ILLINO I S UNIVERSITY OF ILLINOIS AT URBANA-CHAMPAIGN PRODUCTION NOTE University of Illinois at Urbana-Champaign Library Brittle Books Project, 2013. COPYRIGHIT NOTIFICATION , In Public Domain. Published prior to 1923. This digital copy was made from the printed version held by the University of Illinois at Urbana-Champaign. It was made in compliance with copyright law. Prepared for the Brittle Books Project, Main Library, University of Illinois at Urbana-Champaign by Northern Micrographics Brookhaven Bindery La Crosse, Wisconsin 2013 III. - On the Sounds and Inflections of the Cyprian Dialect. BY CHARLES E. BENNETT. DR. ISAAC H. HALL in the Proceedings of the American Oriental Society for October, 1877, stated the wants existing at that stage of Cyprian study as I) a complete collection of inscriptions, 2) a correct syllabary, 3) a compilation of the best interpretations; after which, grammar and vocabulary. The first and second of these wants have been admirably met by Deecke's publication of the existing inscriptions with an appended syllabary in Collitz's Samnmlung der Griechisc/en Dialekt-Inschrften. Bd. I., Heft I, 1883. The excellent foundation laid by this brilliant and thorough work has encouraged the present attempt at a systematic treatment of the grammar of the dialect. The inscriptions made use of, in addition to those published in Collitz's Sammlung (212 in number), have been the following :-I. The two inscriptions with fragments of two others published by Sayce in the Berliner Philologische Wochenschzrift, 1884, No. 21. 2. Three inscriptions published by Hans Voigt in the Studia Nicolaitana, 1884. 3. Thirty inscriptions published by Deecke in the Berliner Philologische Wochenschrift, 1886, Nos. 41, 51, 52. 4. The two bilingual inscriptions of Tamassus, published by Deecke in the Berliner Philologische Wochenschrft, 1886, No. 42; 1887, No. 12. 5. Meister's new reading (suggested by Deecke) of inscrip- tion 41 in Collitz's Sammlung, in the Berliner Philologische Wochensczrift, I887, No. 52. UNIVERSITY STUDIES, Vol. I., No. 2, OCTOBER, I888. I3I Charles E. Bennett, 6. The two inscriptions published by Deecke in Bezzenberger's Beitrag xi., p. 315 f. e, 7. The reading of COLLITZ 134 as published by Prellwitz in Bezzenberger's Beitrdge, ix., p. 172. The inscriptions discovered in Cyprus during the last year have not as yet been published. It is to be hoped that they may add to our knowledge of the dialect. As regards the inscriptions published by Deecke in Collitz's Sammlung, I have been compelled to doubt the general correctness of one or two of the longer ones, and mention this here that more weight may attach to what is urged below against particular forms occurring in these inscriptions. The inscriptions are Nos. 68, 69, and 126. No. 68 is the longest of the inscriptions in the Cesnola collection in the Metropolitan Museum of Art, New York. The characters are quite clear in the main, to judge from Hall's fac-simile (Yournal of the American Oriental Society, x., Plate iv., 13). The divisors are also plain. But unmistakable as several of the words of the inscription seem to be, e.g. XpeTe, line I; a eCOi', line 2; (v)Opcore and 0ewih, line 3 ; ird(v)ra and d(v)Opwrot, line 4, yet there are other words exceedingly doubtful, especially 7ro't, F47rc), Eclo-iS in line I; 6 EpepalE'va and ra(v)7a copao-Tos in line 2; o, E7ro-Tr , X (X) eTvx' aIp ~ in line 3; and Kvlep)pat and cpovcowl in line 4. These words are doubtful not only from the uncertainty of some of the characters contained in them, but more especially in view of their peculiar and irregular formation (see below for the separate cases). Furthermore the interpretation which Deecke seeks to establish for the whole inscription (see Bezz. Beitr., vi., p. 78 ff.) is so forced and far-fetched, that I cannot believe the reading which yields such a sense to be correct. Several words as orrte and deta-als, even if formally correct, cannot have the signification which Deecke attributes to them. Even the metrical structure of the verses (Deecke claims four hexameters) to which Deecke appeals for the confirmation of his results, is extremely harsh, involving the lengthening of the final L of r5 t, the lengthening of 132 Sounds and Inflections of the Cyprian Dialect. the Lof T , the shortening of the second syllable of eiro-Ta, , 1 0 E5s the crasis of O 2 4X(x)', besides two striking instances of hiatus. In view of all these difficulties I cannot believe that the true reading of this difficult inscription has yet been reached, and I have therefore felt it unsafe to attempt to base any grammatical conclusions upon it, at least for the present. An irregularity or two in an inscription otherwise certain (e.g. a XowV COLL. 60, 14; a&aO&t 59, 4) may be easily ad- mitted; but to admit the existence of irregularities in any number in an inscription which is thereby made to yield only an unsatisfactory sense, reduces the probability of the correctness of any one word to a minimum. No. 69, though apparently accepted without reservation by Hall (Journal of the American Oriental Society, xi., p. 22I), seems to me to be open to precisely the same objections as urged against No. 68, including faulty metrical structure. No. 126 is uncertain in several of the characters, and a number of the words as read by Deecke involve principles at variance with the usage of the dialect. The sense too is not convincing. No. 41 is now taken by Deecke (Bezz. Beitr., xi., p. 317) as reading from left to right, instead of from right to left as formerly. This gives an entirely different text, which has not yet been fully made out. Meister, following Deecke, has offered in the Berliner Philologische Wochenschrift, 1887, No. 52, a new reading of the inscription. But this is largely conjectural and to be accepted with caution. Nos. 122-125 have been shown by Voigt (Bezz. Beitr., ix., p. 168) to be in.all likelihood cleverly executed forgeries, and will accordingly be left entirely out of consideration. 1 This is Deecke's explanation, but it is simpler to assume aphaeresis of the initial E. 2 Equally harsh would be the assumption of synizesis in ew~, with shortening before the initial vowel. 133 Charles E. Bennett, SOUNDS. VOWELS. 1. aL.. Cyprian a corresponds in general to primitive Greek a and a(v)T( COLL. 60, 5; 3ao-keli 17, a of the other dialects; e.g. I; Fdva 18, I. 1. dkrov COLL. 60, 26 appears in other dialects as 8XTo70, being derived from the name of the letter delta (rb &XT7a). The Semitic name of the letter, however, is daleth, and it is doubtless owing to the influence of the Phoenician dialect of Cyprus, that the Cyprian Greeks employed the form S&X7ro while the others said e' -roc. 2. Whether lap6o as in Doric, Elean, Boeotian, Thessalian and Arcadian (in the latter by the side of iep6s) really exists in Cyprian is as yet uncertain. Of Deecke's three forms cIapch(v)Bay COLL. 118; tap7o a70o 41, I; and japd 72, 2, the first is entirely uncertain, and the second no longer maintained by Deecke himself (see Bezz. Beitr., xi., p. 317). Only the last of the three, japd, can lay claim to serious attention. Whether japd can be for ljapd (i.e. iapd; see § 18, I) is extremely doubtful. The only theory on which we could account for the disappearance of the initial L,would be that it merged in some way with the final L of the preceding This may be correct in spite of the divisor, 'AwrdoX(X)cov. a ". Cf.COLL. 26 e mi I o "la"o" ja viz. a po"lo ni Cf Deecke, Bezz. Beitr., vi., i.e. possibly 'IoX&o. - 7'7 p. 83. The regular Cyprian form is ieph (ijepbp); e.g. Ijepeds COLL. 40; iepfo9 38, 3, et pass. None of the forms in ijep- ever show any tendency to lose their initial vowel and to appear as 3ep-. 