BULLETIN OF THE UNIVERSITY OF TEXAS NUMBER Q7 REPRINT SERIES NO. 3 NOVEMBER 15, 1907 Studies of Sanskrit Words From the Jour. of the Amer. Oriental Soc., 1906. p. 402-417. Latin Word Studies From the Trans. of the Amer. Philolo11ical Ass'n, v. 37, p. 5·24. Greek and Latin Etymologies From the Amer. Jour. of Philology, v. 27, p. 306-317. By EDWIN W. FAY Profe•sor of Latin in the University of Texas Published by the University of Texas semi-monthly. Er;tered as second-clriss mail matter at the post office at Austin AUSTIN, TEXAS PUBLICATIONS OF THE UNIVERSITY OF TEXAS BOARD OF EDITORS WILLIAM JAMES BATTLE, Editor-in-Chief PHINEAS L. WINDSOR, Secretary and Manager KILLIS CAMPBELL, The University Record WILLIAM SP_ENCER CARTER, Galveston, Medical Series LINDLEY M. KEASBEY, Humanistic Series THOMAS H. MONTGOMERY, JR., Scientific Series PmNEAS L. WINDSOR, General Series The publications of the University of Texas are issued twice a month. For postal purposes they are numbered consecutively as Bulletins without regard to the arrangement in series. With the exception of the Special Numbers any Bulletin will be sent to citizens of Texas free on request. Communications from other institutions in reference to exchange of publications should be addressed to the University of Texas Library. EDWIN W. FAY Studies of Sanskrit Words From the Journal of the American Oriental Society, Vol. xxvii, 1906, pp. 402-417. Studies ef Sanskrit Words.-By Enw1N W. FAY, Professor in the University of Texas, Austin, Texas. I. Arv§.fic­ br his German etymological dictionary (English translation, s. v. Erde) Kluge, with some hesitation, connects Lat. arvum 'field, ploughed land' with Gr. lpat£ 'earthwards.' In his Eng­lish Etymology, s. v. earth, all this hesitation disappears, in view, I take it, of ON. jgrvi 'ground,' not cited in the former work among the cognates. It is also to be noted that V ergil (Aen. 12,681) uses arvis as a virtual rendering of Gr. xap.01;,£ 'to the ground.' There is a very obvious difficulty in the comparison of arvum with lpat£, viz., the vocalism. This difficulty may be sur­mounted, to the best of my knowledge, in but one way, by assuming a base er-'arare,' to which the European ·base ar­(from ar-) would belong. This solution I have offered in my essay entitled A Semantic Study of the Indo-Iranian Nasal Verbs (Am. Jr. Phil., 26,389). There is, it must be admitted, not very much to justify a base er-. Skr. §.ra 'awl': Gr. &p{'> 'auger' is dubious because of OHG. ala. Latin ora 'edge, point,' is also ambiguous, and so is er 'stachel-schwein,' which may well be for 'her', if Hesychian X~P is genuine Greek, with a genuine X· Not very conclusive is Skr. frii:iam (?from ar-) '' bruchiges, unfruchtbares land." A vague connection with the root might be made out for Gr. lpicpo<; 'kid' (if=" pricket "): Lat. aries 'ram' (cf. Umbr. erietu): J.,ith. eras (withe) 'lamb.' But in any case, the vowel relations of arvum and lpat£ are no harder to admit than those of Lat. aries: Umbr. erietu. As a means of testing the correlation of arvum with lpat£ I have made a study of the Sanskrit word arv§.iic-, as it appears in the Rig-Veda, and I attach some importance to the isolation of its formation, with the adverbial suffix -afic-, as compared with the isolated Greek word lpat£. I conclude from the data that follow that arv§.nc-and the adverb arv§.k do mean "earth­ward(s)." Vol. xxvii.] St-udies of Sanskrit Wm·ds. 403 1. The Petersburg Lexica define our word as follows: a) herwaerts kommend, hergekehrt, zugekehrt, entgegen-kom­mend: h) diesseitig (only in grammatical and lexicographical sources, save the adverb arvak, which has literary attestation): c) unterhalb befindlich, nach unten gerichtet: d) in der nahe von (advb. only). The earliest citations for the sense c) are from AV. and SB.' and consist of passages in which arvaiic­is contrasted with urdhvas. This contrast I would, however, see in RV. vii. 78. lbc (B2) 1 : urdhva asya aiijayo vi srayante, u~o arvaca b~·hata rathena-vamam asmabhyam vak~i, "On high her rays are spreading ahroad, U ~as earthward in thy mighty car wealth to us fetch." This hymn may helong to a late middle period of RV. tradi­tion as Arnol~l believes (see the table in P AOS. xviii. 353, and KZ. xxxiv. 341), but the chronology of the Vedic hymns can hardly tell against a word proved to be proethnic; and a hymn of the earliest period, if liable to popular rather than hieratic use, may have retained in all its working-over very early linguistic material (cf. Bloomfield, Proceedings, xxi. 41 ff. pace Arnold, ibid. xxii. 309 ff.). 2. Contexts are not wanting in RV. where a special sense seems warranted for arvaiic-. Thus in i. 92. 16 (A), -vartir asmad2 a . . . arvag ratham . ni yachatam ''unto our house-earthward-your chariot bring", ar·dg might well be taken as a mere adverbial repetition of vartis, and passages are still to be pointed. out in which arvak seems to mean ''ad nostrum fundum." In some of these con­texts previous translators have recognized the sense I would give to arvafic-, and Sayai:ia glosses arvak in v. 45. 10 (see 5 below) by avaiimukhaJ:i 'face-downwards.' 1 After each hymn I put Arnold's indication of its age: A, archaic: B., early middle period: B,, late middle period: C,, early late period: C,, later late period. ' This takes asmad as a poss. adj. == 'nostrum', identical with the compounding stem asmad. [1906. 404 E. W. Fay, viii. 14. 8 (B2): ud g[ ajad aiigirobhya avi~ k~l,lVan guha satfl;t arv[iicam nunude valam, '' Showing the hidden, he drove forth the cows for the Aiigirases, And Vala he cast headlong down" (so Griffith). That arv[ficam nunude means 'struck to the ground' =('to his feet,' cf. RV. i. 32. 8, B2) seems to me most probable. Note the combination of the root nud with urdhvam 'upwards' in i. 85. 10 (B2); i. 88. 4 (A). Wilson, cited by Griffith, paraphrased x. 71. 9a (C,), ime ye n[rv[ii na paras' caranti, by "Those who do not walk (with the Brahmans) in this lower world, nor (with the gods) in the upper world.' 3. In the following passages arv[fic-is combined with the root sad ' to sit.' iii. 4. Bed (B2) : sarasvati sarasvatebhir arv[k ... barhir Mam sadantu, "May Sarasvati and her confluent rivers earthward (come) ... and sit down upon this grass." x. 15. 4a (C,): barhi~adal;t pitara uty arv[g, "Grass-sitting Manes come earthward to our aid (or with aid)." 4. In the following the contrast of earth and heaven is clear. i. 45. lOab (A): arv[iicam daivyam janam agne yak~va, "Fetch-by-sacrifice to earth the celestial kind, 0 Agni." v. 83. 6cd (C,) arv[ii etena stanayitnunehy ap6 ni~ificann asural;t pit[ nal;t. "Earthward in this thunder come, dripping water, god our father." vi. 19. 9cd (B2): [ visvato abhi sam etv arv[ii indra dyumnam svarvad dhehy asme, "From every side let him come earthward: Indra, bestow upon us heavenly glory." 1 On paras, see 6 below. Vol. xxvii.] Studies of Sansk1·it W01·ds. vii. 83. 3 (B,): sam Lhiimya anta dhvasir~ ad!k~ata indravarui:ia divi gh6~a ~ruhat asthur jananam upa m~m aratayo arv~g avasa havanasrut~ gatam. "The earth-ends were seen dust-bewhirled; Indra-Varuna, my cry mounted to heaven: My enemies among the peoples encompassed me: Earthward with help, hearing my cry, ye came." x. 83. 6a-c (C,): upa mehy arv~ii-manyo vajrinn­" Come earthward to me, lightning wielder, Manyu." 5. In the large majority of instances arv~fic-is of vague significance. The gods are called upon to come and bless the worshipper. Obviously "come hither" makes a satisfying ren­dering fot arv~ii yahi, but " come down, come earthward" were equally satisfying. Arguing from the antiquity of the agri­cultural rites of worship, attested for instance at Rome in the ritual of the Fratres Arvales, and mindful of the etymological sense of 'plonghland ' found in arvum, one might feel disposed to make something of iv. 57. 6ab (CJ: arv~ci subhage bhava sfte vandamahe tva, "Earthward, gr~cious one, turn thou, Furrow, we greet thee." Still, as so many other deities are summoned earthward, we car. hardly lay much weight on the summoning of the "Furrow" earthward. The following passages, though vague, have seemed, for one reason or another, 1 worthy of consideration. ii. 37. 5ab (B,): arvtficam adya yayyam n!vthai:iam ratbam yufijatbam iha vam vim6canam. "To come earthward to-day your man-bearing Car hitch up; here is your unhitching." Here arvaficarn is not attributive, but predicative, a sort of factitive to yuiijatbam; unless construed closely with yayyam. Chiefly because they show arviific-in combination, not directly with a verb of motion, but with one equal, by a sort of zeugma, to a verb of motion. [1906. 406 E W. Fay, iv. 4. 8a (B,): arcami te sumatiri:J. gh6~y arv[k, "I sing thy grace; sound it to earth." v. 45. lOcd (B;): udn[ na ntvam anayanta dhira as~·i:i.vatfr [po arv[g' ati~than, "Like a ship in water the wise launched him (the sun); The hearkening floods to earth and stood . vii. 18. 3c (B,): arvaci te pathyi raya etu, "Earthward (? ad fundum nostrum) come thy path of wealth, . vii. 28. lb (A): arv[iicas te harayalf santu yuktalf, "Earthward be thy steeds yoked, ." viii. 61 (50). 1 (A): ubhayam s~·i:i.avac ca na indro arv[g idari:J. vacah satr[cya magh~va s6mapitaye dhiy[ savi~~ha [ gamat. "Let Indra hearken earthward to this our. double song; by our unanimous prayer, let Maghavan, the mighty, come hither to drink soma." x. 89. 5d (B,): n[rv[g indrari:J. pratim[nani debhu]f, ''Nor have any counterfeits decoyed Indra to earth." x. 89. 16d (B,): tir6 visvan arcato yahy arv[n, "Past all (other) praisers (?), come down to earth" (ad fundum nostrum: cf. vii. 18. 