewein was a passinge faire childe, and bolde and 
hardy," Merlin (E. E. T. S.), II 238; "For she was passing 
weary of his love," M. Arnold, Tristam and Iseulte (cited from 
the Cent. Diet., s. v. passing) ; per-Jidus 'unfaithful' mirrors just 
as accurately napa SUrjv 'contrary to right,' napdvopos 'lawless'; per- 
rexi 'I went forward' ; pro 'before, in front of (for defence)'; 
prae 'before'; for prae" in prae-clarus cf. per-magnus above; 
prae-ter 'past, beyond, more than, except.' Lat. perntx 'nimble, 
fleet' may well be a derivative of the nj per. A stem per no- is 
extended to perniqo- > pernic-. 

Now the Aryan \] ir has had much the same line of develop- 
ment as I. E. per-; tr was employed of vertical motion, per" of 
horizontal, and the former doubtless extended to any motion in a 
vertical or obliquely rising direction We have seen above in the 
semasiology of trans how ' over a mountain ' became ' over a river ' ; 
a bird 'rises, shoots up, shoots through the air, crosses the sky'; 
'over the mountain' was 'out of sight.' Sk. tirds has all these 
significations, 'through, on through, past, beyond, except, cross- 
wise, secretly'; tirds+^hr 'overpass, surpass,' etc. For Greek 



472 AMERICAN JOURNAL OF PHILOLOGY. 

we have one stage of development in W W 'except,' to be explained 
as an ace. adverb from an -<7 stem. Certain it is that n\dv can 
have no phonetic relation with the stem of n-X/Qoic. In n\d-yws 
' cross-wise, deceitful ' we have another shade of meaning. Certain 
elements in the semasiology of n-Xafo/uu might connect it as well 

with nXdyios as with ireXayos. 

For the semasiology of h'rds+ nj kr 'despise, look down upon' 
we may compare Grk. {mtp-fypoviai 'despise.' Eng. over-look has 
over in precisely the same force. 

It has been the fashion, where the Europeo-Armenian group 
shows /and the Aryan group r, to attribute /to the parent-speech, 
I suppose on a sort of democratic plan, but this is, after all, a mere 
convention (cf. Brug. Gr. I, §254). In Sanskrit / gains on r con- 
stantly (Wh. Gr. 2 , §53 b~). Why, then, believe there was an earlier 
Aryan tendency when r gained on /? With the explanation of 
trans as belonging to a Eur.-Arm. V ' /el, we gain a new point of 
view. Let us present to ourselves a state of things in which there 
was an r 2 verging toward /, then a root ter 2 would pass into lei-, 
iol-, but the difficulty of // made the stage tr° a laggard. Any 
isolated form might then loose the bondage of phonetic law ; or 
we might put it that t>\ regularly went into tr°, unless dragged 
by tel-, tol- into //". Now, trans is just such an isolated form, for 
Meister, by his explanation o( rlppa, Gr. Dial. II 213, has deprived 
it of its supposed Greek and Latin congeners. 

Interesting testimony for a serial cl" < 11°, eel is furnished by 
equating celer with lards 'quick.' The suffixes are, it will be 
seen, identical; kAijs 'race-horse,' with which celer is generally 
compared, has been explained above in a different way. 

clam 'secretly' shows a very close kinship of meaning with 
tirds and of form with n\dv : (hand) clam me est ' it is (not) 
unknown to me' is a close parallel to Eng. 'it is beyond me' > 
'past my comprehension.' In combination with V dhd iiras 
signifies 'drive away, conquer.' If we suppose clam to be 
extended by the 5 so common with prepositions and adverbs, e.g. 
4k : f'§, then we may explain chides 'disaster, defeat' as out of 
*clansdi-s. To this formation Sk. furnishes abundant parallels ; 
e.g. anlar 'within ' + dhi < V d/id= 'concealment, disappearance'; 
pari ' about ' + dhi = 'enclosure'; ud 'up' + dlii = 'seat of a 
wagon'; ni ' down ' + dki = 'setting out (down) food.' tiro-dha 
'concealment' < 'setting aside' is a closely allied formation. In 
*clansdi- we have a different meaning, but one very close to liras 



STUDIES IN ETYMOLOGY. 473 

+ v' dha ' conquer.' Indeed, the grammarians report a.*lirohita 
'one who has taken flight,' which presents the same facet of 
meaning as clddes 'rout' 