134 Sounds and Inflections of the Cyprian Dialect. 6 As to the relation of the two forms iepo' and iap s, it seems quite probable, in view of Skrt. isird-, that the latter is the primitive one, and that iepl is 6f secondary origin, with -epos for -apotafter the 'analogy of op/ep 6s, oXep6o, Tpoepds, etc. This is Osthoff's view. See Morphologische Untersuchungen, 2 iv., p. 149 ff. Cf.Meyer, Gr. Gr., § 94. 3. icaTadomo-e COLL. 127 cannot be a Cyprian form for icado'~ao-e. The syllabic text seems to give ka ta se 'te se . But the principles of the Cyprian syllabary would demand ka ta sa" te se" to represent icaTa-o o-e. This has led Voigt (Bezz. Beitr., ix., p. 170) to conjecture a mistake of the stone-cutter, by which the ta" and te' in the second and fourth syllables were interchanged. In that case we should get ka te" setse" se", i.e. ta rdrao-e,- the regular form, found frequently in other Cyprian inscriptions, e.g. COLL. 27, 2; 28. 4. Equally uncertain is 7ra(v)Taico'pao-os, the reading proposed by Deecke in COLL. 68, 2, which he takes (Bezz. Beitr., vi., p. 79) for ravratcopeo-Toso, i.e. an emphatic d~cdpeoro , for which Deecke compares 7wavTcipw-row, etc. But this change of to a is difficult to justify either physiologically or by any etymological combinations. Moreover, the word is suspicious in its composition. ravdpLo-rov, which Deecke compares, is not sufficiently analogous to give much probability to his view of the word. We ought to have instances of some verbal with alpha privative, to which wravn- has been prefixed, such as 7ravT-df/a7To, wrawr-dacpTro, before crediting so remarkable a form as wrav'-a-copeoro, even did it occur with e, and not a, as here. Hall (Jour. Am. Or. Soc., xi., p. 220) after a careful re-examination of the inscription in New York reads here 7rd(v)a Xopa i~, taking Deecke's sa " as an i, which he insists is correct. But 8w(; surely cannot be right. 5. As to the possible origin of the peculiar ending -av, in the acc. sing. of consonant stems, from -nv (cf.ETra/ov for '-Tqsy-ov), see below, under Inflections, § 29, I. 6. The a in &vot COLL. 71, 2 must be taken as short, if the inscription (with Allen, Versification in Greek Inscri 135 icaT e -beginning pepvac Ckarles E. Bennett, 6 tions, p. 46) is really to be regarded as metrical, which I doubt. Allen's view requires us to assume the addition of dyo; in verse I and the interpolation of either wrat or e6 in verse 2, along with the shortening of a to a in iemvapevot. This seems to me improbable, especially as thereby we gain only rough verses at best. 7. EaOtl, Hall's reading of COLL. 70 (Jour.Am. Or. Soc., xi., p. 221), which he takes as imperative of PdEw (" do sacri- fice ") is not in the smallest degree probable. 2. Cyprian a appears in ebxowxa COLL. 59, 4; 'o-rdaav 71; icapv 65, et pass. 1. Final -aL (i.e. -a) sometimes appears as a by the disappearance of the L. For the examples, see below, under Dizpthongs, § II,4, 2) ; 13, 3. Whether this change ever occurred in the interior of a word is extremely doubtful. Deecke thinks he finds an instance in "Apr COLL. 126, which he takes for "ASrltt (i.e. But the other difficulties of the passage rb(v) dge(v) "A(t)8(t) tLoadoTw (see below, § 23, 4; 26, 3) are so great that small probability attaches to the correctness of this particular form. 'Atsy). 3. Instances of regular 2; ye 56 ; OVEOi7Kc e are ydVotry COLL. 60, 29; &e ftwL 37, 72, I. COLL. 71; ETao-icpeTeo (gen.) Studia p. 68; Tt ocp&ov Berl. Phil. Woc/., 1886, No. Nicolaitana, ibid. vii. ; TtlofcpdE[eos] Berl. Phil. 41, II., viii. ; Woch., I886, No. 52, xxi., have E where the other dialects ", have a ('Apta-7oKpdryi etc.). COLL. 148 has ...ke " re "te apparently the conclusion of a proper name in -Kcp&rs, the COLL. 121 is very TLOotocpifrT first part of which is lost. uncertain. Besides forms in -icpdery, forms in -npcdry also 7ao-TiKpd7eo 18, 2. c7ao'-ticpar, COLL. 17, I ; occur, e.g. 1. 'Apc -roi T p dLXoicpe&reo9 'se 136 Sounds and Inflections of the Cyprian Dialect. These two formations represent two different forms of the (strong) and icpaT- (weak). The original inflec- suffix, IcpeCtion, nom. -icpirngv, gen. -icpdTeos (for *-icpaTeor, i.e. *-crtE6rov), has become modified by the "levelling" process (cf Wheeler, Analogy and the Scope of its Application in Language, p. 21 ff.). In most Greek dialects the levelling took place in favor of the strong form -icper-. The Cyprian is peculiar in that it has levelled both ways, and so developed two inflections, -IcpeTrs -icpaTS -Ipereos -icpCteos as shown by the above examples. (Cf. the Anglo-Saxon praeterite sang, plural sungun,; whence by similar levelling we get in modern English the two inflections sang and sung.) These were probably local differences. The close relationship of the Arcadian to the Cyprian is shown by the occurrence of proper names in both -icp&~'r and -icpdaTr in that dialect also, e.g. AiTOcpeVry COLL. 1246, D, 17; KaXXKtcpTyi" 1246, B, 15; ~OiEcpiT) B, 3; 'ApLo-ro0/cpdr 1181, A, 123I, C, I ; KaXXLKcp eo 1246, 12. 2. Ice COLL. 6o, IO, 23, 29. As primitive form of this particle we must assume icev found in Homer and Lesbian. This was doubtless originally orthotone. By its side stood the weak form ca (i.e. Icy), enclitic, preserved in Bceotian, The Cretan, Heraclean, Laconian, Elean, and Locrian. form Ice can only be explained (with Spitzer, Laut. Ark. Dial., p. 8, and Osthoff, Geschichte des Perfects im Indogermanischen, p. 328) as a compromise between these two forms icv and Ica, a " Contaminationsbildunzg." Parallel with Doric, Bceptian, and Elean Ica, as weak form of IcKv, occurs Thessalian a (i.e. pn) as weak form of fe'v, in the sense of e.g. COLL. 326, 3 ; 345, 20, et pass. Cf Prellwitz, De dialecto Thes- 8E; 2 salica, p. 48; Meyer, Gr. Gr., § 24, I). So also the Homeric and Attic pd as a particle of asseveration, e.g. va /h d6e ric7rrTpov A 234; vai h Ata Ar. Achar. 88. 137 Even in the 8 Charles E. Bennett, strong form p v, we see in Homer, and occasionally in Attic, unmistakable evidence of the same affirmatory force, which was original to this particle. That in these latter instances a " Contaminationsbildung" Le has not been developed, must be ascribed to the early differentiation in the meanings of fpv and ~pc, and the consequent feeling that they were separate words, while KEcZ and Ica, as long as they existed side by side, remained identical in signification. 3. eXaptvl[wv], on coins, COLL. 176, 177, and ZeXauivLoS, Sayce in Berl. Phil. Woch., 1884, No. 21, have while laXa- e, ivtow, COLL. 148, represents the vulgar formation. The E is probably attributable to Semitic influence. Deecke (on 176) compares eXaldv, a town in Galilee. The reading of COLL. 121, where Deecke suggests aXaalvtoq, is quite uncertain. 4. cpio-e'rv (aor., = copltoaTo) COLL. 126, I has not developed its from a by any phonetic process, but is simply an illustration of the tendency, occasionally exhibited by the sigmatic aorist, to assume the thematic formation. (Cf. the c same phenomenon in Homer, e.g. E7rt$p'eTo ' 78; 8 6o-e o P 328). Perhaps the Cyprian form is due directly to Homeric influence, as is undoubtedly the case with a number of words in this dialect. See § 20, I. 5. o-ja COLL. 41, formerly taken by Deecke (Bezz. Beitr., a vi., p. 71 f.) for doeia, another form of oo-cos, is now read by him quite differently. See Bezz. Beitr., xi., p. 317. e 6. For Deecke's in the inflection of nouns in -ev;, rather to be written, e.g. laouLtXFos not paa-LX'FoS. q is See below, under Inflections, § 28. 