3, and paragraph 2, above). x. 129. 6c (C,): arv[g dev[ asya visarjanena. "'l'he gods are later than this world's production" (Griffith). No translation of this passage is likely to win conviction, even from its proposer, but I venture on " gods to earth at its creation." 6. I have reserved two passages for separate treatment. i. 164. 19ab (Ci): ye arvtncas t[l:'i u paraca ahur ye paraiicas t[l:'i u arv[ca ahur, 1 Note the gloss of Sayai;ia, mentioned above at the beginning of 2. Vol. xxvii.] Stitdies of Sanskrit Words. "The down they call the up ; The up they call the down ." In this stanza we have some astronomical or cosmogonic riddle, and a real solution I do not pretend to offer,' but it is probable that paraiic-here repeats para-in stanzas 17 and 18, where para-' connotes 'heavenly,' as in the same stanzas avara-, echoed in stanza 19 by arv~iic-, connotes 'earthly.' This interpretation shows points of agreement with Wilson's paraphrase of x. 71. 9a (see in 2 above). viii. 8. 23 (A): trfni pad~ny asvinor avil:i shti guhii paraJ:l kavi ~·tasya patmabhir arv~g jivebhyas pari, '' The three regions of the Asvins are revealed in hiding in-the-far-heaven; The two seers of righteousness wing-their-way earthward unto the living." The translation of paraI:i by "in-the-far-heaven" (=German '' jenseits "), rather than by ''before," seems to me beyond question; and this would seem to fasten the sense of "earth­ ward" upon arv~k. After the above tests of the special applicability of the ren­dering ''earthward," it seems not amiss to regard Skr. arv~iic­as a cognate of Gr. (pat£, Lat. arvis (in Aeneid 12. 681). Postsc1·ipt. The editors have asked me to add, for completeness' sake, a word on (1) arviicina-(arvacina-) and (2) arvavat-. It is habit­ual to render (1)-construed like Lat. snblimis (Gildersleeve­Lodge's Gram. §325. 6)-by 'hitherward' (='to the worship­per, to me'), rather than by 'earthward,' and in all the usage of the word (fifteen cases) there is nothing, as Professor Hopkins observes, to prove the inadequacy of the usual rendering. The 1 But we may note the Vergilian usage, Aen. 6. 481, of superi ='qui in terra (supra terram) sunt,' for the usual superi = 'caelestes.' •See Grassmann's Lexicon, s.vv. para-, parama-. 408 E. W. Fay, (1906. one difficult use is RV. vi. 25. 3 (A, m Arnold's system of dates): indra jamaya uta ye 'jamayo 'rvacinliso vanu~o yuyujre tvam e~am vithurli savansi jahi vf~11yani kri:iuhf paraca~, as to which Grassmann remarks in his Lexicon that here alone arvacina-is used of other than friendly approach. But there need be no question of approach at all, for we may well take jamaya~ and ajamaya~ as adjectives and arvacinlisa~ as a substantive, in formation something like arvales, but in sense like vicini (fini­tumi) or Landsleute. The stanza does not lose in point thereby: Indra, our kindred and non-kindred Neighbours, that as enemies have united,­ Do thou in sunder their mightiness Rive, their prowess; make them as strangers (=drive them afar). In the study of arvavat 'proximity,' the salient fact is that it is never used save as an antonym of paravat 'distance;' but, after a consideration of all the examples in RV., I feel free to say that 'distance' is not the only signification of paravat (and its kin). Grassmann's Lexicon s.vv. para-parama-, suggests 'heaven' as a rendering, and renders parastat by '-oberhalb '; while both Grassmann and the larger Petersburg Lexicon inter­pret tisras paravata~ by 'the three regions' (sky, air, and earth). The connotation of 'sky' or 'air' ('aloft, on high') seems to me probable-what connotation is mathematically demonstrable? -for this group in the following instances. In iv. 26. 6 (B2) the falcon brings the soma-stalk from the paravat (a-b), having taken it div6 amu~mad uttarat 'from yon remote sky' (d), and padas ab, without the interpretative clause d, recur in sub­stance in x. 144. 4 (A). In iv. 21. 3 (A), Indra is besought to come from (1) diva~ 'the sky,' (2) prthivyii~ 'the land,' (3) samudril.d . . puri~at 'the sea-of-air,-i. e:, from the three regions already mentioned: the stanza then adds (4) svar11arat ' from the light-realm ' and ( 5) paravato vii sadanad rtasya; I interpret (4) as a substantial repetition of (1) and in (5) I take sadanad rtasya, which Sayai:ia glosses by meghalokat 'from the cloud-space,' as a synonymous (explanatory) apposition with Vol. xxvii.] Studies of Sanskrit TVoi·ds. paravatal?-(which Ludwig renders by an adjective): thus (5) = 'from paravat, the cloud-space.' In vi. 8. 4 (B,) Matarisvan is said to have brought Agni down from (his hiding place in) paravat, while in x. 187. 5 (B,) Agni's birthplace is given as pare rajasal?-'in the far-off of the air.' In v. 53. 8 the Asvins are summoned from the sky (dival?-), the air (antarik~at), and from here (am~t), and besought not to remain afar, paravatal?­'from (=in) the paravat. Further, note viii. 12. 17 (A): yad vii 8akra paravati samudre adhi mandase asm~kam it sute ral_lii sam indubhiJ?., "Whether, 0 Might, thou joyest in paravat m the sea ' [So Griffith supplies] Delight in our pressing," etc. In addition to these examples of the connotation 'sky' (air) for paravat (piiram), we may note the contrasting pair avara­lower (and) parama-' highest,' especially in i. 164. 17 (C,) aval?­parei)a para en~ 'varel_la'. In the light of such instances we may note t.hat in the remote Celtic branch Ir. eross, which Stokes (Fick's W oert.' II. p. 37) gives as a cognate of Skr. para-, means 'height,' which would tend to vindicate the sense of ' high ' for proethnic, pero-. If paravat means 'sky,' what of its antonym arvavat? Note Vlll. 13. 15 (A): yac chakr~si paravati yad arvfivati v;·trahan yad va samudre andhaso 'vited asi, "Whether, 0 Might, thou art in paravat, or in arviivat, V;·tra-slayer, Or else in the sea , thou art the protector of the Soma-stalk." If we are right in taking samudre of the 'air,' then paravati and arvavati are the sky and earth, respectively.' 1 [Observe, however, that the same words are repeated in viii: 97 (86). 4, but filled out (after the invitation is given) in 5 as follows: yad viisi rocane diva~ Samudrasya' dhi Vii?tapi, yat piirthive sadane Vftrahantama yad antarik~a a g:i.bi, "or if thou art in the sky's brightness, (or) on the sea's expanse, (or) if on earth's seat, (or) if in the interspace (air), do thou come hither," where 'sea' is distinct from air, and earth and sky are separately contrasted. ED.] 410 E W. Fay; (1906. The semantic problem may be stated as follows: para-meant (1) 'distant, far' but came, by a connotation which may have been proethnic, to mean (2) 'high, in the sky;' its antonym, arv~iic-, meant (1) 'earthwards, towards (on) the ground' but developed, under the influence of para (1), the meaning (2) 'near.' The following illustration furnishes an approximate parallel. In Latin, domi (domum) and apud me (ad me) became, in a restricted sense, synonymous: ' at (to) my house.' These syno­nyms must have played a role in the upgrowth of domo doctus for a me doctus and of domi habeo aliquid for mihi est aliquid (cf. Lorenz ad Mil. Glor. 194), wherein the sense of 'domus' has nearly vanished. Similar is the generalisation of Ovpa'E 'out,' French fors /hors ('Lat. foris), from which the sense of 'door' has vanished, almost or wholly; and in French chez the sense of Lat. casa is all but gone; and we no longer think of a hill when we say down or adown. In general, on such prepositional words (direction adverbs) derived from nouns, consult Steinthal­Misteli, Abriss cler Sprachwissenschaft,• II, §4, p. 11 ff., noting especially Skr. parsvam parsve-'adversus, ad, apud, prope.' To say briefly what I think of the morphology of the group under discussion, I explain arv~fic as a terminal accusative *arvam (or plur. *arv~n)+ a deictic particle *-c(a), comparable with Gr. -oE; *-c(a) may be compared with Lat. -ce, and if it belongs to a different guttural series, the reason is that *arvams­has been attracted into a group with the other direction adverbs in -anc-. Alongside of arvacina RV. exhibits a pretty large group of which pracfna and praticina may be taken as representa­tives; arvavat is not to be explained as from arv~iic, but merely as a counterpart of paravat. 2. Nahus-. In RV. viii. 8. 23 (above), the words trii:ii pad~ni call for in­terpretation: what are the three padas? Sayai:ia interpreted them as the three wheels of the Asvins' chariot. Griffith says heaven, firmament, earth. But the hymn itself mentions three places from which the Asvins come, viz., nahus-(stz. 3), antarik8a­(3, 4), dyaus (4, 7). Dyaus we know and antarik~a-we seem to know, but what is nahus? I believe nahus to be 'the night,' Vol. xxvii.] Studies of Sanskrit TV01·ds. cf. Gr. vilxa · vVKTwp, lvvvxo>. This interpretation yields good results when applied to vii. 6. 5bcd (B2): y6 aryapatnir u~asas cak~ra Sa nirudhya nahu~o yahv6 ag11fr visas cakre balih~·ta~ sahobhi~, "Agni made the dawns noble-spoused, Driving off the nights, strong Agni Made the peoples tribute-bringers by his might." Here note the opposition of u~asas and nahusas. The base to which I ascribe Gr. vvxa and nahus is s)n6(w)-gh-, alternating with s)n~(y)-gh-, and refer for my conception of the phonetic problems involved to Am. Jr. Phil. xxv. 371 ff. 379 ff. Stripped of "root-determinatives," the base in simpler form is s)n~(y)-/s)n6(w)-, and meant "to wrap," cf. Skr. snii.