But in clandestinus 'secret' we have the very force of tiro- 
dha. Clandestinus is composed of clam + a stem -des-. For the 
stage *clandes-, Sk. vayo-dh&s- i) adj. 'health-giving,' 2) nom. 
'strengthening,' purodkas '*one set before' > 'house-priest,' 
*payo-dhas ' water-holder, sea,' reto-dhas 'semen-implanting' are 
sufficient testimony. The next stage in Latin was the addition of 
the /0-suffix, as in mod-es-tus : modo-s, vetus-tus : vetus, st. *vetes-; 
cf. Brug. Gr. II, p. 392. * Clandes- was ' concealment,' *clandeslo-s 
was ' one concealed.' To this the suffix -ino- was added, as divinus 
: divo-s. Libertinus : libertus is a precisely parallel formation. 

The connection is thus broken between clam, celo and oc-culo, 
with their Germanic congeners O. H. G. helan, e. g. Oc-culo 
could be phonetically connected with clam, but a reason for not 
doing so lies in the fact that Sk. tarali never shows a force 
'conceal,' either in or out of composition. 

Pro-cella 'hurricane, onrushing wind' shows the same meaning 
as Sk. tarali ' pressing forwards,' and so does percello, e. g. ventus 
percellit 'the wind rushes past, overpowers.' It is possible to 
connect Lat. celsus 'lofty' with collis, callis, above explained, out 
of I. E. hj qel. It is possible, too, to explain from I. E. V ter % 
'rise,' trans, 'raise.' 

In Lithuanian also the group // became kl (Brug. Gr. I, p. 288), 
and thus I explain kUti ' raise ' and kUta-s ' elevated.' 

Let us turn now to a consideration of the words for 'star,' which 
I believe must be associated with the group above discussed. 
The comparison of n-Xftdfiey 'the pleiades' and "(Hones in sepiem 
triones 'the seven stars, the great Bear' has not heretofore been 
made. Cf. King and Cookson (Sounds and Inflexions, p. 203), 
who compare 'triones with stella. The phonetic question is to 
be solved as for n-XaV, trans, above. nXeidfi« is perhaps an 
extension from an -iien-stem, just as "triones. The -et- is an 
affection of popular etymology, perhaps, from 7rX«a> 'to sail' 
The lengthening in Horn. n-XijiaSes is doubtless due to de Saussure's 
lot rythmique, e. g. o-o^corepoy < *o-o<£oT«pos. Greek d-orep- never 
became *d-<rr<rX-, because it was felt to be an agent noun in -ier-. 
So the retention of r in the Germanic languages is to be explained. 
Latin stel-la may be from *sler-la (cf. agellus to ager-), or it may 
be original. The Armenian is a-st\- where X is a tertium quid, 



474 AMERICAN JOURNAL OF PHILOLOGY. 

neither r nor /. Its phonetic worth is unknown to me (cf. Brug. 
Gr. I, p. 27). In e\n 'stag' : ZXacpos and in aXues 'fox' : Sk&mfe 
this X agrees with Gr. X. The only other occurrences of this X 
cited by Brug. Gr. I, p. 216, are in the combination \b = I. E. 
bhr, where we may explain the affection as due to the labial. By 
my explanation a-st\- falls with Gr. X, not with p. 

In Sk. taras beside sirbhis we have the initial variation as in 

reXXw : ot«XXg>. 

We reach from these comparisons the sense ' riser, mover across 
the sky' as the primary one of the words for 'star.' Sk. tardni 
'sun' has had the same semasiological development as taras 
'stars.' 

By the equations submitted I do not wish to deny all connection 
of (o-)reXX<» with Sk. V car. Certain meanings of the former, e. g. 
'send, despatch,' correspond with Bbht.'s P. W. definition of the 
causative of nj car. 2) 'put in motion,' 5) 'cause one to practise 
something'; with the sense of 'dress' (a-ToXff) we can compare 
colo, which has the same connotation. It is interesting, too, to 
know that Sk. nj car appears in the Maitrayani Sarrihita in the 
form 1) gear-, that is to say, with an initial sibilant. We may 
regard (o-)re'XX&> as containing relics of both the roots gel and ter v 