7. The Cyprian name of the town of Citium was Ki&tov, as seen in KerLov (gen. sing.) COLL. 59, I; KE7LTFEs 60, I. also the abbreviated KCTs. 57 and Ke. 195 on a coin. So 4. 1. to Cyprian a' corresponds regularly to primitive Greek n and of the other dialects (except to that Attic and Ionic q '1 138 Sounds and Inflections of the Cyprian Dialect. a), e.g. icao-tyyro -cpTdry~ 17, I. COLL. 71 ; ijaT-pav 60, 3; :ra 1. "AAS, is read by Deecke in COLL. 126. He takes it for "A(t)Bq(t) (see § 2, I; 13, 3, b). But this involves question- which has arisen from primitive Greek able principles in the case of the word itself, and the context is uncertain, so that Deecke's reading can only be regarded as conjectural. If correct, the word might be referred to Homeric influence. See § 20, I. 2. Moto-18po9 COLL. 127 is very uncertain and can be correct only on the assumption that it is an Ionic name. On icaTrdcayTOe in the same inscription, see above, § I, 3. COLL. 68, 2, taken by Deecke (Bezz. Beitr., vi., 3. Ede ol' p. 79) as gen. sing. of *EFE o, i.e. ''-o (cf. Hom. viec oe-at for *dFo-at) cannot be correct in its 11. The inscription, moreover, has elsewhere a, according to Deecke's own reading, e.g. Ovarotc, . On other objections to the word, see below, under Diphthongs, § 12, 2. 4. iOovlici COLL. 41, 3, formerly taken by Deecke (Bezz. Beitr., vi., p. 71, 5) as for iOvvic (' straight victory'), disappears with the changed reading of that inscription (see Bezz. Beitr., xi., p. 317). a3wo'dr (Jour. Am. Or. Soc., The Ionic 9 in Hall's E;no 229 = COLL. IOI) cannot be correct, especially with xi., p. the non-Ionic a in the second syllable. Ttpo&sopry ibid. 5. p. 231 (COLL. 121) is doubtful and improbable. 6. e I'r7OV1iO COLL. 128 is apparently correct and, if so, to be explained as an Ionic name. 7. o-v1r5o~ (for o-vXr~oj; see § 13, 3, b) COLL. 126, 2, is to be referred to a-vXEo, not o-vXcio, and hence its q presents no irregularity. The tendency of verbs originally ending in -a. to change to verbs in - o is abundantly illustrated by the (i.e. Lo 'wv) evidence of other dialects, e.g. Cretan inscription, II., 21 ; rTt/Linstead of youcdcwv in the Gortynian po.odtov ovo-a (i.e. rtGovo-a) for rTtdovo-a, Cauer, Delectus 2 132, 22 ; ovXE'v (i.e. avX-ev) Bulletin de Correspondance Helldnique, 1885, 10, 8 (cf. Herforth, De dialecto Cretica, in Dissertationes Halenses, 1887, p. 279) ; Delphian o-vXEovTES Cauer, Delectus,2 211, 17, et pass. 139 Charles E. Bennett, I0 8. The name of the city Idalium appears in Cyprian always as 'HcdXhov ; so COLL. 60, 1, 27; 'H3aXtlov (gen. sing.) 59, I; aX 'HatXLoL (loc.) 62, I; the inhabitants, 'H 8 LtFe 60, 2; 'H8aXLtjL 60, 3I; abbreviated 'H 3aX . 205; 206. In 60o, i6, 26 t Deecke (adloc.) takes the syllable e " as standing for the adj ective 'HGhXta or 'H&aXaKd. The only reason for transcribing Xov the e" here as 'H- instead of 'E- lies in the fact that 'I8P as found in classic Greek always occurs with long initial vowel. Deecke in COLL. 68, I reads Frrc as aor. subjunctive (= Att. iirro). Ahrens (Philologus, xxxvi., p. 17) had already 9. proposed FErT0, which he explained as present indicative. Deecke suggested Fr rr) on metrical grounds. But the word cannot be correct. The Cyprian form would not be )7Aro, as Deecke maintains (Bezz. Beitr., vi., p. 79, 5), but Felt7o, with EL as in Attic. Cyprian has q only as the equivalent of the Attic EL arising by compensative lengthening or contraction (see below, § 14, 7; 15) not as the equivalent of the genuine diphthong EL, which ebTrov had. That the EL did not arise here by contraction from *d-F-e r-ov or *'-e-Fr-ov, but was the genuine diphthong EL,is shown by the Old Attic EIIEN CIA. iv., 22, b, 4 (450 B.c.) and frequently (see Meisterhans, Granmmatik der Attischen Inschriften1, p. 79, Anm. 648); also by Lesbian reLt7rv, the tradition in Alcaeus 55, Sappho 28 (Bergk, Poetae Lyrici Graeci3). o10.On 1 for IL (i.e. q) in the 3d sing. of the subjunctive, see below, under Dipht/ongs, § 13, 3, b. 11. On , arising by contraction, see § 14, 7. 12. On , arising by compensative lengthening, see § 15. 0. Cyprian o answers regularly to primitive Greek o and to 0 of the other dialects; e.g. 8cXTov COLL. 60, 26; OTe 60, I; 59, 2. Tdoe 102; Sorvac 60, 5, 15; drayopev 1. oveOC'e COLL. 72, I ; 74, 2; 75, 2; 120, 4 corresponds 140 Sounds and Inflections of the Cyprian Dialect. I to the vulgar avitOyce. Meister's conjecture of o(.)/3i7[vt] (Ber. Phil. Woch., 1887, No. 52, col. 1644) is not at all certain. The origin of this monosyllabic form (frequent also in Lesbian, e.g. ovTee7r v COLL. 31I, 8, 34; ivOE&Ta 31 I, 39, and in Thessalian, e.g. drypa eZ COLL. 361, A, I I; B, 24; vypietv 8v345, 21) is not clear. Whether v-a, Ev, represent three originally different forms of the same root (i.e. weak, strong, and ablaut), bv- finding its correspondent in German an, and iv being for v (cf. Avestan an-a, for ~zn-a ?) is a question too difficult and complicated to be entered into here. One thing, however, seems certain, that unless iv and 8v- do stand to each other in the relation suggested, they are not etymologically connected, but originally different words, like /edc, re8- ; 2 As to the use of d'v and tv-, Meyer (Gr. Gr., § 55) thinks that 'v- was the form originally employed before consonants, av before vowels, and that dv- occurs before vowels, as in Thessalian and Cyprian, by a subsequent extension of its proper use. But this view lacks sufficient foundation. It is based upon too slender evidence, drawn from the Lesbian, which certainly admits of other interpretation (cf. Meister, Griechische Dialekte, I., p. 50). Beside the above-quoted Cyprian forms with 8v- we find also Av vice COLL. 17, 2; 76, I ; in the second instance in an inscription from the same locality as 72; 74; 75, which all ~/ce; also ve-, i.e. ave'Oyxe, in one of the fragments have ovEv published by Sayce in Berl. Phil. Woch., 1884, No. 21, where comparing Homeric (vov Sayce erroneously takes Ave- as ~ve, Nicolaitana, p. 69.) d 7 496. (Cf Voigt, Studia yv On v0rlKce, COLL. 45, 3, for avOyice, see below, § 9, 4. Deecke's earlier reading of COLL. 2. The o for in iOovicr, 41, disappears with the changed reading of that inscription (see Bezz. Beitr., xi., p. 317). v 3. 'A6(v)ra, COLL. 147, if correct, would speak for the sim- ilarity of o and v in this dialect, especially before nasals. below, on OvuBtOlce, § 9, 4. 141 Cf. Charles E. Bennett, 12 6. wo. Except when arising from contraction or compensative lengthening, o corresponds regularly to primitive Greek w and a of the other dialects, e.g. ebwXaci COLL. 59, 3; SkicOL 60, a'voyov 60, 2. 16; 1. 2. mo On o for (i.e. w) in final syllables, see § 13, 3, c. On o arising by contraction, see § 14, 5, 13. 