­yati 'wraps,' Lat. nuit glossed by 'operuit, texit.' The word nahus belongs more closely with nahyati 'binds, wraps,' while Gr. vi!xa has the vowel color of nuit. Lat. niger 'black' and noegeum 'amiculi genus' attest the -~y-diphthong. In all this it has been assumed that the night was the " binder" or "wrap­ per up" of the day (cf. Am. Jr. Phil. xxv. 386, note 2). The base for "snow," with a different final guttural, s)ne(y)-gwh-, bas a cognate meaning, snow being conceivable as that which "wraps" (covers) the earth : cf. also A vest. vafra-: ' snow ': the root vap-''to strew, weave." 3. vedhas, 'worshipper, pious; faithful, true.' Uhlenbeck in his etymological lexicon groups together vida­ tham (with deaspiration) 'congregation, assembly,' Yidhati 'worships, honors, dedicates (to a god),' and vedhas as defined above. For none of these words does he suggest further cog­ nates, not even Avestan ones. In view of the uncertainty in some few Sanskrit words, even the oldest (cf. Whitney, Ve1·b Roots, sub the root v~·h, and Wackernagel, Altind. Gram., § 161), between band v, we may provisionally etymologize on our words as though they began with b. Then if we set down *bedhas-'fidns, pins,' it becomes immediately apparent that *bedhas and fidns are etymological cognates, which differ only in their stems, the former being an -es-/-os-stem, the latter an -e-/-o-stem. However, it must be observed that in AV. xix. 3. 4 the stem vedha is found in a 412 E. W. Fay, (1906. variant reading for vedya, while in old Latin fidusta (from *fidos-to-) occurs, defined by Paulus as "a fide denominata, ea quae maximae fidei erant," a definition that would lead us to infer an Italic stem *fidos-: cf. also foedus and confoedusti. The derivation of vedhas here suggested also accounts for vidatham, if etymologically defined by ''federation." But vidhati presents a harder problem. It would not be very well defined by -rr£{(fo, but is fairly well matched by Germ. betet, beten and bitten, being, according to many,' cognate with -rrd6u, fidit. But if Kern is right in referring these German words to Skr. b~dhate 'premit' (cf. the citation of the footnote), then it might be necessary provisionally to separate vidhati from vedhas, and rather put it in a group with b~dhate. I have tried, however, in Am. Jr. Phil. xxvi. 179 ff., to reunite -rr£{6n and b~dhate under the still remoter base bh~(y)-d(h) 'to split> plvM (Ovµ.ov) nv[ (nvo>) by 'to open one's mind, convince,' comparing our own idiom ''to open one's eyes;" fidit and 7rl7rot8£ may be rendered by 'opens (one's own heart) to (another), trusts.' The concrete sense of 'splits' is perhaps retained in Iliad 15, 26 7r£11't8ovua Ovl>J,.os, which may be defined by ' findens procellas.' The locu­tion µ.tuO together.' The seman­tic opposition of "to split" and "to join" is only apparent, and comparable with the conflict found in the pair sticks 'stecht' and sticks 'steckt' (cf. also stitches 'stecht, stickt ;' and see Kluge's TVoert. s.v. stechen). But in demonstrating a root bheid(h) 'to split,' with the con­notation 'to join,' the last word has not yet been said for foedus 'truce.' Touching foedus, I think of some primitive form of contract by indenture, some breaking of a tessera hospitalis, in which the breaking of the token was the chief symbolic act of 1 I share Uhlenbeck's doubts whether Goth. beidan can be directly connected with Lat. fidit 'trusts.' But in view of MHG. stecken 'to remain fast, stick, bide' we may connect beidan, Eng. bide directly with bheid(h) to split, pierce.' •Cf. Fr. resoudre 'to persuade,' from Lat. resolvere 'to open up'(?). 414 E. W. Fay, [1906. the treaty-making. Th11s do we best account for the idiom 6pKr.a 7rt fissifacere" rather than to "foederis causa sacruficare" No doubt, however, the cutting up of the animal sacrificed for distribution among the compact-makers was a part of the cere­ monial (cf. Aristophanes, Lysis., 192; ? V ergil, Aen. 8. 641). Returning now to vedhas 'fidus,' I conjecture that its ortho­ graphy with v for b was primarily due to the association of forms of b(h)eidh 'to split, pierce' (cf. Goth. beidan, Eng. bide, Gr. 7r£m0ov, 'fidns' has been assimilated to viddhas 'di-visus, apertus ;' though it is of course not to be denied that vedhas, defined by 'apertus, etc.' is susceptible to immediate derivation from vidh­ yati 'peirces.' 4. vadhri. The close kinship of vadhri and its Greek synonym 'I.Opt> ' -roµ.{a>, castratus ' is not to be called in question, despite their phonetic divergence. The phonetic difficulty is resolved by deriving vadhri from the Sk. root vadh 'to beat, slay,' and 'I.Opt> from a base widh-, found in Skr. vidhyati 'pierces,' Lat. di-vidit ' divides," and further attested, I believe, in l (from *widh­tmos, with -tmos from the root tern), '(mare) dividens.' The parallelism of vadhris (*we (widhris)' throws light on the Skr. root vyadh (not attested in RV.), which I take to be a blend of the root vidh (with grade forms in vedh) and the root vadh. 5. sp~·sati 'touches, grasps, feels, besprinkles.' Uhlenbeck (ai. W oert. s. v.) finds no sure cognates for sp~·sati, but mentions the possibility that Goth. faurhts 'fearful' (with­outs-) is identical with the ptc. sp~~ta ' touched, stirred, moved.' [Professor Hopkins calls my attention to the fact that sp~·~ta appears as p~·~ta in RV. i. 98. 2. ] I doubt not, however, that 1 On the close correspondence of Latin and Sanskrit in parts of their vocabulary, see Kretschmer, Einleitung, 125, ff. Vol. xxvii.] Stitdies of Sanskrit -words. 415 Gr. <'terror-> smitten.' Wharton (Etyma Latina) sets down spurcns as a cognate of sp~·sati, and I believe this to be correct. We may again illus­trate by derivatives of tangit, viz., contaminat, contingit 'defiles,' contactus 'defiled; cf. also tangit 'smears.' The German word ferch 'dung' (without s-) perhaps belongs more closely with spurcus. With these we might group Lat. porcus, supposing the pig to have been named (1) 'the dirty creature,' and not (2) 'the rooter-, (porca 'furrow'). Still another possibility for porcns is (3) 'spotted, dark,' cf. Skr. pfsni 'spotted, a cow,' Gr. 7rEpKo<>, 7r£pKYO'> ('spotted), dark,' 7T'EPK"f/ 'perch' (a dark or spotted fish, cf. our fish-name of "spot"). There is no incon­sisten?y between (1) and (3), 'dirty' and 'spotted' being closely related notions, as Lat. maculosns, e. g., shows. And if porcus meant 'rooter' (2), it may still be a cognate of CT7rapcf.CTCTn 'tears, rends.' I see no reason to doubt, either, that p~·8anf, defined by Boehtlingk as "sich anschmiegend, zartlich " (mulcens) belongs with sp~·sati (cf. also Whitney, Roots, Verb-Jiorms, etc, etc., sub p~·s), cf. upa-sp~ati "zartlich beriihrt, liebkost." With this group we may classify the cognates of Lat. procus 'suitor,' precatur 'entreats, 1 presses (a snit, request); lacessit, flagitat.' 6. khudati 'futuit.' Uhlenbeck defines khudati by "stosst hinein (kap~·tham, sapam), and (s)khidati by "reisst, stosst, driickt." No cog­nate of khudati seems to have been pointed out. 2 If the long diphthong gradation -~(y)-/-6(w), already referred to in this paper, is correctly assumed, then khndati and khidati go back to a common root (see Am. Jr. Phil. xxvi. 396). So far as signi­fication goes, khudati would seem but a specialization of khidati, and we might explain its vocalism as something individual, due, to use the metaphor introduced by Bloomfield (IF. iv. 78), to a blend of khidati and its synonym tndati 'stOsst, sticht, stachelt.' 1 Eng. entreats derives from Lat. tractat 'handles;' cf. further, Goth. bidjan: Skr. b~dhate 'premit' (supra, p. 412). •But now cf. Prellwitz, Woert. 2 s.v. Kvcrl7or; I would derive Kvcrllor from khudhtos, Lat. cunnus from khudhnos or khudnos : base khud(h). 416 E. W. Fay, [1906. But the infection of khidati by tudati may as well have begun in the primitive period as in the separate life of Sanskrit. Uhlen­beck remarks s. v. khidati, "verwantschaft mit chinatti is nicht undenkbar." In Latin both (per-)scindere (= chinatti) and per­tundere occur in the special sense of khudati,1 the former in Priap. 15. 5, 54, 77. 13, and the latter in Catullus 32. 11. If scindit and tundit thus cross meanings in Latin, we have some confirma­tion of the supposed association of ideas tliat changed khidati to khudati under the influence of tudati. If Lat. cudit 'strikes, beats' corresponds with Skr. khudati ' stosst hinein,' the recognition of the Italic cognate would for­bid us to regard khudati as khidati inflected by tudati. It would not forbid us to suspect that primitive khudeti is khideti, with the vowel color of tndeti, though we should be bound to admit three roots meaning 'to strike, thrust, pierce, split,' whose weakest forms are; 1) (s)khid', 2 (s)khud, 3) (s)tud, (cf. Uhlenbeck, op. cit., s. v., tomarns). The derivation of (2) from (1), inflected in its vowel color by (3), is purely glottogonic; not in any case a phonetic question, but rather a psychological question. Provisionally, leaving out the possibility that khudati is cognate with Lat. cf1dit ( : Germ. liauen, cf. Brugmann, Grund. 1' § 639), we may include khudati / khidati among cases like those pointed out by Bloomfield in the essay referred to. Ulti­mately, perhaps, a psychological treatment of the vowel alter­nation in the spirit of W undt's Die Spraclie (I', p. 335 ff.) may be arrived at. 7. Skr. ambaram. Uhlenbeck asserts that no satisfactory explanation has been advanced for ambaram, but it seems to me that an easy one lies at hand. The meanings we have to account for are (1) ambitus, vicinia, (2) amictus. It is phonetically allowable to connect amb-with &.p.cpt, Lat. ambi-, cf. Skr.