In Sk. Ir I believe we have also relics of I. E. ter 'penetrate' 
(cf. reperpov 'gimlet,' ropvos 'lathe-chisel,' Lat. terebra 'auger,' Sk. 
tiras 'through') and ter^ 'rise' as discussed above, lurdtt is the 
phonetic representative of trr 2 , and iirdti of trr-. There is, 
doubtless, no trace of this difference of signification in the verb- 
forms, for the notions of motion over (obstacles) and motion 
through (obstacles) enabled the verbs to thoroughly assimilate 
even in the non-transferred meanings. The epic torana 'arch' 
has the sense of 'rise' implicit in it; tula 'scales,' which occurs, 
according to Whitney's Verb Roots, in Brahmana, shows the 
sense 'lift,' causative to 'rise.' The vocalization in tor ana and 
tolayati ' weigh ' is a secondary analogical gradation to tur- < trr^ 
and tul < tur < trr^. 

11. 

Splendidus and its Congeners, with an Explanation of 
Vrddhi in Sanskrit. 

Sk. prathitd, prathas : n\aros : splendidus, splendor : O. Ir. less. 

ir\ados : lalus, planus : Lith. plhlt. 

prthii : wXarvs : Lith. plates ; O. Ir. lethan. 

wfio-nXdrai : lalus ; O. Ir. less : O. Big. plasti, ple'ste. 



STUDIES IN ETYMOLOGY. 475 

The Sk. hj prath means 'broaden,' its ptc. prathita 'broadened, 
wide,' and in a transferred sense 'glorious, famous, splendid.' 
With the latter sense splendidus agrees in its so-called transferred 
meanings. The primary meaning is retained in lat-us 'broad.' 
The transferred meaning of splendidus may, however, be that of 
'shining,' as when we speak of a 'glorious day, sun,' etc. Chro- 
nologically, to judge by the citations in Lewis and Short, the 
sense 'shining' emerges in Latin literature earlier than 'glorious.' 
Perhaps, on this account, we had better regard 'shining' as an 
extension of 'spread out': an 'outspreading' that is a 'brilliant' 
body — the sun, say, like the rayed pictures one makes of the sun. 

The phonetics involved in the equation oiprathitd to splendidus 
is as follows ; The I. E. root was ^r^atk. Lat. splendidus did not 
become *lendidus (cf. lien : mr\r)v), because of its use in compounds, 
e. g. re-splendeo. It must be borne in mind that Sk. nJ prath 
combines very freely with prepositions. In Latin the nasal verb 
system has forced its way even into the ptc, cf. junctum. In the 
present case the nasal was an affix, *spla x nfo- > *spla*ndo, as in 
pando 'open out' : pateo 'be open' < *patno\ cf. Brug. Gr. II, 
p. 152. From *spla"ndo came a participle *spla x nditus, whence, 
by progressive assimilation, *spla x ndidus 1 ; cf. the regressive 
assimilation in coquo < *quequo < *pequo. 

We have now the more difficult question of the vowel to be 
recognized in this root. Bechtel, in his Indoger. Lautlehre, pp. 
242, 244, on the basis of n-X^or 'multitude, extent' and Lith. pllsli 
'make broad' : plaths 'broad,' makes it fall in an e-a series. 
According to the nomenclature of the Brugmann school this a is 3. 
If Bechtel means to compare nXaris directly with Lhh. plattis, as 
he seems to do, he severs the connection with Sk. prthu. The 

1 The large class of Lat. adjectives in idus may have had this origin. Splen- 
didus would have easily influenced candidus 'shining' : candeo, nitidus 'glitter- 
ing' : tritco, rubidus ' reddish ' : ruheo, sordidus ' dirty ' : sordeo, etc. In many of 
these words the -do- < -to- may have proceeded by an independent assimilation 
to a preceding d, as in sordidus, or //, as in pallidas, which latter would be 
dialectic, as in oleo 'smell' : odor 'scent' ; cf. V. Henry's Comp. Gram., p. 65. 
Zumpt meant perhaps to recognize the participial nature of these words in 
-idus by giving them in §176 of his Lat. Gram., along with their corresponding 
verbs. In §249, however, he does not explain himself in this way. I agree 
with V. Henry, Comp. Gram., p. 162, note 3, as to the improbability of a 
connection between this suffixal -do- and the \/do- ' give.' Sk.Jalada 'water- 
giving' is doubtless a popular etymology; cf. Brug. Gr. II, p. 383. 

The purely adjectival sense of a ptc. to a verb meaning 'to be such and 
such' would aid in the transfer of splendidus, etc., to a purely adjective category. 