7. Cyprian t corresponds in general to primitive Greek Lto-O0v COLL. 60, 4; to L of the other dialects; e.g. 37, 2; i 1Cdi20, I, etpass. t and 8ewo 1. In a number of words i has been changed from an origiErtcd(v)ra COLL. 60, 9, 19, 22; Frqja (Ion. AErea) nal E, viz. 60, 26; Os&& 37, 2; 61; 75, 2; O 6 v 6 0, 27; id(v)ra 60, 23; 'iwo- 60, 31 ; T7pXvja 60, 9, 18, 22; &7reXLja 60, 23 ; IcaTEcjav 6o, 27; 9lut Ber. Phil. Woch., 1886, No. 42, col. 1323. It will be seen that the phenomenon is confined to those cases where the E was followed by a, o, or o; so also in Boeotian and Doric (cf. Boeotian eos COLL. 425, et pass.; Avileav 414, I; Heraclean ascLe'ov I, 138). The j in the Cyprian forms carltjav, FErLja, dTEXja, rTpXvja has been developed Cf. Pamafter the change of to L; see below, § 18, I. phylian &(v)6pjwva for AvSpe0cva COLL. 1267, 8. Forms which retain the E before a, o, are about as frequent as those which change it to L, viz. 'ETeos4/a COLL. 135; eedmcop 126, I; Oe& 2, I; 3, I; 15, I; 16; OJcco 27, I; 3; 68, 4; 72, 2; edo 74, I; 78; 40, 2; Oeo7c 68, 2; lcc e co t68, eeoTlJov icXEoS 42 ; 35 ; eoicKXoS 126, eo8oGpcov 42 ; I; TL/poXc Feo9 36; 64; Tt/aso- T7ao/-tKcp~eo 18, 2 ; Ttoicpeov Berl. Phil. Woch., I886, No. 41, ii. ; ibid. viii. ; JLXOocpE&eoS ibid. vii.; E;Fd(v)Oeo9 COLL. 162; TraCo-tcpi7reo Studia Nicol- aitana, p. 68. 142 Sounds and Inflections of the Cyprian Dialect. The change from E Lis to 13 confined almost exclusively to the two inscriptions COLL. 60, 61, both from Idalion. Outside of these it is found only three times, always in Ged', viz. O i6 COLL. 37, 2; b[ t] 75, 2; louh Berl. Phil. Woch., 1886, No. 42, col. 1323. Of these the first is from Palaipaphos, the second from Athienu; the last from Tamassus. Spitzer's statement therefore (Laut. Ark. Dial., p. 16) that every before a or o becomes L in Cyprian, was evidently a conclusion drawn from the Idalian Bronze Tablet (COLL. 60) alone, and needs revision accordingly. On aTreXjv COLL. 60, IO (acc. sing. from aTeX-c where we , e should expect cTEXja for aTeXea; cf acc. plu. &T-eXi'a in line 23) see below, under Contraction, § 14, 6, and Declension, § 29, 8. 2. Another peculiarity is the preposition Iv for Ev. This occurs always in the form i(v) (see § 23, 2), viz. in COLL. 17, 2; 27, 2; 28; 31, 4; 37, 3; 59, 4; 60, I, 3, 8, 9, 17, 20, 31; 72, 2 (twice, once with the dative and once with the accusative); probably also in the compound ivaXaXko-p va 6o, 26; very questionable is icwda 126, 3. The closely related Arcadian dialect also has the same peculiarity, 'e.g. iv COLL. 1222, 2, 4, 20, 37, etpass.; the comp7cot (Att. dyceXtpounds 'iyyvov ('yyvor) 1222, 36 ; I7KyEXc prjcot) 1222, 12; iCSbaiv (Att. ep baivev) 1222, 24. The Arcadian also has dv several times, e.g. dv 'OXvvrat COLL. 1235, 5, etpass.; but only before a vowel. It is an ingenious theory of Spitzer (Laut. Ark. Dial., p. 14) 1183; iv lpvat that Iv developed in Arcadian from iv before initial consonants, and he adduces analogies for this change from other languages, Old German and Latin; e.g. Lat. tingo for *tengo 4 (cf T'ycyo), quinque (i.e. *pinque) for *penque (cf r w7e), though it must be confessed that such words as ventus, offendimentum (Idg. bhendh-) furnish puzzling exceptions. The fact that dv has survived in Arcadian is sufficient evidence that iv developed in that dialect only under certain conditions (otherwise do would have disappeared altogether), and Spitzer's theory that this was before consonants is highly 143 14 Charles E. Bennett, probable. According to him the old formula v wroX tot ica' iv Ipdva (COLL. 1233, 5) represents the proper use of dv and iv respectively in Arcadian. .(Cf. the similar relation of Es and eliin Attic, the former of which was orzginally used before an initial consonant, the latter before an initial vowel.) At the same time, Arcadian iv has already begun to encroach upon the legitimate territory of iv; e.g. iv atpalt COLL. 1222, 4; IvayodvTr 1222, 19. In Cyprian, 1(v), when used alone as a preposition, occurs only before initial consonants, never before a vowel, rejecting iv 'AU(v)mrc, COLL. 41, in view of Deecke, Bezz. Beitr., xi., p. 317. In composition we have probably one instance of Ivbefore a vowel, viz. ivaXaXto-va COLL. 60, 26. ivLrrd COLL. 126, 3, is too doubtful to admit. Deecke now (Bezz. Beitr., xi., p. 319) reads ta i "ne "ta "li "o"i i.e. Ta lv 'H&aXiot, in COLL. 62, in place of his previous reading ta "i ' Ie ta "li" o i., i.e. rat 'HcXtoZ(locative). This would give an instance of iv before a vowel. But the character which Deecke now wishes to take as ne ", while perhaps not a perfect e ", certainly entirely is different from the ordinary character for ne ", seen not as only in Idalian inscriptions, but others as well, and the mark after ta "i" as given in Schmidt (Sammlung Kyprischer Inschriften in EpichorischlerSclzrift, vii., 2) which Deecke wishes to join with the character in question bears every evidence of being a divisor. I can hardly believe therefore that Deecke is right in this new reading, whatever may be the difficulties of the old one. The form iv has not as yet been brought to light in any Cyprian inscription, but, under the circumstances, this must not be regarded as conclusive evidence that it did not exist side by side with iv just as in Arcadian. The only place in which lv might fairly be expected to occur would be before an initial vowel (assuming Spitzer's theory to be correct), and but a single instance (itself not perfectly certain) of this sort can be cited (viz. IvaXaXta-p va COLL. 60,26), which of course so far as it goes contradicts Spitzer's theory when applied to the Cyprian. '44 Sounds and Inflections of the Cyprian Dialect. 15 Hall's latest reading of COLL. 76 (Jour.Am. Or. Soc., xi., p. 223), which he has again examined in the Cesnola collection in New York since the appearance of Collitz's Sammiung, is rav FeLKdva rd(v)Be 'v 'Ard[X()wva], - i.e. 'to Apollo,' in which he takes 'v for iv with aphaeresis of the I (cf. of ('v) for of 1(v) COLL. 60, 31). His reading, if correct, would, in view had disappeared of the preceding Tr(v)Se, indicate that rather than L, and might be taken as furnishing some slight evidence of the existence of Ev; but in view of the incompleteness of the inscription and the possibilities of combina- e tion, Hall's reading cannot be considered safe enough to base conclusions' upon. Yet it is quite possible that the form Iv may have existed in Cyprian and may yet be brought to light. At all events, until instances of iv before vowels are discovered, we have no right to declare that iv had driven Iv out of use in Cyprian; any more than we should be justified in claiming the same for the Arcadian dialect on the basis of Arca- &p4iepats COLL. 1222, 4, and ivaydvro 1222, 19, assuming that instances of iv before vowels in Arcadian had not yet been found. Arcadian iv &d ppaLs and ivayovrm when viewed dian iv in the light of iv 'Apica8la COLL. 1200, 3; v ipdvat 1233, 5; ayioa- 1231, are seen to be encroachments of iv upon the domain of Iv. Cyprian ivaXaXLao-pva is perhaps most safely explained in the same way. Further light is thrown upon the question by the Cyprian forms tv COLL. 71 (tpv 'o"Tao-av) and ai COLL. I, I; 2, 2 are unquestionably ( aTre'OKe). These forms ,Ev and for p , the acc. sing. of the first personal pronoun. On the origin of Lv from t, see § 31, I. 1 is certainly to be considered as pd(v) (see § 31, 2), and as developed from /Ev before a consonant, just as iv from iv. Only the initial vowel of oraoaaav in COLL. 71 has preserved to us the form LEv. Can we doubt that, if we had preserved to us instances of the preposition (iv, iv) before initial vowels, it would appear as iv ? On Hall's reading of j(v) in COLL. 45, I, before an initial vowel and Voigt's reading of iv in 45, 4, also before an initial vowel, see § 23, 4; 9, 4. 