· ambu I ambhas 'water' for the variation b / bh after nasals. By this explanation ambaram (subst.) is morphologically comparable with avara­ 1 Cf. Gr. 1' are the 'expectant flames' unless, instead of jur~1aya]?., we supply samidhal:i 'kindlings, faggots.' As t"o svapaka, Ludwig's 'selbst garer' suggests to me 'self­cooker, self-kindler.' • Extractedfro~ tke Transactions of Ike American Pkilolo~cal Association, Vol. xxxvn, 1907. I. -Latin Word-Studies. BY PROF. EDWIN W. FAY, UNIVERSITY OF TEXAS. (I) accersit or arcessit. WALDE, in his Latein. etym. Woert., s.v. arcessit, avows a preference for Brugmann's (IF. XIII, 88 sq.) derivation from . *aifaccssit 'herbeischafft' as compared with Thurneysen's.; (Archiv, XIII, 36 sq.) from *arvocessit. To the derivation• from •ar'lJocesso Brugmann raises both phonetic and semantic ; objections, waiving the latter, however, in view of Eng. hales.· (= 'drags into court, summons') : Lat. calat 'summons,' Gr. 1'aXe£ 'calls.' In view of calfaccre (Cato, Petronius), from califacere, califaccre, the phonetic difficulties might also seem solvable, -arcesso, from •arucesso, from *arvocasso; but the form -voccsso is itself gratuitous. There are, however, psycho-phonetic difficulties in the reduction of *aifacesso to arcesso. It would seem that com­pound verbs are so liable to "recomposition" that in only four cases 1 have they entirely lost a representative of their root vowel, viz. : in surpere ( : rapcrc; cf. usurpare ?), pergere, porgere, surgere (: regere), in all of which the group, vowel + -rr-+ short vowel, was reduced to vowel + -r-(see Vendry es, lntensitl, p. 26I). A semantic difficulty with the *mfacesso derivation is to account for the change fromfaccssit 'makes 1 The compounds of iacit exhibit -icit, an'1 are on a somewhat different footing. s 6 Edw£n W. Fay. off' (neuter) to arcessz"t ' fetches, brings ' (causative, not merely transitive, pace Brugmann, I.e. p. 94). The most satisfactory derivation, as I see it, starts with accersit and connects with Skr. kdrs,ati 'trahit, arat.' This explanation is older than Pott, who seemed to pooh-pooh it in his Forschzengen, I. 699, but again coquetted with it, £bid. IV, 361. Latterly this etymology has been defended by Nazari, in Rz"v. dz" Filologz"a, XXIX, 269. Nazari's startform is *arcer­sz"t, rightly criticised by Walde, I.e., p. 698. To justify the -rs­of -cers£t Nazari derives from *cerssit (see on the verb suffix -so-Brugmann, KvGr. § 678-679), but we might also operate with a base s)KERAx-s-, whence a present stem s)KER -Es-, reduced in Latin to cers-. True, for the Sanskrit forms the base KERs-(Hirt, Ablaut, § 723), i.e. KER-s-(Walde, s.v. curro writes KERE-s), is sufficient, but this does not assure us that in other linguistic territory a base KERAx-s-may not be found (cf. Reichelt, KZ. XXXIX, 31 sq.). Either of these ways of accounting for -rs-in Latin ought to satisfy, so far as -cersit is concerned, all who do not still yield allegiance to the comparison of Skr. kars,'tts 'furrow, pit' with Gr. Te'A.uov 'turn-row,' an identification now given up by Prellwitz in the second edition of his Woerterbuch. · There is no cogent historical proof that arcessit is an older form than accersit, and the contention that the -r-of arcessit, while it was yielding to the analogical influences of com­pounds in ac-c-, sprang forward and produced accersit seems to me, as to Nazari, very improbable. On the other hand, assuming that accersit was the genuine old form of the word, the reverse process seems to me likely. Pairs like jworS11s /prosus, rursus/rusus attest a period in the history of the Latin language when secondarily derived -rs-threatened to yield to -s-; and periods of fluctuation, of begun but arrested phonetic change, imply a certain popular conscious­ness of the shifting pronunciation. Accordingly, I assume that an individual language-user may have had it upon his mind whether he should pronounce the new (and perhaps slightly "tabooed") *access£t or the old (and "standard") Vol. xxxvii.] Latin Word-Studies. accersi't; the result of his hesitation was arcessit, which he may have reached through *arcersit. The process was psy­cho-phonetic, identical in principle with "anticipation," say vowel-infection in Old Irish, Germanic Umlaut.1 From my own speech experience I can cite an almost per­fectly analogous case: a favorite flower of my boyhood was the yellow jessamine, and about the time I went to college I began to affect "jasmine " for "jessamine," with the result that I often caught myself saying" yallow jasmine." 2 So one halting between "accersit" and "accessit" may have fallen into *arcersit and arcessit. It remains to speak in conclusion of the general semantic problem involved. Skr. kdr~ati 'pulls, drags' has even been separated-in Cappeller's lexicon, for instance -from kdr~ati 'ploughs.' But inasmuch as the same meanings are found associated in Gr. eXK€t 'pulls, draws' : Lat. su/cus 'furrow,' 0 . Eng. su!h 'plough,' this separation must not pass unchal­lenged. The parallelism of the eXK€t-group and the kdr~ati­group is thorough-going, cf. €A.1Coµ,€11oi; (Aristophanes, Nub. 1004) '[ad iudices] accersitus,' eAKW UE (ib. 1218) 'te [ad iudices J accersam.' The question arises whether 'pulls ' or ' ploughs ' is the prior meaning. Easy as it seems to be to derive the sense of 'pulls' from 'ploughs' (i.e. pulls the plough), on the other hand KERS-'to plough' would seem but a simple extension of s)KER-'to pierce, cut, shave, scrape.' Here let us consider Germ. reisst, which offers a curiously exact parallel, for it occurs in the technical language of the farmer in the locution eincn acker reissen = 'wuest gelegenen Boden zum ersten Male pfluegen' (Heyne's Woert. s.v.), and the ordinary meaning is 'mit Gewalt ziehen oder auseinander gehen.' According to Paul, Dmtscltes Woert., the funda­mental sense is ' einen Einschnitt worin machen '; and a further development, traced in detail by Paul, yields 'zerren' ( = 'heftig ziehen '). Again, zerrt is a specialization of zehrt Vowel assimilation in Latin sometimes proceeder', it must be admitted, from beginning to end of a word, but then chiefly had the effect of a resistance to vowel weakening (see Brugmann, KvGr. § 331, B). 2 But the other day, speaking of a 'bang-fringe,' I miscalled it 'bange-fring.' 8 Edwin W. Fay. cf. also Germ. raufen, •zerreisst' (: oepet 'flays, tears'); 1 1'upfen 'to pluck, pull' : Lat. rumpit. In line with the above, we may put e71.Kor; 'wound, sore ' and Skr. srkds ' lance' in the eA.Ket-group~ Here a general word on semantic questions, a propos of Walde's separation of Lat. caed it from scindit (see s. v. ), because the former means ' schlaegt,' and the latter ' spaltet.' True, 'strikes' and ' splits ' may seem very different to us now, but both acts are performed with an axe, and develop alike from a sense ' axes.' The chief difference between caedit and scindit is that scindit is especially set apart for chopping in the long rather than the cross direction; but we may note Ital. largo 'broad'/Span. largo ' long,' specializa­tions of Lat. largus 'copious, abundant.' In questions of semantics a good motto is chercltez le denomi­natif; for whether the verb was denominative to start with (as I often think) or not, the cognate nouns are perpetually engrafting their senses on the verb (Brugmann, Gr. II, § 794, Anm. I). By all means precision in definition, but not a precision too narrow for the facts of usage. As to the notion ' cutting,' two French verbs are of interest, as showing how very restricted the notion may be to start with, and how general­ized it may become. Thus Fr. couper starts with Graeco­Latin colapum ' a slap or blow on the face' ; but now the developed verb has a range indicated by the following defi­nitions : 'to lop, cut, fell, cut out, clip, pare, cut away, Similarly from Lat. talca ' shoot, intercept, divide, chop.' cutting,' comes tail/er' to lop, trim, prune, cut, cut out, carve, hew.' 2 If in neither of these verbs the sense ' to split' is 1 In connection with zerren, the suggestion comes to me that Germ. zer· 'apart, asunder,' comes from D ER· ' to split,' and is comparable with the other preverb los-(see A.JP. XXVI, 173, n. 1). A verb like zerreissen looks very much like a composite of zerren and reissen. The final solution of the problem rests, The English preverb of of course, with the interpretation of O.H .G. zer-. similar m~aning is to-(see Skeat, s.v.), which may be derived from the base DE(Y)-/no(w)-'to split' (see A.JP. ib. 178 n.). 