476 AMERICAN JOURNAL OF PHILOLOGY. 

form nXados labors under the suspicion of being hyperdoric or 
hyperaeolic; cf. Cauer, Delectus, 437, 18. On the Cretan decrees 
granting- rights to the Teians, who were Ionians, the tendency of 
the Teian stone-cutters was to substitute Ionic ij for a. The form 
7rXS0or in such a decree, Cauer 123, 18, resisted the tendency to 
'hyperionism.' The form jrXijflor is, however, amply vouched for 
by Doric and Aeolic inscriptions. On my assumption that nXadus 
is the original, the explanation of n\rj8os is very simple : nothing 
closer in language than the notions of fulness, multitude and 
extent. nKrjdos in Doric and Aeolic was a popular etymology 
with wXij-pijr, where the >j is Indo-European; cf. ple-nus 'full.' 
Now, in Lithuanian pllsti the same association with the V pie 'to 
fill' has been at work. \JA\i. plaths, Sk. praihu-s 'broad' are the 
same formation, prathii-s may have been an affection of prthti 
by prathista. Grk. wharos 'breadth' and Sk. pr&thas we may 
regard as belonging to the normal grade. Further examples of 
the normal grade are Lat. lalus < *splatus (cf. lien : an\rjv) 'side' 
< 'broadside' and Grk. w/ion-XaTat 'shoulder-blades.' Of the 
deflected grade we have examples in n\ad-os 'breadth' > 'multi- 
tude,' latus ' broad ' < *spldtus, a formation entirely analogous to 
Grk. ahis, Lat. sudvis < *suddu-is, and O. Big. plasti < *pldtti 
'mantle, covering for the shoulders' \pleste 'shoulder.' The 
vowel in pleste is derived possibly in the following way: An early 
Slavic gradation a : d became : a ; another gradation was e : o. 
By mediation of o, interchange between a and e was easy ; in this 
way pleste may come from *pldlio-. 

Old Irish preserves this stem very faithfully. Stokes has 
already derived l/ss 'light' < *plent-to and compared splendor, 
BB. 14, p. 313. In *plent-to le is the representation of I. E. /, 
which sometimes appears in Irish in this form ; cf. Brug. Gr. I, 
p. 238. Windisch, in Curt. Gr. Etym., compared O. Ir. lethan 
'broad' with ttKutos, etc. It proceeds from pltnno-, as does prob- 
ably irXdravos 'plane-tree.' Less 'hip, haunch' proceeds from 
piles- ; lat-us in the normal grade has a cognate signification, 'side, 
flank,' represented for O. Ir. by leth 'side, half.' 

Lat. Idtus calls for some especial explanation because of the 
report of Paul, ex Fest., p. 313: stlata, genus navigii latum magis 
quam altum, et a latitudine sic appellatum sed ea consuetudine 
qua stlocum pro locum, et stlitem pro litem dicebant. We know 
that the ships of war were long and narrow for speed. Juvenal's 
stlataria purpura 'imported dye,' i. e. 'costly' (cf. McKinley 



STUDIES IN ETYMOLOGY. 477 

Bill?), suggests that the stlata was the ship of commerce, adapted 
to bearing loads. This stlata may very well be from the same 
root as tXijtoj. I give it an active meaning, 'bearing,' which suits 
very well the kind of ship indicated. The active use of the suffix 
-to-, though not common in Greek and Latin, need not surprise 
one. It is quite common in the Avesta as a suffix of agency ; cf. 
e. g. Vendidad, II 7-10: vlsanhi me yima srira, [vivanhana] 
m?rjlo bsr3taca daevayai ' come unto me, famous Yima, thou 
learner and upholder of the faith.' tX»jto? 'enduring, patient' is a 
Greek case directly in point. The three stages of the treatment 
of stl" in Latin are mirrored by sills > sits (twice on inscriptions) 
> Us. 

Tlatie, Umbr. gen. sg. to the proper name Lat. Latium, has 
been formerly connected with latus 'broad,' so Brugmann, in his 
Grundriss, I, p. 281. The connection with rXtp-os is quite proper, 
but, as we have argued, latus comes from *spldtus. For the 
sense of Tlatie Biicheler, Umbrica, p. 114, compares Tt\ea<f>6pos 
'fruit-bearing.' I would so explain Tlatie, Latium as the 'bear- 
ing, fertile land,' not the 'broad land.' Roman popular etymology 
had doubtless established a connection with latus ' broad.' 