145 v 'OXvvrrlat I I83, 6; 'tl Ev ,~l Charles E. Bennett, 3. 20, 2, On the Lof the ,carTEOo-av (= Att. KcaTreo-av) COLL. see below, under Conjugation, § 32, 5. 8. Where it occurs, t corresponds to primitive Greek i and to t of the other dialects; e.g. Aa1Zovco COLL. 151; 179; 'EXeiF eWV 38, 2 ; OeorTpLv 42; ivcs Bezz. Beitr., xi., p. 316; riLOt COLL. 135. 1. ipovLt COLL. 60, 8; 31, if correct and to be connected with iep' in the sense of 'consecrated district,' probably had t, which must be explained in the same way as the per6 plexing Homeric pd Lesbian Lpos. Osthoff (Morpf/ologische s, for Untersuckungen, iv., p. 151) assumes *~o--pos the original form, as otherwise it is impossible to account for the in Lesbian; a primitive * opos would have given *ippos in that dialect. Ahrens (Philologus, xxxv., p. 42) reads the L with the preceding Tr, i.e. TWe tlovi, or according to his principles Tro wovl (Tro locative; see below, § 27, 3) ' in the plain.' 2. S0bioX6s is Deecke's reading in COLL. 126, 3 for bet&oXoc. But the change of primitive EL to i in Cyprian is altogether improbable, since ELwhether original or of secondary z origin is elsewhere retained; e.g. ret'oet COLL. 60, 12, 25; EIret 59, I ; 60, I ; 'rEL 76, I ; aipei 60, 31 ; 'Arrelowv Berl. Phil. Woch., 1886, No. 42, col. 1323. Moreover, one or two of the characters of which the word consists are quite uncertain. 9. Cyprian v corresponds in general to primitive Greek v and to i~ of the other dialects; e.g. En--gvXe COLL. 59, 4; caipvy 65, I ; o-v 60, 28; prypco 60, 6, etpass. 1. 8vpdvoF (for 8v-dv-ot, with parasitic F, see § 17, 2) COLL. 60, 6 seems to be from the root 8v- ' give,' seen in Lat. du-im, 146 Sounds and Inflections of the Cyprian Dialect. 17 du-int, and not to be confounded with 3wo-,8o-. Cf DeeckeSiegismund in Curtius' Studien, vii., p. 248. 2. Final o in Cyprian when preceded by a consonant invariably changed to v. The instances are awtd COLL. 60, 8, 17; yEyvorv 60, 29; EFPr7yarcaT 60, 14; evFprao-aTv 60, 4 ; cpr-erv 126, I. So in Arcadian; e.g. COLL. 1222, 4; XXXv 1222, 40. The Thessalian and Lesbian also have ardt, and the Pamphylian shows the change of o to v not only in case of final o, but also elsewhere; e.g. &oX ao-rv COLL. 1267, 8; FOLKcvroXLe 'rt 3 1267, 14; /cXaXlcvv9 1267, 13. Arcadian caiv (for icaT') COLL. 1222, I I, 29, has not yet been found in Cyprian. Nor does ica - itself occur. 3. So also in -ao the ending of the gen. sing. of masculine -a- stems, o usually changes to v, preparatory to undergoing contraction to -au, e.g. Oetlav COLL. 66; Mapa'cav 29; but we find ao- in Kvrpay6pdo COLL. 79 and Aajario-do 58. On these see below, § 14, 4. 4. 1vPOice is read by Deecke, COLL. 45, 3, as a local variation of OvE~40ce, i.e. AveO8Ice; see above, § 5, I. The only difficulty with this reading is that the character for u " (IA/) has a superfluous horizontal line drawn over its top. This has led Voigt (Quaestiones de Titulis Cypriis, p. 282, and later in Bezz. Beitr., ix., p. 166) to conjecture an error of the stonecutter, whereby the horizontal line was made over, instead of under, the rest of the character. With that change we should get the regular syllabic sign for mil. This combined with the other characters gives a v B0qce, in which Voigt takes But I) it seems more natural to as the equivalent of /p. regard the horizontal line above the u' as an accidental scratch than as a mistake of the engraver. 2) ~Lv for ~ev, i.e. ~e, before an initial vowel, is not admissible (see above, § 7, 2, adfin.). 3) ~'Oqce is not the proper word for a dedicatory inscription, as this evidently is. The regular word is VOrflKce or aveoyKce, which occurs frequently (see above, § 5, I). 4) Voigt's objection to the form of the word (vy- for hv-) is not well founded, and is the result of a false conception of 147 pV Charles E. Bennett, 18 the relation of 3v- and av-. Voigt takes dv- as derived from ad- by some phonetic process, and refuses to believe that after becoming dv- could still further progress to vv-. 0v-, however, must be taken as an independent form (see above, § 5, I), and that it should become bv-, in a dialect where the relations of o and v are confessedly very close, is not to be regarded as surprising. That these relations were close is made evident not only by av- the regular change of final -o to -v, as noted above, but also by 'Atd(v)ra COLL. 147, for 'Ap ,(v)Ta (if correct; see § 5, 3), and vvOi for dVEOfIKce. Cf. the same phenomenon in Boeotian ice 'Apvav (for 'Ap wvras) COLL. 603; NLveivtLos (for NtLo/~ vtov) COLL. 485, 24. It is noteworthy that in Boeotian too the phenomenon seems to occur chiefly before nasals. 5. Spitzer (Laut. Ark. Dial., p. 17, note) cites Cyprian &vFavot COLL. 60, 6 as illustrating the change of o to v in the interior of a word. But 8oFe'vaL in the same inscription, lines 5, I5, certainly does not speak for this change, nor do other words in the dialect; so that the reference of the word to root as above (1), is undoubtedly correct. tv-, 10. v. We find oXjo'rj ~ in XiDo' (i.e. Xao-) COLL. 60, 29; Xo-at 60, 28; (i.e. o'0-X05a) 126, 2, where it corresponds to of the v other dialects, and presents no peculiarities. DIPHTHONGS. 11. aL. 8oFevat 1. Primitive Greek as appears in alFei COLL. 60, 31; (Att. 8ovat, for 6o(F)dvat) 60, 5, I5; ljao-ae 60, 3. 2. iJFats, the accepted reading in COLL. 60, io, is taken by Ahrens (Philologus, xxxv., p. 54) as from the preposition u (= dwri; see below, § 33, 5) with the adverbial ending -aas 148 Sounds and Inflections of the Cyprian Dialect. appended. 19 On the parasitic F see § 17, 2. This ending -a s, which appears nowhere else in Greek, Ahrens identifies with the Skrt. termination -dis as seen in uccais, fdnais. But these are instrumentals from -o- stems (see Whitney, Indische Grammatik, § 1112), and are formally identical with the socalled dat. plu. of -o- stems in -OLS (for *-Ocs, Idg. -ois; as Ze;U 2 for Zyss; vaiJ for vai, etc.; Meyer, Gr. Gr., § 298). On a more probable explanation of fpais, see below, § 33, 5. 3. Interesting is aXcov, COLL. 6o, 14, = Att. A "ov. This is by epenthesis for a primitive *aXo (Lat. alius), whence *aiXoio, alXo. " Cf the Hesychian gloss aiX6Tporov • aXXoLodrporov and the recently discovered Cyprian form 'AreiXov (for *'Aw~'Xov) in the inscription communicated by Deecke in the Berl. Phil. Woch., I886, No. 42, col. 1323. 4. has also been assumed by Spitzer (Laut. Ark. Dial., A p. 26) in preference to -a (i.e. -a) as the ending of such singular forms as ,dXa, -Xat , FoXryia, etc. Spitzer first (ibid., p. 25) attempts .to demonstrate for the Arcadian that the forms in -at in that dialect have the a short and not long (-CL not -a). He is convinced that -a could not have remained unchanged in Arcadian, but would have lost the iota and so have appeared as -a. His grounds for this are that final -4, (i.e. -,) loses its iota in Arcadian and appears as -9; e.g. Tv yxdar7, COLL. 1222, 14, for v'vyxavy. He also adduces Arcadian 'AgetCo, which he takes as for 'Ageyc , COLL. I 185. But this last is by no means certain. Spitzer's reasoning, however, is not conclusive, since final a, as, a do not necessarily all develop in the same way; and in fact even in one and the same diphthong sometimes retains the e.g. Ionic 7r PovXX (for T7 j3ovX^) -r , Mylasa, 355 B.c.; but 7 Uvry dialect one and the same and sometimes drops it; Erythrae, 394 B.c.; 8 7Lo- L , Samos, 322 B.c., Cauer, 2 Delectus, 510, 6. Hence it is quite possible that the Arcadian might have retained final -a&,and that such forms as Terye'a COLL. 1222, 34; 'Apfca&lat 1200, 3; 'OXvvrclau I183, 6; ap1lat 1222, 18, should be considered as ending in -a, so far as any phonetic necessity is concerned. 