2 KeJler, Latein. Etym., p. l 13, assumes a low Latin verb *taliare, citing Varro's (ap. Nonium, 414, 30) intertaliare. Vol. xxxvii.] Latin Word-Studies. reached, yet it is in briser 'to break,' intransitive 'split,' and we may imagine this intransitive turned causative. The sphere for an interchange of the ideas of caedit and scindit is stone working, -the neolithic stage, to wit, -where chop­ping is as much 'splitting ' as ' striking ' ; cf. the locutions erz /zauen (' einschneidend schlagen ') and erz scheiden (: sdn­dere). The generalized sense of 'divide' arises for most verbs of 'cutting,' and from 'divide ' back to the concrete ' split' is an easy step; thus the substantial identity of secat mare with jindit mare might give to secat a concrete use as ' splits,. cleaves.' English cleaves does in fact mean 'scindit,' and its Greek etymon ryA.ucpEt 'sculpat' may be rendered in Latin by 'caedit, incidit' (cf. particularly caelmn ['engraver'sJchisel'). The specific senses of scindit 'findit' and caedit 'secat' are combined in Lucretius' (i, 533) findi in bina secando. (2) avdry1C17 : neccsse. The old comparison between avdry"11 and necesse has been given up by the latest authorities (see Prellwitz and Walde in their lexica). But neither word enjoys an entirely satisfac­tory definition in its current explanation. Thus we have to regard avdryic17: evEryKEiv 'portare' as a sort of imperious, not vacillating, Fortuna (: ferre 'portare '), -which were all very well if we did not have to account for avdry1C1) 'tie of blood'; ol avary/Catot ' necessarii, affines, connections ' ; 'To avaryica'iov 'prison '; OECTµO<; avaryicaio<; (Theocritus) 'vinculum necessa­rium'; cf. also ~µap avary1Ca'iov (fl. xvi, 836) 'day of enslave­ment, bondage ' ; avaryKaia TVX7J (Sophocles, Ajax 48 5) ' lot of enslavement, bondage.' In all these locutions the notion of ' binding, bondage, constraint ' lies clear. Not but that I think that EIJf"/ICf'ilJ 'portare, is a true cognate of avdry1C1); only 'portare' seems to me a secondary meaning, while the primary meaning, lost in the Greek verb, was ' pangere, nectere.' To the base ENEK-belong the following: ( 1) o'Yicor:;1 'barb of an arrow,' Lat. 1mcus 'hook,' Lith. dnka 'knot, loop, noose' (cf. ory1Co<;2 'knot of hair'), 0.H.G. ango 'point, arrow-point'; ( 2) oyico<;2 'mass, weight,' -£.e. 'pack.' It is from 'pack' that I hope to clear up the definition of ' portare.' The first IO Edwin W Fay. definition of pack in Stormonth's dictionary is 'a bundle or bale tied up for convenient carriage' ; and so the Hatzfeld­Darmesteter-Thomas lexicon defines French paquet by 'as­semblage de plusieurs choses liees, envellopees ensemble.' It is immaterial whether with Koerting (Lat.-roman. Woert.2, s.v. pac-, pag-; cf. also Skeat, Concise Diet., s.v. pack) we ultimately derive paquet, pack from Lat. pango or not, -the point is that a pack is 'a tied bundle,' and that in English packs means 'carries (a bundle), bears.' Note also that Ger­man trag-riemen 'carrying-strap' furnishes a semantic line from Lith. dnka direct to €very1Ce/,v ' portare.' The base I have written as ENEK-does not differ from the base usually written ENEK-(cf. Prellwitz, s.v. €very1Cetv, and Walde, s.v. nancz'scor). The -k-form of root finally got the upper hand for reasons set forth by Hirt in BB. XXIV, 287. The base originally meant 'to pierce, strike, cut,' ­a chain of meanings that may be exemplified by Gr. 71"auua­A.o<; 'nail, peg ' (Lat. pangere), -whence it passed over to the sense of 'iungere, nectere,' which we may exemplify by Skr. pddfa(as 'rope, lash' (for the signification, cf. the author in AJP. XXVI, I 77, L.). The sense of 'bend,' often con­ceived as the primitive of Lat. uncus 'hook,' may have devel­oped secondarily from the sense 'binds' (i'b. 378, T. /3), but in view of O.H.G. ango 'point, arrow-point,' it is not at all unlikely that the sense of 'bent' originated by metaphor from the (barbed) arrow-point (cf. r(A.urp('>). But the sense 'ferre, portare,' derived above from the notion of 'pack,' may have come directly from the sense of 'strike, hit.' An important carrying act of the neolithic man must have been the bringing to his abode of the spoils of the chase; and for heavier burdens he would have employed a pole swung on two shoulders, that pole possibly the shaft or spear with which the game had been slain. With this consideration in mind we may develop the notion 'portare' either from 'nectere' (by means of a trag-riemen 1), or from 1 An intermediate term between 'caedit' and 'fert' would be 'trahit,' perhaps (cf. Span. trae 'portat' = Lat. trahit). This allows the identification of Lat. trahit' zerrt ' with 0. Bulg. truzati' reissen,' Skr. tnie:Y-(cf. Sanskrit dhiydte). Thus ¢&.pm and Lat. /eris, though of a different grade in their "root" part, may be equivalent in their "stem" part. On the eis-and ... of the nasal verbs see AJP. XXV, 387, n. 2. s Let who will think that the neolithic Aryan thought in terms like "Vollen­dung einer nach vorwaerts gerichteten Handlung, Durchdringen durch etwas." 12 Edwin W. Fay. [1906 (that is, strictly, 'consanguinitas, necessitas '). Accordingly, I am convinced that in the locution (mihi) 11ecesse est we must seek for some original sense like that of the English ' I am bound.' 1 Morphologically, necesse is, I take it, a locative plural from a stem neces-, attested for the base ENEK-(in a different sense, to be sure) by Gr. 7r007JV€1Cer; 'foot-reaching ' (cf. 7roo~p1Jr;).2 A comparable formation is the adverb temere (loc. sg., see Walde, s.v.); syntactically, the plural may be illus­trated by compedes, casses (in the literal sense), and by nuptiae, sponsalia, indutiae, vindi"ciae (in a figurative sense). Thus I would understand the locution necesse est hoc facere to adum­brate an original sense like 'it is in the bonds to do this.' It remains to say a word about the derivatives of necesse. Since the serious defence of necessis as a genuine abstract noun by Brugmann, Skutsch, and Zimmermann (references in Walde), it seems necessary to avow the conviction that scholars like Munro were right when they explained the gen. form necessis (Lucretius, vi, 8 I 5) as gen. to necesse, conceived as a neuter adj.; the writer who could say natura ... inanis (i, 363) and plus ... inanis (ib. 365),-inanis being in both passages a gen. of inane, and in both a line end, -may be supposed to have used necessis in magna vis ... necessis (by certain conjecture for. necessest) as a gen. to necesse. But Lucretius, who used momen for momentum, seems to me quite capable of having used neccssis as a short form of necessitas outright; just as Henry Porter, no mean dramatic poet, uses, in The Two Angry Women of Abi11gdon, 'enter­tain' for 'entertainment,' 'attain' for 'attainment,' 'maintain' for 'maintenance,' 'depart' for 'cieparture,' 'persuase' for 'persuasion,' 'suspect' for 'suspicion' ; and he would have had a sort of pattern for necessis in ravi·s ' hoarseness ' ; note also pairs like facul/facultas, volup/voluptas, as (incorrectly) interpreted by Donatus (cited below). The remaining forms to account for are necessum (Plautus 1 The older etymologists were right, in my opinion, when they connected oeZ 'necesse est ' with o, if for uJCipo<>, to the base s)K(H)E(v)-'cae­dere' (see AJP. XXVI, 396, and Walde, s.v. scio). With uJCtppo<> 'hard ' belong uJCtpov ' parings of cheese' (cf. the 18 Edw~·n .W. Fay. sense of ' coating, cover, scum,' in uKtpoc;), uKtpoc; 'copse­land ( = French taille), a stump.' As to the suffix, I assume a verb *severe(? saevere, a doub­let of sacvirc) whence severus, as oKV7Jpdc;: oKV€'i (cf. Brug­mann, Gr. Gram.3 § 202, 3); cf. also avarus: *a11are (doubletof avere, as densare of densere). Inasmuch as Lat. sc11erus forms a conceptual group inmost minds with Gr. av(jT7Jpdr;, a few words over the latterhere. Both words mean ' astringent, bitter,' in a concrete way, and' strict' as moral terms. In the Plato context whereit is defined 'astringent' (Tim. 65 D), it denotes a lesserdegree of astringency than (jTpucf>vdr;; 1 but uTpucf>vJr; is to beconnected with (jTpecf>oi ' turns, twists '; cf. Du. wrang 'acid,sour,' cognate with Eng. wrings 'twists,' "because acidswring the mouth" (quoted from Skeat, op. cit. s.v., wrong); soLucretius (ii, 401 ), speaking of astringent tastes, observes,foedo pertorquent ora sapore. Is there likewise a concrete etymon for au(jT7Jp6<; ? Itwould seem from Skr. nif{ltura-s 'asper' (see asper in Walde,and note Terentian asperum vi11um) that ·(jT7JPO'> and -f{httrasmay derive from a · common source, viz.,-a base s)TER-'topierce,' found in Lat. terit 'rubs,' terebra 'auger' (see Walde,s.1'. stringo). The same base (not a different one, pace Walde,l.c.) means 'to turn, twist,' -a metaphor derived from boringwith an auger; and so ·(jT7Jpo<;, -f{lzuras may be cognate withLat. str-z'c-tus (cf. astrictus 'astringent'): stringit.2 To the base s)TER-'to pierce' belongs stercus 'dung'(named from its pungent odor), sterilis, with a sense ofstrictus ( = compressus, compact, strong; cf. Skr. star-i-s' strength '), and a further derived sense of ' barren ' ( = noncompos Veneris? cf. Skr. star-i-s 'vacca sterilis '); for thesense of 'rough' note Gothic and-staurran 'to be shrewish.' 1 On the v of eni? :::o Edwin W. Fay. Giving to amarus 'bitter' (cf. also Skr. amla-s 'sour') aprimitive sense of 'biting, piercing,' the semantic correlationwith amat 'loves' has an explanation suggested in the fol­lowing: "'Biting' is a well-known gesture of physical pleas­ure (cf., e.g., for the Roman poets, morsus in Pichon'sDe Senn. Amat., s.v.), and plays a role in the sexual lifeof animals" (AJP. XXVI, 201 ; see also Havelock Ellis, p. 107 of the work to be cited, where the horning of does bystags is mentioned). This subject is treated at length by Havelock Ellis in hisStudies in the Psychology of Sex, particularly as regards"Sadism" (p. 88 sq.), or "algolagnia" (p. IOI). The HinduArs Amatoria has a chapter on striking as a love stimulus(ii, 7, 15-16, in Schmidt's Kamasutram); cf. further on thelove-bite, Ellis, I.e., pp. 65, IOI, and also note the entriesproelium and pugna, with their respective verbs, as givenby Pichon in the work already cited; e.g., Propertius, ii, 1,45, angusto versantes proelia lecto; Ovid, Ars Amat. i, 665: pugnabit primo fortassis et "improbe" dicet;pugnando vinci se tamen ilia volet. Resting on this physiological foundation, there is no reasonto reject off-hand the suggestion that amarus 'bitter' may becognate with amat 'loves' (i.e. eparni). Note also Skr. ari-s'cupidus Xhostilis,' and epaTal 'amat' : epl<; 'rixa,' all of whichbelong to the base h:E(Y)-'caedere' (see AJP. XXVI, 389).To this base apoEl and arat, both verbs of sexual connotation(see the lexica), also belong.1 Shakespeare may also be citedfor "He that ears my land spares my team," though this maybe of Hebraic origin; cf., in the Vulgate, Jud. 14, 18, nisiarassetis in Vitula mea : apoTpov 1 plough' Of the organs Ofgeneration.2 1 On the relation of the ideas of' arat' and' coit' ( = amat), se.e Meringer,IF. XVI, 181 sq.,-though his order of semantic development is just the converseof mine (AJP. ib. 407, n. 4). 