Lat. planus I also connect with the ij ^ath < pldtno-. The 
treatment of the group °tn° in Latin is not a little difficult ; cf. 
Feist, Got. Etymologie, s. v. apn. Lat. anno- is derived by 
Brugmann, Gr. II, p. 137, from at-sno, cf. penna : Old Latin pesna 
<petsna-. But Festus, as cited in Lewis and Short, gives petna 
equally as an old form. Who shall say what is the relation 
between petna and pesna? The most natural development oi°t?i° 
would be nn. But we can operate on pldt-sno-, whence, seeing 
the vowel is long, we would have only a single n, as in mist < 
*tnitsi beside missus < *mit'lo-. aenus < aes-no might lead one 
to expect pena < pesna <pet-sna, but the chronology can doubt- 
less be suitably arranged. Prdnus < *prdd-no- (?) and ra-men- 
tum < *rddmento- shed light on planus < *pldt-no. 

I draw attention to the value of this etymology for gradation. 
The number of examples in the a : d row is not very great. In 
Greek Sya 'lead, drive' : Kwayos 'huntsman' : S-yixos 'furrow' with 
prothetic o : &dos < *&aPos 'fire-brand' : Se'firje < *6'«Sa.£e 'it burns' : 
Sir) 'misery.' These are about the only examples where Greek 
shows all three grades. Between Greek and Latin all the grades 
can here be made out: latus 'side, flank,' a^onXarm 'shoulder- 
blades' : talus 'broad,' nXados 'breadth' ; nXaris 'broad.' 



47§ AMERICAN JOURNAL OF PHILOLOGY. 

It remains to point out, in this connection, a possible explanation 
of vrddhi in Sanskrit, where d represents the o of the European 
languages. I do not agree with Brugmann in explaining Sk. a in 
open syllables as the representative of I. E. o. Let us assume for 
the proto-Aryan period a series an, an, n, n being symbolic of 
zero, z, u the liquids and nasals, and beside that a series en, on, n 
> an, an, n where the normal and deflected grades reach the 
same value. It is obvious that the two series become identical in 
the normal and weak grades : what easier, then, than assimilation 
between the deflected grades? We have, for example, to the 
tjbliaj, bhdjati in the normal grade, babhaja, abhdkslt, abhak, 
bhajayati in the deflected grade. Influenced by such forms we 
have babhdra, abhar§it, abhdr, bhdrayali. Sometimes the influence 
of the e-o series was predominant. Sk. *j prath has no forms 
where we should expect prdth", save the caus. pr athayati. This 
may be explained from the prevalence of the middle voice in this 
verb, where we have always a weak stem ; cf. Whit., Verb Roots, 
s. v. s] prath. The Sk. roots of the d-d series seem all to have 
ended in a single consonant, or with a semi-vowel (j, u, etc.), as 
the examples in Hiibschmann, Indog. Vokalsystem, show. Hence 
it is that the assimilation did not take place in closed syllables. 

The existence of vrddhi in the Europeo- Armenian period has 
been deemed possible on the basis of 'text, rexi, text, O. Big. riesu, 
rcchu, pogresu, etc., Brug. Gr. I, p. 256 ; Bechtel, Indog. Laut- 
lehre, p. 157. There seems to me no cogency in this opinion: 
the Lat. forms are more than easily explained as of secondary 
origin. Egl < *eagi, sedl < *sesdi are lengthenings of an organic 
nature. A very large proportion of perfects in "si were from roots 
with long vowels (diphthongs), e. g. dixi, diixi,fixi,frixi. There 
was every enticement to lengthening "lexi, etc. The simplex legi 
beside "lexi makes it quite likely that text, etc., are syncretic 
formations from *tegi + *iexi. It is noteworthy that the forms in 
question are confined to stems in g. The popular etymology of 
°lexi was doubtless */eg-2i, for which °lexl may have been the 
orthographic representation. "Lexi shows a syntactic contamina- 
tion of pf. *legi, aor. *lexi, as well as a morphological, which is 
perhaps a way of accounting for 'Pure' and 'Aorist' Perfects. 
Surely no one dreams of interpreting the quantity in tectum, etc., 
as original. 