149 The only reason Charles E. Bennett, 20 for not taking them as ending in -& is found in the corresponding forms from -o- stems; e.g. p/pyot COLL. 1222, 49; dupo-a-ot 1222, 25. These latter must necessarily be regarded as locatives (to explain them as datives, with -oL shortened from -wo against all principles of Greek phonology), and so is after the same analogy the forms from -a- stems are most naturally taken as locative, and as ending in -- L. A confir- mation of this view is found in the similar Boeotian forms from -o- and -a- stems; e.g. &84y (i.e. &4LoL) COLL. 380, 3; Tra[l7 (i.e. Taidat) COLL. 385, 5. The above considerations, therefore, are not intended to show the incorrectness of Spitzer's conclusion in regard to the Arcadian forms in -aL, but simply the unsafeness of his method in reaching that conclusion. The same theory (viz. that -aI cannot stand in Arcadian) applied to the closely related Cyprian dialect, as Spitzer (ibid., p. 26) does apply it, leads to a false conclusion. Let us first look at the facts. We find in Cyprian the following dative forms: I) forms in -aL (whether -aL or -aL is to be determined). a aX atalkL COLL. 37, 3; 59, 4; 'AXa(pt)rpjdTaL 60, 8; lpodpaL 60, 20; 'A po8/Tat I, 3; FoXylaL 61; &L 6o, 8, 17, iaxat 60, 3; IIakat I, 3; 7r-eciat 24; MaXavijae 60, 17; 60, I8; llepo-e&raL 45, 3; T& I, 2; 40, 2; 60, 3, 6, 8 (twice), 17 (twice), 18, 24; 61 (twice); 62, I; Tnxa I7, 2; 27, 2; 28; 31, 4; 33, 2; 37, 3; 72, 2; Berl. Phil. Woch., 1886, No. 42, col. 1323; 1887, No. 12, col. 380; 'TXdTra COLL. 27, I; 28; 31, 4; 32, 2 ; 'AOdvat 17, 2 ; 'AXaLO-uTac Berl. Phil. Woch., 1887, No. 12, col. 380; 'ApLo-rayd pac ibid., 1887, No. 52, col. Deecke's earlier reading in COLL. 41, is now no 1644; 8oja&, longer maintained by him. (See Bezz. Beitr., xi., p. 317.) 2) forms in -a. 'AOdva COLL. 62; TvXa 74, 3; 120, 4; 62, I; 7~ 17, 2; 60, 8, 17; 'r IHacia, the correct reading of COLL. 9, according to 27, 2 is best Hall ('our. Am. Or. Soc., xi., p. 212). eXOXwX COLL. 135, which Deecke taken as nominative; 'ETEoMd4a (ad loc.) says may be taken as either dat. or gen. (with omitted -s; see § 20, I) is best taken as vocative; do- ja, 150 Sounds and Inflections of the Cyprian Dialect. 21 Deecke's earlier reading in COLL. 41 is no longer maintained by him; see Bezz. Beitr., xi., p. 317. Spitzer's conclusion with regard to the above forms (whether he had them all before him is doubtful) is this. Those in -as, while used as datives, he considers may be morphologically either locative or dative formations. Both these formations, he holds, were in case of -a- stems originally the same, the locative -a arising from primitive a+ , the dative -ai from a+ a. Either of these, according to Spitzer, must develop to -a in Arcadian or Cyprian. The forms in -as on the other hand he takes as locatives and as ending in -aL. This -aL he regards not as a primitive locative formation, but as developed secondarily from the primitive locative termination -aL (for a+L), after the analogy of the locatives in -os from -o- stems (e.g. otlcot). This may be expressed by the proportion: : 0/l L O Tvxa : Tvxat. Against Spitzer's theory must be urged I) There is no evidence that the -a- stems ever formed a 2 locative in a+L, which might give -a&. (Cf. Meyer, Gr. Gr., § 351.) Hence the locatives in -aL from -a- stems are not the successors of an earlier locative formation in -ac, but are best explained as entirely new formations. This being the case, the Cyprian forms in -a- could originate only from a dative -a&, not from a locative -aL. They are therefore datives. 2) If we view the forms in -a as locatives (i.e. as ending in aL), we shall have the anomaly of the locative taking on the function of the dative, and being used in precisely the same phrases and formulas, along with the continued use of the dative itself. The improbability of this fact is sufficiently great. Wherever one inflectional form takes on the function of another, it is to the exclusion of the latter, at least in the same function. Thus Arcadian &pyoL, locative used as dative, has supplanted Upyp; OaicLa similarly has supplanted al ca-t. So also Attic 7reiq, dual (borrowed from plu.), has taken its 151 Charles E. Bennett, 22 place in the dual to the exclusion of the regular formation relXeL (for Teliee) ; cf Att 'e /yet CIA. II., 652, B, 26 (Meisterhans, Grammatik der Attischen Inscriften,1 p. 61). But according to Spitzer's view, in such an inscription as Ta 'A6ivat COLL. 17, 2 we should have a dative article Ta limiting a also 6o, 8, 17 Ar; 7Tat at 7 /ao-tCLXFos locative noun. (Cf. ra 1(v) Co ipo(vt, where similarly Ta, dative, would stand in apposition with a locative 7at p.) Doric ra, Attic 7r) COLL. 60,4, 12; 3) The adverb 7rat (cf. and shows clearly that final -aL in 71, can only be for rr-a, Cyprian did not necessarily lose its L, and that other forms with -aL may therefore exist in Cyprian. The forms in -aL are therefore to be considered as datives, hence as ending in -aL, while those in -a are also datives, with the -a developed from -au, as frequent in many dialects. Ahrens assumes a locative in -&, and a dative in -a and -a. See below under Inflections, § 25, 5. 5. AXtjcOet(s) COLL. 74, I (cf. dAcLFeOetLL 60, 21) is obscure in its form and probably incorrect, as the inscription seems to be carelessly written. 12. EL. corresponds regularly to primitive Greek EL and Cyprian EL EL of the other dialects in ailFe COLL. 60, 31; FeLcva 76, 2; rrelcet (Att.1 Telo-eL) 60, 12, 25. 1. EL by epenthesis appears in the form 'AirrelXov in the to bilingual inscription communicated by Deecke in the Berl. Phil. Woch., 1886, No. 42, col. 1323. The original formation *'Arrh oXv, became first *'ArredXtwv and thence 'AvrelXov. Cf Pamphylian 'AriX(X)ova (for * 'Airrwova) COLL. 1267, 30; Syracusan 'Arrn-tX(X)cov Roehl, Inscriptiones Graecae Antiquissimae, 509, and the Arcadian proper name 'Arr'n-Xov COLL. 190, all of which represent the same form of the 1 Often incorrectly written Trifo; but REL- the is regular form of the root for the future and is assured by Attic inscriptions of the best period. 1 hans, Grammatik der Attischlen Insckriften, p. 24, 88. 152 See Meister- Sounds and Inflections of the Cyprian Dialect. 23 7reX-. root syllable, viz. The ordinary Cyprian form 'ArdX(X)ov represents the ablaut of the same root, while Thessalian "AwrXovv (ov = c), seen in "ArXovvL COLL. 368 ; 372 ; "AwXovvoc 345, 22, represents the weak form. Cf.the similar "Abstufung" in the name Iloo-et~ov, Laconian Ilooi& vt (i.e. Ilooc6av) Roehl, Inscriptiones Graecae A ntiquissimae, 83; Corinthian IIorts&v ibid., 20. 2. The EL of the form EFeLo-7)y COLL. 68, 2 cannot be justi- fied. Deecke (Bezz. Beitr., vi., p. 79) takes this as the equiva- ( o-ac, i.e. lent of the Homeric eio-q (cf vi es *EFt~oat). But or assuming this to be correct the change of t to EL the opposite (see on ZcoXdos, § 8, 2) remains to be proved for the Cyprian dialect. All the existing evidence shows that no such change took place. Moreover, the primitive form of the Homeric word was Flo-Fo9, as shown by the recently disFLF6otPpoov x., 53 ; FIl-FOV covered Gortynian inscription. Frag. B, 2. Hence the probable reading of the Homeric It is clear text is d-o-oat (for *d-Flo-.at; the E prothetic). that a form E-FLF- could not give Cyprian d-Fro--. The Ionic 11 too, of the termination, discredits the word, and the phrase, n0or' (for 7rwo', i.e. rpdos) eFelIo-' is not elsewhere found. On the general uncertainty of the context, see p. 2. 