2 The metaphor of the sexual ploughing is fully developed in Lucretius, iv,1272. For the Plautus examples of arare in this sense (viz. Asin. 874, True. 149, to which we should perhaps add Mere. 356, cited above), see the Thesaurus,II, 627, 55. In The Spanish Curate, II, 3 (p. 245 in the Mermaid text), Beau· Vol. xxxl'ii.J Latin T„ord-Studies. 21 The base to which I assign amarus, amoe1ws, and amat is to be written AME(v)-, and given the primary sense of 'figit­ caedit' (i.e. 'pierces'), and a secondary sense of 'figit-pangit' This is the base I have writ­ (ultimately =vincit, iungit). , ten ME(v)-in AJP. ib. 176, and which is gi\'en as MEYE-by Hirt writes it as EMAx-(Ablaut, 335 ), but Walde, s.v. moenia. this does not account for the -f-of Skr. amzti, nor the ay of The sense of (I) 'caedit, JC67rTet,' with the A vest. amayavii-. natural transfers to the mental and emotional sphere, I find in the following: /1µ17 'mattock,' aµdpa ' pit, ; Skr. 11/a)'ft k !ta-S 'peg, ray,' Germ. ameise = 'ant,' µfrv'A.o<; 'curtus,' Lat. ama­ rus 'bitter,' Siu. amitram 'hostis' (accent not normal for a compound with a-privative, cf. Whitney, Skr. Gram. § 1288, d), Skr. mitlw-s 'false.' Generalized from 'caedit' is (2) 'premit, We have this sense in Avest. ama-'impetus, might' urget.' (morphologically comparable with epo<;/epOJ<;: ERE(v)-), also used of sexual virility (cf. Bartholomae, iVoert., s.z1.). Here In the Rig Veda we also A vest. amayavii-'labor, dolor.' have dmavat-of the Soma press-stones, and contexts like 8, 77 (66), 10, 'Z'araham emu~dm 'verrem urgentem,' 5, 87, 5, vffii 'urgens taurus' (?), 7, 24, 2, (1lathi!ty amitriin dmaviin ­ ab/ii ye no ... amdnti 'caede hastes ( = prementes) qui nos ... comprimunt.' Further differentiated is (3), a sense like 'pangit, vincit.' This sense ·is well attested for sam +v'am-in Sanskrit 1 ; cf. further Avest. miBra-'compact,' µfrpa 'infula, ~wv17,' µfro<> 'thread, string,' (?) Lat. red-imztus 'vinctus,' Skr. mithund­' pair, ~eilryo<> ' ( = iuncti amici), mitd-s 'vinctus.' Again (4), with mercantile sense, -but whether this came or was rather developed as from the sense of ' compact ' mont and Fletcher echo Judges 14, 18, in the line 'Plough with his fine white In view of the sexual ploughing, I see no reason to doubt the connec­ heifer.' tion of rpa-ra< with arat, unless we mean to divorce Cymric erw (see Stokes i,1 Fick's Woert.4 II, p. 41 ) from Lat. arvum. The e-vocalism appears also in th : I agree group of which ipfr11s 'oarsman' may be taken as the representative. with Meringer (IF. XVII, 122) that it is semantically incredible to divorce the 'rowing' from the •ploughing ' group, especially in the light of /Eneid, iv, 399, fronLhe. t eindringlich.' 22 Edwin W. Fay. in Germ. sch!ag (see AJP. ib. 176, n. 3), it were hard to determine -aµm(3~, µotTO') I µo£vo<; 'recompense' (cf. Lat. munus 'present'); aµ€{(3EL, Skr. mdyate, Lat. miUat 'ex­ changes' (?here Lat. emit 'buys'). This notion of exchang­ing may have given rise to the notion of 'friendship,' but Ithink rather that 'amicitia' arose after ' i!pwr; ' (cf. Siu. ari-s,aryd-s, entered by Prellwitz S.1). epaµat), perhaps merely aspairs ( coniuges) grew old. In the order of moral evolution ( r) amz"ca '~ f.pwµev17' (? µotxor; 'paramour') gave way to (2) amica, quasi' comes amata.' The word lust has had inGerman (see Heyne's Woerterbuch) a very similar semanticdevelopment. Our English lust still corresponds, morally, to (I) amica, while Germ. lust is in the stage of (2) amica (cf.Skr. mdyas-'gaudium '). As to the form of amoenus, the diphthong in the post­tonic syllable causes difficulty; so does the tonic oe ofmoenia, in which Sommer recognizes (Lat. Gram., p. 89) apsycho-phonetic influence, and the post-tonic oe of oboedz"t (see Havet, Mem. Sue. Ling. IV, 4ro); cf. also lagoena'flagon,' borrowed from A.dryuvor; (lagoena 'wine-flagon' hasperhaps been affected by a popular connection with oenopo­limn 'wine-shop'). But amocnus need not be explained asan immediate counterpart of primitive AMOINOS. I suggest,in view of Avest. amaym:ii-( = Skr. dm'ivii-), cited above, thatwe put as the Lat. startform *dmoye(s)nos > dmbenos > dmoe­nus, but aµHVO-would suggest a different startform (cf. Prell­witz, s.v. aµ€{vwv). It remains to add a few words on other Latin cognates,or possible cognates, of this group. The gloss amoenavit'densavit' might be interpreted, not as a denominative toamoenus, but directly as a verb of nasal flexion, from thebase AME(v)-'premit.' Foremit(if from 1tlm-) 'buys' the older meaningwas 'accipit'(so Paulus-Festus); cf. the compounds adimit 'ad se accipit,' smnit 'takes up,' senses that all derive very naturally from'exchanges.' Not but that the sense of 'takes' (gets) may have been directly derived from 'strikes,' as in Lat. nancz'scitur(cf. AJP. XXVI, 193); and note Lat. capit 'takes':"o7rT€£ Vol. xxxvii.] Latin Word-Studies. 23 'strikes, cuts,' for the notions of 'striking' (cutting) and 'taking' (obtaining) are, pace Walde, s.i1. capio, capable of being united; cf. Eng. hits 'strikes, lights upon, attains to' : Swed. lzitta 'to find.' The sphere of activity in which these notions come together is in the chase (cf. AJP. I.e., adding ~ -; ~ ' ) 1cvpH 1cvpei to TuryxavEi . In view of aµij 'mows, reaps' (i.e.' cuts,' with a long-grade 1 a-), we may wonder if ( r) Lat. amentum 'strap for binding ' and (2) times' notch, furca,' luimus 'hook' (if with inorganic h-), do not also belong to AME(Y)-in the senses of ( r) 'pangit, The vincit,' (2) 'caedit' (the notch being 'what is cut in'). words ansa and amp!a 'handle,' referred by Walde to a root am-'to grasp,' may also belong to our base, whether we think of a handle as a 'notch' or as a projecting 'knob' or 'peg' (cf. Skr. kanfli.-s 'ear, handle,' Graeco-Latin diota 'two-eared jug'), i.e. 7racrcraA.oi;. Attention is also called to the following rhyming synonyms of AME(Y)-, viz., Lat. premit (if from prmmeti) and Skr. vkam­' amat'; for in rhyming synonyms w~ may suspect, at least to a limited extent, a common origin of the phonetic elements common to all members of a group (cf. my remark on Lat. apio, capio, rapio in AJP. XXV, 373). . In the Dlzatupa(!ta Skr. yam-is defined as 'ehren,' no wide remove from amare and y'kam-. (6) frangit, fregit. Has no one yet suggested that in frangit we have a blend of the bases of Skr. b!tandkti 'frangit' (: y'bhaF1j-) and of Mryvvcri ' frangit ' ? Postscripta : P. 7. Cf. tractabant = arabant, occabant ( ?) , Luer. v 1289. P. 17. Lucretius, v, 1190, writes signa severa for ste//ae i11erra11tes. P. 20. On the love-lick, also see Crawley, Tlze Mystic Rose, p. 353 sq.; noting, "When analyzed, the emotion [loveJ always 1 But Greek tvaµ.µ.a-if not sug-"ested, as its time of appearance would admit, by amentum-would seem to make for Vanicek's derivation of ammentum from •apnuntum. 24 Edwin W Fay. come;; back to contact" (p. 77); "Primitive physics, no less than modern, recognizes that contact is a modified form of blow" (p. 79). P. 21. Hirt writes ema• with e because of 0. H. G. emazzig, whose ~' it would seem, may be mutated a; cf. the forms ~mizzig and ~miz in Kluge's Woert., s.vv. emsig and ameise. AevTepai cppoVT{oer;. P. 6. A trace of the specific sense 'pulls,' assigned as the pnm1­tive signification of accersit, would seem to be extant in a proverb-like phrase in Plautus, Am. 327: illic homo a me sibi malam rem arcessit iumento suo, "the fellow is pulling a beating from me 0:1 his head with his own team." A very special usage of accersit, well attested in Plautus and Terence, is in the sense 'fetches the bride forth' (see the T/zesaurus, II, 452, 5 r ), spoken of the bridegroom and of others who bring forth the bride. Its special appropriateness to the Roman bridegroom is clear from Festus, p. 289, rapi simulatur virgo ex gremio matris (cf. also Catullus, !xi, 3, 58; Apuleius, Met. iv, 26; Macrobius, Sat. i, 15, 21). The Plautine instance is Aul. 613: ne adfinem morer quin ubi accersat meam extemplo filiam ducat domum ; that is to say, being interpreted, that there was an accersio by the bridegroom prior to the dommn deductio. The bride's natural shyness and reluctance may have made it necessary for others also, as well as the bridegroom, to pull her about as preparations for the wedding were making. For instances of such reluctance in all manner of savage tribes see Crawley, Tlte M)'Stic R ose, p. 354 sq. P. 7. Schulze (Latein. E igennamen, 209) asks whether the name Perternius has not arisen beside Petenzius, owing to "Vorausnahme des r." P. 15. Schmidt's H esychius has the following entry : vi~a> · -ra u-rpwp.ara (Aeol. ?). The current assumption, that in Latin ,jnedh­yielded nectit under the influence of plectit, is not more reasonable than that .ynedh-'vincere' was affected in the proethnic period by the b:1se of Skr. par ' vinculum,' whence the secondary (?) root nek-. [Reprinted from AMEUICAN JOURNAL OF PHILOLOGY, Vol. XXVII, No. 3.] III.