The O. Big. forms are likewise possible of explanation without 
the resort to vrddhi. In boda : basu the a-d relation obtains ; in 



STUDIES IN ETYMOLOGY. 479 

cita : lisu we have i and ei ; in vriiza < *virzq : vresii < *versu 
we have r and er. The transition from these roots with r to roots 



without r was perhaps made through rekq. The impv. (opt.) 
rtci < *rqois stands in the same relation to aor. rex& as zipi to 
zpkxu, however the vocalism of the pres. reka is to be explained. 
The extension was now become easy reka : re\& (< *erxu?~) :: 
nesa : nesu. 

Again, the relation of e-e may have been patterned on the 
proto-Slavic d-d. 

It has been seen, then, that Latin and Slavic aorists give no 
help for a belief in European vrddhi. The 3d sg. pf. act., where 
in Sanskrit vrddhi is at home, shows for Greek always the 
deflected, not a lengthened stage, save in, so far as I know, the 
example yeyave 'is capable of being perceived, heard' — surely an 
insufficient evidence. 

The Indian grammarians 1 report that the 2d sg. perf. was liable 
to accentuation on any of its syllables, and forms like daditha and 
tenitha, not in the earlier language, amply support this view to 
the believer in the origin of gradation from musical accent. 
Where there is any gradation as between the persons of the sg., 
the 2d person is weak. Perhaps the grammarians attributed 
accentual variations between the 1st and 3d persons to the 2d. 
Sk. babhaj'a (3d sg.) < *bdbhdja, Gr. 8«'Sa(.P)«, but for the earlier 
language, in the 1st pers. almost exclusively babhdja, etc. The 
I. E. speech certainly had strong 1st and 3d persons for the non- 
thematic present system, and their accent was on the root. What 
wonder that this accentual relation stamped itself upon the perfects 
also ! The primordiality of Sk. accent is certainly open to 
suspicion under conditions where analogy was sure to produce 
assimilation. 

We may represent the original conditions to ourselves as 
follows : A. In the d-d series, 1st pers. *b(d)bhdgm, Sk. babhdja ; 
3d pers. *bdbhdge, Sk. babhaja ; the accent of the 1st pers. pre- 
vailed ; the reduplication and ultimately the vocalization of the 
3d. B. In the e-o series, 1st pers. t(e)-tepm, Sk. tatdpa ; 3d pers. 
*tdtope, Sk. tatdpa, cf. Grk. reVoiee ; 1st pers. *r i i-r 1 e'iqtg, Sk. ri- 
ric-a ; 3d pers. *r % eiroiqe, Sk. ririca (with reduplication affected 
by the 1st pers.), Grk. XAowre (with reduplication generalized 
from the t«Vo« type). A'. In the d-d series, 1st pers. i(d)-idgm, 
Sk. *iydja (to be inferred from titydjd) ; 3d pers. *idQ)idge, Sk. 

'Whit. 2 , p. 283 fg. 



48O AMERICAN JOURNAL OF PHILOLOGY. 

*yayaja (to be inferred from vavdca, R. V., vavdpa, vavdka, epic) ; 
3d sg. middle *ia((a)gai — Sk. yeje; 1st plur. r\eirj.qmh, Sk. 
riricmd, Grk. *5<=§tfi<Fi/ > hZ&ipev. B'. In the £-0 series, 1st pers. 
*u(e)ueqni — Sk. uvdca; 3d pers. *ue'uoqe, Sk. vavdca (R.V.), 
but usually uvaca, where the 1st person reduplication prevails. 
C. In regard of the ij vid, it may be remarked that elb&s, dhivat, 
el8S> may be interpreted as formed from a 1st pers. *Peiba before 
its assimilation to Foihe. D. As a corollary to the above explana- 
tion of yeje, we may set up an explanation for Sk. perfects of the 
petiis type (Wh. 2 794 g) ; sediis < *sezdiis or *szdiis was another 
starting-point, papata -.petiis :: sasada : sediis :: *ydydja :yeje. 

In Grk. and Sk., as in the parent speech, the reduplicating 
syllable was felt to give the temporal distinction, and was general- 
ized, and so the deflected grade was extended from the 3d pers. 
throughout the sing. But in v/da : Foida, which had reached a 
present signification in the parent speech, the reduplication was 
lost. In Goth, the vowel-change was felt to give sufficient 
temporal distinction, and the reduplication was, for the most part, 
lost. 

in. 

Grk. n4p6a ' sack, destroy, kill ' : Lat. perdo ' destroy ' : Sk. 
nj sprdh ' strive in rivalry, contend, fight.' 