3. The ELin the first member of ALtFe Ole, COLL. 6o, 21, where some claim an old dative, is difficult of explanation. Attic Acetpi~fb CIA. I., 447, III., 53, et pass. is probably kindred. 4. On ELarising by contraction, see § 14, 9. 13. v, OL; L L, , L wL ; U . 1. In one or two instances Ev has developed from E before ;FP7Cc-aTv (for pro'TP-arv) COLL. 60, 3, and KcevevF,viz. in el This points o'v(for cevepodv; cf Homeric ceveoc) 20, 2. clearly to F as a bilabial and not a labio-dental spirant in between ev and a Cyprian, as does also the development of following vowel (see below, § 17, 2); though that it points to r 153 24 Charles E. Bennett, that pronunciation of F for all Greek dialects, as Meyer (Gr. Gr.,2 § 230) seems to conclude, cannot be admitted. There may have been a labio-dental as well as a bilabial F in Greek, just as in Germany in case of w. The v which was doubtless heard in the spoken language between every E and a succeeding F is not expressed in inscriptions except in the instances above cited. Elsewhere we find dFp/Tado-aTV 46; 47; icaTEFOp/Cov COLL. 60, 14; 60, I; 'Fpega 71; 'ETEFd(v)8po veFOOrTarTa 59, 2; NxcolXKX 4 40, I. The same development of o to on before F probably existed, but existing inscriptions show no evidence of any attempt to indicate this refinement of pronunciation. Cf.6oFevat COLL. 60, 5, I 5; 'AptLTo coFwv (questioned by Hall, your. Am. Or. Soc., xi., p. 216) 45, I. 2. oLhas also been assumed by Spitzer (Laut. Ark. Dial., p. 24, Note) as the termination of those forms from -o- stems which Deecke transcribes as -w (i.e. -w, dative). Spitzer takes these as locatives, holding that -o could not remain unchanged in either Arcadian or Cyprian, but must always become -o. That -oL did frequently lose its L in. Cyprian is beyond question. This is shown clearly by the frequent dative forms in -,(for the instances see below, 3, c). But it is not true that -wL always lost its L any more than did -aL (see seem both of them to be passing above, §I I, 4); -a&and -,o through a sort of transition period in the dialect of our Cyprian inscriptions. Moreover if, with Spitzer, we transcribe Cyprian -o " i" by -oL (i.e. locative), we shall be forced to admit a serious inconsistency in such phrases as Tol Oedo COLL. 74, I and 70o Geo 7O TX dTaL 27, I, where we should have a dative article limiting a locative noun. There is therefore not only no phonetic necessity for admitting -oL instead of -oL, but to do so would lead to an absurdity. The view of Ahrens, who and -o, involves claims a locative in -OL and also a dative in -oL no phonetic considerations and will be considered below under Inflections, § 26, 3. 154 Sounds and Inflections of the Cyprian Dialect. 25 3. The diphthongs -aL, -1L, -wL (i.e. -a, -5, -,) often lose their L and appear as -a, -4, -o. a) The examples of -a for -aL have already been given above (see § I I, 4, 2). So far as can be seen they reveal no law. Yet as we find forms in -a and -cI side by side in the same inscription, it is natural to assume a phonetic origin for the shorter forms. These may have originated before initial vowels, while -cawas retained before consonants, though the evidence is not sufficient to make this at all certain. b) Final stands regularly in the 3d sing. of the subjunctive for -1L, viz. in XMo-,r COLL. 60, 29; o-vXo-, 126, 2; ~4 -1 8pi4y' 60, 12, 24, 25. Cf. Arcadian Tvryadv? COLL. 1222, 14; 2 Cauer, Delectus, 44, 69. 'X77 1222, 26; Cretan icaraXv Deecke's "ASt1 COLL. 126, 2, which he takes for'"Asp is not s, certain (see § 4, I). Final LI is nowhere retained in Cyprian, so that Deecke's suggestion of /'Tg-tas the reading of COLL. 124, is un-Cyprian, apart from the general uncertainty as to the genuineness of the inscription; see p. 3. It is noteworthy that while we have frequent instances of -aL and -,L in Cyprian,' - L nowhere occurs, but always -, The Cyprian accordingly exhibits the same teninstead. dency as other dialects, in which -gL is the first of the improper diphthongs to lose its L. Cf. in the Therean inscrip2 vi., 20; e;rr7 viii., tion (Cauer, Delectus, 148), 7radly II., 28; 9; but TL KOLV&t ii., 9; 'Av8payodPat iii., 2. This is physiologically natural, as the L being more closely related in sound to a than to w and a, would more easily be absorbed by a preceding 1 than by either of the other two vowels. (Cf Brugm ann, Grundriss der Vergleichenden Grammatik, I., p. 121.) c) -WL (i.e. -w) loses the L and appears as -w in the following 'A7rdXXcwL COLL. 74, I; 7-&L O&CL 7 instances: TrC Oej 'TXdTra 27, I ; Tai CTXd77at 28; 'Arr6X(X)ow) 7vt 'rTXdraL 31, 4; v at 'AL/ckwt 'Oript) 45, I; ra 'A/rdA(X)w Tr'&olpL (i.e. 7 rG vpcevw 'H8a'AXa(L)wrptjdrat 60, 8; r7O '7 59, 3; r ipowtave vT& Xt j 60, 31; T0 'Oo'ip 72, I; T7t, EG&tL 'A'roN(X)oVt 72, 2; 0G[&t] T 'AAoX()Owvt 75, 3; TO) 'A76X(X) wvt TO Maytpio O 120, 2, 3; 7T O~cs'Berl. Phil. Woch., 1886, No. 42, col. 1323; TO) 155 26 Charles E. Bennett, 'Apto-raydpat 7O 'Ovao-tFolco Berl. Phil. Woch., "A(t)68(t) 1644. aCLoadro COLL. 126, 2 and 1887, TO No. 52, a(v)Opvrow 126, 3 are very doubtful. An examination of the above instances almost tempts to the conclusion that -C originated from -)L before vowels. Re- jecting the last two forms as uncertain, all the others accord with this inference, except T7 Ot&Cw Berl. Phil. Woch., 1886, No. 42, col. 1323; T7o Mayrppco COLL. 120, 3; Oe ToT I. 74, But forms in -,m also occur quite numerously before vowels; e.g. 7~CO'AXa(!t)rpLJdraTa COLL. 60, 8; 7roL Xe qjL 60, 31 ; rw Xet 60, 9; 7~T 'Ha- Berl. Phil. Woch., 1886, No. 42, col. 1323, so that the existence of the law suggested cannot be established. The absence of such a law in the occurrence of -a and -&a (see a, above) is also opposed to its existence here. 4. On av by contraction from -ao, see § 14, 4. 'AreiXowv 14. Contraction of Vowels. 1. a+ gives a in ijao-Oat (for ijdeio-aL) COLL. 60, 3 and by crasis in raw (for tra dr) 37, 2. 2. -+o. A Xdovo 9 COLL. 63; I8'paov 69; Aao'a 83 are all too uncertain to be considered here. 3. a+ o. 'APpoOa'dL COLL. 129, 130; 'AvoE 97; 7I/OH& 69 are all uncertain. 4. a+o. Final -ao in the gen. sing. of masc. -&- stems (cf. Homeric 'ATpelSdo) contracts to -a as in Arcadian (cf. Arc. 'AwroXXwviav COLL. 1231, B, 16), viz. in 'Apto-rayopav COLL. 28; 'Apcort'av 20, I; 21, Oeflav 66,; Mapcdcav 29; Nao-drav 2; 'Ovao-aypav 60, I, 22; '~rao-Ljav 17, I ; Ttayopav Berl. Phil. Woch., 1886, No. 41, viii.; 'Oao-ayo'pav ibid., x.; raco-aro'pav iv.; IIvvrayo'pav Berl. Phil. Woch., 1886, No. 51, xv.; T ~cayopav ibid., xvii. Kvrrpaydpoo COLL. 79 and AaarTo-ao 58, however, remain uncontracted. Deecke on the latter regards the termination -ao as also diphthongal, which is perhaps correct. Cf. Ionic abros, rarTa, advTo (for azbrot, etc.) Cauer, Delectus,2 510. 156 Sounds and Inflections of the Cyprian Dialect. 27 'Awgvajd, COLL. 6o, 18, gen. sing. from 'Atpygvja, 'Apto-rj a Berld. Phil. Woch., 1886, No. 52, xx., and EbFaYdp, 153, 154, gen. sing. of EbFayopa (cf. 'Ovao-a7 6pav above), are difficult of explanation and due perhaps to foreign influence. See below, under Inflections, § 25, 3. 