-GREEK AND LATIN ETYMOLOGIES. I. THE BASE SEP-/ SOP. 1) Lat. sapit: Gr. £1m, etc. The Latin verb sapit means (1) 'tastes', (2) 'has taste, per­ceives'. Greek ihru1 means 'fastens, binds' but iirrTErn1 means 'grasps, touches'; cf. Lat. stringit (1) 'touches', (2) 'brings in touch, binds'. Further, Cirrr•rn• has the derived senses (3) 'per­ceives' ( = 'grasps mentally") and (4) 'has sexual intercourse with'. In view of English tastes' sapit, gustat 'which is identical with German fasten 'tangere, tractare ', both borrowed from older French taster' to feel, taste' (cf. tate-vin. 'wine taster') ; and in view of Lat. ta11git 'touches', but also specifically 'tastes, eats and drinks'; we may suspect the ultimate cognation of Gr. iirrrerui 'touches' and Latin sapit, 'tastes'. This brings us to a consideration of the Sanskrit verb sapati, defined by Whitney in List of Verb-Roots, etc. by' to serve'. But in fact sapati seems almost to resist a single definition. The Vedic commentator Sayar:ia defines sapati by sPrrati 'touches', and the compound rtas'ti,pas• (nom. plur.) c keeping the ritus' is hardly to be felt as different from the compound rta-spf{as (nom. plur). Supposing Saya9a's definition of' touches' to be correct­and it is borne out by the Avestan idiom hafsz zasta (=Sanskrit sapasi hastena) 'thou boldest (graspest, touchest) with thy hand', cited by Uhlenbeck (Etym. Woert. d. ai. Spr.)-sapati invites comparison with CirrreT"ai 'touches, grasps' (cf. Bartholomae, air· Woert. col. 1764), and with sapit 'grasps (mentally)'. Inasmuch, however, as sdpati is cognate with Homeric orru 'tractat ', the problem of correlating lirrura1 with £rru presents itself for solution. It is unfortunate that €rru is as vague a word as sdpati and equally defies precise definition. But the cognation of 7Tfp&tea'A'Aia nvxE' E7TOVra, acnriaa 1~Ta. It will, therefore, be necessary to explain the

8'1" beside E't>..1]µµa' and l'Aficf>O'I"• (: 'Aaµ{3cim 'takes,' wins') might well have resulted in the introduction of the q, of Ei>.'lcf>a into forms of c'irrm (cf. 2d aor. '7cf>'lv ).2 Especial attention may be called to the rhyming pair acf>vuun and 'Aacf>vuun; acf>vuun 'draws (wine, water,) -trahit ' seems a specialisation of a# 'tractat ', while 'Aacf>vJun 'draws in, swallows greedily, quaffs' is a specialised sense of Eifl.'lcf>a (cf. trahit pocula in Horace). The meaning 'draws, tears-quaffs' is exhibited in Greek by urran (i. e. urr-au-n, cf. the Homeric aorist cnrauuaTo), which belongs ultimately to our root S~P-'to touch, take '.3 It remains to call attention to yet another Greek derivative of the root sep-'to touch', viz: drra'AOs 'soft'. For the signification cf. Gr. µa'Aateos 'soft' : Lat. muled 'touches gently ', mulcat ' touches rough! y '; cf. also 8&{3p6s, glossed by Hesychius with arra'A6s, which will thus belong with o,yyavn 'touches' (from a stem *dhigw-ro-, cf. Lat.fivere lfigere). Cogm.te with arra'Aos, perhaps, is German sacht (with -cht for -ft, cf. Kluge's Woert., s. v.) 'soft'. Independent of the breathing, drra'Acls may be explained from EP­(see Am. Jr. Phil. I. c.). 2) /lvfJp6>rros. In Proc. Am. Phil. Assoc. 25, vii fn. I made a suggestion in passing(" peripherisch ") as follows: tlv8p6>rror 'human' from civapo 1 Unless, resorting to determinatives, we write a base SE-BH-/SE·P-. •The root-determinatives must have originated, in the main, from such rhyming assimilations. s Conversely German zielun 'to draw' (: Latin dzieere) seems to have de­veloped, when borrowed for use beyond the Rhine, the sense of 'to touch' (= Fr. toucher), 21 310 AMERICAN JOURNAL OF PHILOLOGY. +¢For (: cf>v"') 'possessing the nature of a man', where transfer of aspiration is to be recognized". This attempt to account for the eseems to me not to have received the attention it deserved. What I had in mind was the possibility that the parasitic dental appearing as a in avapo-was not so definitely a media as to exclude its aspiration: and I even fancied that the transfer of aspiration might have happened when DH rather than e was spoken by the Greeks. I based -cf>For on Skr. vibhvan 'egregius ', abhvas 'unmenschlich ', Lat. superbus 'uebermenschlich '., cf. vrTEpcf>laXor (Brugmann, Gr. Gram.8 § 24. 4). It might now be suggested that *avapocf>or was an animal name in -cf>or (: BHA-'to shine', cf. Prellwitz, B. B. 22, 76 sq., cited with approbation by Hirt, Gr. Gram. 284, c), and meant 'having the appearance of a man'. Semantically, it would make .little difference whether we operate with -cf>For or -cf>or. But a difficult phonetic question remains: how account for "' in *avlJp6>-cf>or? A lengthened vowel in the first member of com­pounds is no stranger to Vedic Sanskrit, and A vestan and Greek parallels are found (cf. Wackernagel. ai. Gram. II, § 56). We might justify avlJp"'-as we justify Skr. rathti-. Apter parallels are virti-qah-"Manner beherrschend ", narti-farhsa-"der Manner Lob" (?). Note the nearly exact proportion Lat. vir: Skr. virti-= Skr. nr-: Skr. narti-. For avlJp"'-· in view of Lat. nero-n­' manly', the question arises whether we have not to do with an on-stem. Here, in my opinion, lies the solution. In the Veda, also, -an-stems exhibit a in compounds, e. g. vibhvti-sah-'fortes­vincens ', vrqa-yudh-' tauros-pugnans '. How are we to interpret this a? Wackernagel (I. c., d.) especially denies to the ti-of an­stems a proethnic character. In this I believe him to be wrong, and particularly on the score of Lat. leno-cinium 'procurer­w heedling ',a compound I have treated at some length in Class. Rev. 18, 349. I there explained as leno[ni-]cinium, but when I consider the type of vibhvti-sah-, vrqa-yudh-and vrqti-ravd-s 'bull-screaming' (alongside of vrqa-bhara-s 'Manner-hegend ') I can see no reason for denying to leno-cinium (alongside of vaticinium) the character of a proethnic type of compound. The long vowel got into the compounds, I take it, from phrase­groups like leno canit; thus leno-is, in fact, a nominative. That the"' in *avlJp"'-cf>or is to be connected with the o ofLat. nero -that both exhibit a nasal stem-is to a certain extent attested by iXa-cf>or (with a= f!, cf. Prellwitz, I. c. 100) and Lat. colum-ba. GREEK AND LATIN ETYMOLOGIES. As to accent, *tJ...af>6>-cf>os has the recessive accent of the btihu­vnki type, and that it falls on the d-is no more-and no less­difficult than the accentuation &..ip•s/ tJ..bpu, &..apciu1 / tJ.11bptuu1. This explanation has a phonetic difficulty to acknowledge: how account for -8-rr-from -cp-? A form like Ionic tJ.xavros for tJ.ica118os exhibits shift of aspiration, it is true, but a shift from aspirate ( 8) to surd (ic) is not a warrant for the shift from aspirate (cf>) to sonant (a). So if we were to admit the derivation of tJ.v8pfiJrros from *avlJp"'·cf>or, the phonetic problem is not attested by any parallel example; and, indeed, the fact that the sound -a­inserted between -vp· is parasitic renders the phonetic conditions unique. We must, therefore, feel a great reserve in regard to the derivation of tJ.v8pfiJ1ros from ,.,.d.lJp"'·cf>os. A hardly less unique phonetic problem is presented in the ex­planation of C:.v8pfiJrros from *avlJp-kc.irros (or *avlJp-horros) which was advanced by Brugmann in I. F. 12, 25. We cannot infer from Ti8p11r1rov out of *T£Tp-k1rrrrov (?)1 to C:.v8p"'rros out of *avlJp-h"'rros. The Attic form p.118tts from p.'f/lJ' Els with a probable intermediate stage *p.qT' •ls (cf. Brugmann Gr. Gr.3 § 139, e) is also not a real parallel to -8p· from -lJpk-." Besides, the reconstruction of a Gr. */.,rros/*orros 'face' (from a root SEKW' to see, say'), on the basis of Gothic siuns 'face', Lat. s£gnum,3 O. Bulg. soku 'accuser' must be accounted daring. In view of these reasons we must take an attitude of extreme reserve also to the derivation of tJ.v8pfiJTrOS from avbp-+ *kc.irros' 'face, appearance'. This is particu­larly advisable in view of the fact that the compound *avbp-h"'rros is not the continuant of *N1:l·SOPOS, but must have originated m Greek after SOPOS became *h"'rros. 1 *'rETpa+ lrr7rov seems to me the only correct writing, for I-shows that we have not before us a continuant of KWETR-EKWOM: Ttrp'· from nrpa-by elision? 'I fancied I had found in Homeric TrroA1e8pov, which I was fain to derive from *1rTON·keopo-•city-site ', an analogy for the phonetic change Brugmann finds in ii.v8p=o,, but in view of r.th:fJpov •acre of land', phfJpov stream, 'Pti8pov •Stream-town', iU.fJT;pov 'bolt', the old explanation of -e..fJpov as suffixal seems to be valid. s On signum see below, No. 6. 'Admitting *wTro, we might compare *rrpoa-!wTrov • mask' ('over the eyes') with Lat.per-siina (-siina from SOKW-SN·A, cf. for the -sn-the Skr. stems ak-~r;i­ • eye', cdkJ!ar;ia-•appearance'). Perhaps persona comes from *pres-sona, pres­being a by-form of 7rp6' (cf. Aeolic rrpt{): it is not necessary to derive rrpo'· from r.p6n, but it may well be either PR-OS or PRO·S (cf. Brugmann, Kurze v. Gr.§ 610). AMERICAN JOURNAL OF PHILOLOGY. It is easy in moments ot scientific exaltation to speak of the "Sirene des Gleichklanges ", but rigid adherents to the doctrine of phonetic uniformity have found it hard to contemplate the disjunction of a118p-wrror from a11lJp· (cf. e. g. Brugmann, I. F. I. c.), especially because of the Hesychian gloss lJpwo/ • tMJpwrror (cf., e.g. Meillet, Mem. de la Soc. de Ling. 7. 166). So brief a gloss, without any usage in which llpwo/ might betray a special sense, looks like a modern scientific etymology, to be sure; but we cannot make sure that llpwo/, if more fully defined, might not be seen to be cognate with lipwrrTTJr • TrAa11rfTTJr, TrTwx6s, llparrfrTJs' ¢vyas 1 ( : Skr. drapayati 'causes to run'). Perhaps also in the gloss lJpi.