Pott, KZ. 26, 174, suggested that n-e'p&o was a combination of 
the root of ri%u and an unknown preposition, and compared Lat. 
perdo. 

The agreement of the words under discussion is absolute from 
the phonetic point of view as soon as we recognize the initial 
variants ^. 

The Sk. loc. plur. pr tsii is, I take it, an isolated form from the 
»J *rdh. The root-noun sprdh- 1) 'battle,' 2) 'enemy' has in 
R. V. the forms sprdh-i (loc. sg.), sprdh-ds (ace. plur.), spfdhas 
(nom. plur.) and sprdkam (gen. plur.) Out of the isolated loc. 
plur./r/-sz/ < *{s)prdh-sii a stem/r/was extracted, whence prtana 
'battle.' This must have happened in the Aryan period, for we 
have Zend pesana < *prtand. In Zend verb-forms were also 
constructed on this stem, for which Justi sets up a Zend V parst, 
which in the weak grade is V pes. The form spared- = Sk. 
spardh- occurs once. 

The senses of Zend \j parat are 1) 'fight,' 2) 'hasten on.' Both 
senses derive from the primary sense ' strive in rivalry ' in the 



STUDIES IN ETYMOLOGY. 48 1 

different spheres of battle and racing. Zend pesana means 
' hostile '. The Grk. and Lat. representatives of this root have 
derived their sense of 'sack, destroy, ruin' along the same lines 
of hostile rivalry. Lat. perdo has been assimilated in inflection 
to abdo, etc, A perdo of a not greatly different sense is, however, 
derivable from the combination of per + do-; cf. what has been 
said above in 1) about per, etc. 

IV. 

vi+ V ' bhr (intensive) '*bear apart' > 'move to and fro, bran- 
dish' : vi-bro 1) trans, 'shake, brandish,' 2) intrans. 'quiver, 
tremble.' 

Vidro has been heretofore connected with Sk. i/vip- 'tremble.' 
No phonetic change of p before r is provable for Latin, however, 
and so the explanation of vibro as denominative < a stem *vipro- 
is untenable. I propose instead a division into a preposition 
vi+br-a-. This -bra- stands in the same relation Xofero as rXa : 
T«'XXa) in Greek. Note, too, that the Latin frequentatives are all 
of the 1st conjugation ; vibro is, to be sure, not from a supine 
stem, as the others. 

For the appearance of the preposition vi in Latin compare Pott's 
explanation of Lat. vito 'shun' <vi+ita 'gone apart,' KZ. 26, 

P- 154-' 

v. 

vi-nc-io ' to bind ' : nec-to ' bind.' 

Nec-to is congener to Sk. V nadh, in some way that does not 
here concern us, gutturalized in Latin. A perf. next in composi- 
tion with vi' would give us vinxi < *vi-nc-si ; cf. reppuli < *repe- 
puli, surpui < *sub-rapui. From vinxi to vincio, vincil is an 
easy step ; cf. spexi and specio, specit. That vincio should then 
inflect after the manner of the 4th conjugation is a question to be 
solved for venio and other underived verbs. The effect of vi° in 
the combination is not easy to see. Perhaps the notion was that 
of binding to an object away from one ; we might compare Eng. 
tie up, which comes to mean ' tie to an elevated object ' ; Sk. vi+ 
V sanj = 'hang up, suspend' and simple tf sanj = 'cause to hang, 
attach, suspend'; cf. Delbriick, Synt. Forsch. v, p. 467. vinxi 
is, then, 'tie up,' i. e. to an object above one or away from one. 

1 This explanation of vlto I reached independently, but am glad to be able 
to cite it, on Pott's authority, in confirmation of my recognition of the prep, vi 
in Latin. 



482 AMERICAN JOURNAL OF PHILOLOGY. 

VI. 

vivo : vic-si, virtus. 

I add to my already printed explanation of these words (Am. 
J. Phil. XIII, p. 226) the following note, an explanation of the 
guttural in O. E. cwicu. This I take to be a contaminated form. 
The reduplicated ptc. *cwecwen%- and *cwiwo-, the congener of 
Lat. vivos, Goth, quius, were coexistent at some period. Now, 
cwicu is the result of a contamination of *cwiwo- and *cwecwen§ 
> *cwicen$ under the influence of the weak stem *cwecuiv&. 

Austin, Texas. EDWIN W. FaY.