>o/ • a11lJpwrros we are to recognize the debased sense of 0.11lJpwrros, quasi 'servus': this would give us a reason for suspecting the ultimate cognation of tJpw·Tr· with tJpTJuT~P 'servant' (cf. in Homer the forms lJpi.>01µ.1, -llpi.>wcn [cited by Leo Meyer, op. cit. III, p. 245] and the gloss llpi.>wu1 • ll1aK011oi!u111, iirrTJp£TOvu1). The positive suggestion I have to offer for &11lJpwTros is a deriva­tion from a11Tpo-+ *wrro•· ' cave-dwelling', 'specum pro sepe habens ', basing *wrros on Lat. seps ' hedge, praesepe 'hut, hovel', and comparing Skr. rtasapas which, instead of rendering by "heiliges Werk pftegend ",we might explain in terms of rta-s&d "im heiligen Gesetze seinen Sitz habend." The English word 'keep' will render all the compounds fairly well, rta-sap­' keeping the rfa', rfa-s6,d 'keeping in the ffa ',*a11TpohwTrOS 'keeping in caverns'. A point to which I attach some importance as a support of this explanation is the following: in Homer there are next to no examples of the singular of a11lJpwrros, and the plural so markedly preponderates as to admit of our supposing that the &11lJpwrroi formed a community of 'cave-dwellers', to-wit.2 In the same sense we might derive lJpwo/ from *tJpF-hwrr-'woods-dweller'. As regards the definition of &11lJpwTroi by 'cave-dwellers', though well enough in line with archaeological facts-I refer to the cave­dwellings of early neolithic men, and note cases of historic sur­vival like the Tpwy("A)otJurm of Herodotus-it seems to me to have less to commend it than the earlier exposition as *a11lJpw·¢os, in 1 Cf. also (?with a formative ·P·) op{nrr17V • a'M;rTJv, cognate with Skr. dru-tds 'running' ; OpV7rOAei. bpei(3aui, Opv'l/Joyipovrnr. rovr ar67rovr rrpea(3vrn{ KOL olovet ariµ.ovr. 2 There is accordingly no inconsequence when Homer describes the dead in the islands of the blest as av1Jpw7rot. GREEK AND LA TIN ETYil-IOLOGIES. spite of the clear phonetic difficulties involved therein. However, something yet may be urged in its favor on the semantic side, viz., the relation of Lat. homo 'human': humus 'earth', cf. x8o~&o& 'unterirdisch ',which lets us surmise that homo meant not so much 'earthly, mortalis' as 'underground-dwellers': here note from the philosophical summary of Diogenes of Oinoanda (Rhein. Mus. 47, 440) the phrase oi rim> y~r cpvvrfr [..61rr1r. (1) The Hesiodic Kliic>..61rr1r are lightning forgers; (2) the Homeric Kliic.\61rrfr inhabited caverns on mountain tops; (3) a last variety were wall-builders. The two last may well be one, as­suming that the first walled-towns were built (about caves) on hill tops, as fortifications to take refuge in. If we transcribe KliicA611m by Skr. *cakra-siipas and define by 'discus-grasping' (cakra-, the discus of Vi~!)u) we get a most improbable sense for *icvicXo + h61rr-1r, for Vi~i:iu's discus and Zeus's thunderbolt do not invite identification (cf. the imperial Roman thunderbolt in Duruy, Histoire des Romains I, cxxiv). But *icvic>..o + h61rr-1r, ' wall-joining' yields a correct definition for icvicXo-, and a!:counts for *~rr-1r as a derivative to SEP·/ SOP-(see above) : in view of the recessive accent, however, KliicX61rr1r should be rendered by "rotundas-saepes-habentes." Of course, the mythical "round­eyed" is well enough if the myth is older than the name: but if the name is older than its explanation, then "round-sited" (wall-sited) has, by' disease oflanguage', become' round-sighted' ( ? wall-eyed). 4) ,,;;,pof. This adjective, usually defined by 'shining' or ' noisy', because it is always coupled with xa>.icor, may perhaps be better defined AMERICAN JOURNAL OF PHILOLOGY. by 'man-hedging, man-sheltering'. The definition is certainly apt, as xaXicor, with which it is always found, means 'protective armour', and it seems reasonable to derive from ""'P" 'man' (long-grade, corresponding to the grade exhibited by ciy-~""'P• or else from N~) + ho'TI"-'hedging, protecting'. S) µ£po'f. In the Homeric phrase µ£po11"Er t1118p"'11"0' (flpoTol) the epithet µlpo'Tl"Er is very hard of definition. As a proper name Mlpoy was the name of a famous seer (A 329 = B 831). An examination of the remaining usage yields the following result: (1) the term seemed especially allotted to city-dwellers, as in Y 217, fl"oA'r p.Epo11"6lv civ8pw11""'" 1 (cf.~ 288, 342, 490, r 402)-cf. Euripides, Iphig. T. 1264; in A 250 it is said that Nestor outlived two generations p.Eporr"'v civ8pw11"6lv ; 2 in I 340 the Atridae are asked if they alone µ. ci. love their wives; in A 28 a rainbow is described as a portent unto µ. ci. ; and in B 285 Agamemnon is said likely to be rendered the most disgraced p.Epo'Tl"Euu' flporo'iu,v; in" 49 companies of soldiers are compared to µ. ci., while in 132 Penelope is said to yield respect rather to the worse of µ. ci. than to the better. After Homer µlpof occurs in the 31st Homeric hymn (end) with the following context : lie (TtO ll' ap~aµoor KAyu@ p.Epo'Tl"(A)V ylvor J.vllp6w s I ~µi8i"'"• &iv tpya 8EOl 8v1rro'iu'v V ytvor avllpwv by 'genUS hominum Optime merentiUm ', If we render µ£po11"•r by 'optime merentes' it will not be incon­sistent with the tragic usage of Aeschylus (Suppl. 88, Choeph. 1017) and Euripides (!. c.) From all the passages one might infer one of two meanings, either (1) righteous or (2) civilized. For the first sense we may derive from p.Ep· (: µo'ipa 'destiny, fas'; µopor 'destiny') + ho'TI"-, comparing again Skr. rta-sap-'right-keeping'. If the second definition, 'civilized', 'city-dwellers', be correct, we may derive from p.Ep-( : p.•lpETat 'divides') + *011"-' precinct' (cf. Lat. sep-s): hence µipn'Tl"Er t1118p6>11"0' would m~an 'men who divide their pre­cincts '-those who had advanced from collective to something more like individual ownership (cf. the more individualized 1 In the hymn to Apollo 42, the same phrase occurs, but Mep6rr= is taken as a proper name =a tribe of Coans. Homer distinguishes the city of Cos as well-built. 2 Cf. Hesiod, Frag. 172. 5, aCf. Hesiod, Epya, 109, XPVO'EOV µev rrpiJTtO'Ta yivo~µep61r(,)V av8pf.J11"(,)V I o.eavarot rrolr;C1av. GREEK AND LATIN ETYMOLOGIES. proprietorship described by Tacitus, Germ. 26, with the more collective system described by Caesar, B. G. VI. 22). The accent of f'ip01"1 if not secondarily shifted to conform to the rather large group of adjectives in -orr-(cf. Leo Meyer, op. cit. 1. 486), accords better with the second explanation. There is a bird (the apiaster) that the Greeks called f'Epa1y. To what did he owe this name? Perhaps to his bright color (?from µEp-: Lat. merus 'bright' and -orr-'eye-spot'), but the curious notions recorded by Aristotle (H. A. 9. 13. 2; 6. 1. 6) and Pliny (N. H. 10. 98),-to the effect that the f'ipoo/ (1) cared for his parents, who kept their nests; (2) that they nested alone in holes (otral),1 in the ground, or deep in mud-banks beside rivers­make us suspect again that ,,.Jpot meant either 'optime merens' or 'dividens-specus '. II. 6) Lat. signum: rx,.or; dignus; ignis. In Proc. Am. Phil. Assoc., vol. 26, p. liii (Special Session, 1894) I propounded two etymologies, in the following words, "signum 'statue, sign' : seco 'cut' (cf. szca 'dagger') --dignus 'worthy': ~Elic"vf'al 'greet, honor'." I return to them now, apropos of the question of the vowel-length before the group -gn-, as treated by Buck in the Classical Review, 15, 3II sq. Latin inscriptions attest signum,2 dignus (and ignis), but the Romance languages attest vulgar (?) Latin slgno-, dlgno-(or segno-, degno-?), nor can we say whether i or l was the best Roman pronunciation in these words. The dialectal difference is habitually treated as Italic: it may have been inherited. At any rate, the root to which I have referred signum is, in its fullest form, to be written si'(v)-K-(cf. Brugmann, Grundr. 2 § 549. c.); and the root of dignus: aElicvvTm 'greets, honors' may also be written D~(Y)K-(so Brugmann, Gr. Gram.5 § 340, end); to these the proper -no-derivatives would be S;}IK-NO-M, D;}IK­No-s, i. e., Lat. signum, dignus. Lat. slgnum, d1gnus would 1 If for *orrai, then cognate with Lat. stp-s •[3ofipor '. •No conclusive evidence for an i short by nature can be drawn from the diminutive sig-lllu'11, for what Vendryes (Intensite Initiale § 72) neatly terms the" loi de mlimilla" need not be limited to consonant simplification: the Latin compound verb conscribillo, diminutive to scribo, seems a sure case for antepenultimate vowel-shortening before a stressed penultimate (scribo: acipitJ>or). 316 AMERICAN JOURNAL OF PHILOLOGY. proceed normally from SEK·NO·M (SIK·NO-M), DEK-NO-s (DIK­NO· S), all of which are normal derivations from bases si'(v)K­and DE(Y)K-. That ' cut, mark, form , sign' constitute a betterdefinition for signum than anything to be got by the comparisonwith Lat. inseque, Goth. saihvan 'to see' (cf. Brugmann, Kurze v. Gr. § 309 d) seems to me to need no demonstration beyonda statement. There is no substantial semantic difference whetherwe derive dignus from the root of a.iKv11Ta1 or from the root ofdecet 'it becomes' (=" it honors"), for both, in my opinion, goback to the base nt(v)R-.With the derivation of szgnum from savK-NOM I couple thederivation of 1x•or 'track ' from SIK-SNOS-. For the specialisationof the original sense of 'cut, mark', we may note that whileOvid writes out signa pedum = 'tracks', Vergil uses a bareszgna = 'tracks'. The Greek specialisation was as early asHomer, who uses 1x"'°" (which has the form, though not theaccent, ofa diminutive to i'xvo·r), in the Odyssey only, in the senseof track (cf. p 217, txvf