5. a+o contracts to a in the gen. pl. of -a-stems. The only example preserved is drraoecvav Arcadian 'pywov~v COLL. 1222, 47. COLL. 59, 2. Cf. 6. +a does not contract in Cyprian, but in the Idalian Bronze Tablet (COLL. 60) E becomes L according to § 7, I, always with the parasitic j (see § 18, I), viz. in rpXvja, COLL. 60o, 9, 18, 22; Fsrrja 60, 26; aTcXr La 60o, 23; KcaTL7dEav 60, 27. acc. sing. from cTeXj9, Deecke's reading in COLL. 60o, Io, no exception to the above principle. This is not is aTreXiv, to be taken as a contracted form for rTeXa (cf TEeXLja 6o, 23), i.e. 'reCX?, with added -v, as in case of &(v)8pjd(v)ra-v COLL. 59, 2 (see § 29, I), but is rather the same formation as is seen in Lesbian BaordEX'v (from BapJorTXfl) COLL. 304, A, 44 (see under Inflections, § 29, 8). Hence the form is to be written &rTeX jv. Instead of Deecke's ? ice (= e'i ice) COLL. 60, IO, 23, Meyer 2 (Gr. Gr., § 113, foot-note) suggests ice. (On (v) see § 23, 2.) 17(v) This 'j(v) he takes as the Cyprian contract form of idv. We should thus have the same combination of ice and lv as in Homer; e.g. o3bp' Av ,du Kcev A 187. But it is inad- missible to assume contraction of Ea to q and is moreover unnecessary. Meyer's unwillingness to accept 4 as an independent particle (related to but not identical with el) is not well founded in view of the occurrence of " in this sense in the Cretan inscription from Gortyna, e.g. iv., 31 ; v., 9. Outside of the Bronze Tablet E~ remains unchanged, viz. in Oedowp COLL. 126, I; Nea- 76. 7. E+,E gives in yXe (for *eXe, i.e. *E-cey-e), Att. elxe, COLL. 60, 21. Whether the same contraction takes place in q the infinitive of -overbs is uncertain. 10, 22 writes Deecke in COLL. 60, Xivv, i.e. for *Xe-ev, Att. ExeLV. under Conjugation, § 32, I I. 157 See below, 28 Charles E. Bennett, Ev'Faetre COLL. 56 which Deecke reads as contracted form for e VaFe'ee (Bezz. Beitr., vi., p. 148) cannot be regarded as a Cyprian form. In the group -EEO- does not contract, but the group is EE simplified by apheresis of the first E. Instances of this are eeoic'Xo9 for eeocX'eo, (i.e. eeoicX Feor) COLL. 126, I; TtoevofXc o, for RevoFrceo9, COLL. 1246, cX oc 35 (cf Arcadian R, B, 12; XapacXko9 1246, B, 4); o-reos, gen. sing. for * orreoc (i.e. *o-rd-eo-o ) 31, 2; 32, 2. Deecke reads o-weo, here, assuming contraction of eo to o, but Eo does not elsewhere contract to o in Cyprian (cf Oeoic oc, TLoKXoKo; $7ao-icpdTeoc COLL. I8, 2, etc.), nor in the closely related Arcadian (see above). Hence the Cyprian form must be ordo , even though the genitive thereby becomes identical with the nominative. The Cyprian accordingly bears out the general principle assumed by Spitzer (Laut. Ark. Dial., p. 37), viz. that when of three successive vowels the last two are incapable of contraction, in the particular dialect where they occur, the first of the three disappears. In the same connection Spitzer formulates another general principle intended to apply to all Greek dialects. It is this: When of three successive vowels the two latter are capable of contraction, they contract and no further contraction with the first vowel takes place. This principle I believe to be unsafe and to be contradicted by an undoubted illustration This is gentaken from the Arcadian itself, viz. 8aquoprdy. erally incorrectly referred to a form auatoepyeos. But the second member of the compound as a nomen agentis demands the ablaut of the root, -Fopry-; cf. cXor- 6 'thief'; o-0or-d 2 'spy'; 7roLr- o 'attendant' etc. See Meyer Gr. Gr. , § 9. The Homeric poems, it is true, exhibit 8ctoepydl 383 et pass.; but this is to be regarded as of secondary origin. by the side of *8ato-opyo's. Cf the similar relation existing between Att. wevr~cdv'r-opov 'fifty-oared galley' and Ionic r7rVT?7CVcOrT-epo, -root ep- 'row'. So Attic inscriptions have TptacovT-epos by the side of the earlier Tptaicod-opos. See 158 Sounds and Inflections of the Cyprian Dialect. 29 Meisterhans, Grammatik der Attischen Inschriften 1, p. Io; cf. 2 Meyer Gr. Gr., 1.C. No dialect has preserved any instance in inscriptions of the original formation. -OPFOE in inscriptions written in the old alphabet, e.g. COLL. II1170, 2 (Elean); 1479, 15 contraction of -oopydo as of -oepyos. B&iytovp- (Locrian), if not actually for -opy 6e (as read by Bechtel in case of the latter inscription), may be taken as easily for the So also Attic yos points no more clearly to -ocprdo than -oopy 6s. The 2 Messenian dialect has Baptopyo's Cauer Delectus, 47, 119; so also the Achaean, CIG. 1542; Megarian, COLL. 3094, 19; Pamphylian, 1261, 3. In all these cases 8aat!opyds is to be derived from the primitive form *Bapto-opyo by aphaeresis of the first o. In other words, we have the same law here as in the Cyprian forms Oeo/cKeo9, T/otoicK o mentioned above. The facts I believe authorize us to assume at least for the Arcadian and Cyprian the following law: When of three successive vowels the first and second or the second and third are repetitions of the same sound, one of the repeated vowels disappears. This law also shows evidences of its operation even to a wider extent than these two dialects; e.g. Cretan lIptavo-d~ for -GEES cIG. 2556, 30; Ionic p/op&e for/ 3 ope'ew; so also the infinitives of contract verbs in -ci, -EI, -dO, *TL/aev (whence TrLLav) for *Tt/rLEev; *cXdev (whence kXev) for *4LeXv; puo-0'ev (whence for *t o-00'eeV. The above explanation of Ba, topy~ not only starts from pto-LOov) the form' demanded by the signification of the compound but explains its further development by a principle simple and natural and abundantly illustrated in Arcadian, Cyprian, and elsewhere. Spitzer's explanation (after Ahrens, De Graecae Linguae Dialectis, I., p. 234) refers the word to a primitive whence &atopyo' by contraction; thence, by shortening of the w, &atopyo's. This shortening of a long Sacpto-ep79s, vowel when followed by a liquid + consonant, though maintained by Brugmann (Grundriss der Vergleic/zenden Grammatik, I., p. 463), does not seem as certain, by any means, as 159 Charles E. Bennett, 30 the other instances of vowel-shortening adduced by Brugmann in the same connection, and is to be regarded as doubt6 pvvyt (from *o-T7p-vv-tc) and ful. Even if admitted for ob-T P/3oXXopa, i.e. *l6XvoLa (from *16-votat), it is by no means certain that it operated subsequently to the disappearance of F, as must be assumed for BSato-(F)epye. kXEroppyovro9 Thessalian (i.e. -Odrov) Mitth/eilungen des Deutschen Archaologischen Instituts, vii., 346, which is explained by Prellwitz (De dialecto Thessalica, p. 43) by the shortening of w (for oc) to o, is not certain and probably incor- rect. Lolling, in publishing the inscription, says: "Da der stein hoch eingemauert, musste ich auf sicherstellung der zweifelhaft und undeutlich bleibenden stellen verzichten"; so that confirmation of the form is needed. If correct, XetTopyov7ro might be referred to the influence of SaptopyovvVroV. 8. E+t remains unchanged in @Oerov ico COLL. 128. 9. E+ may be contracted in FETEL COLL. 60, I; 59, I; gXE 60, 9; Tre 76, I, though the character of the Cyprian syllabary makes it impossible to determine whether the vowels were contracted or spoken separately. lo. +o in the Bronze Tablet becomes LO in accordance with § 7, 1; viz. in Errd(v)a (for rred(v)r7a) COLL. 60, 9, 19, 22; i~(v)ra 60, 23. remains unchanged, viz, in OeoicKe'o COLL. Elsewhere 126, I; 'ETeoactia 135; ~eortip ov 42 ; TL/ KrcheFov 36; 64; co p ov 7Tao-tKpcp Studia Nicolaitana, p. 68; Tttoicpeeov Ber. Phil. Woch., 1886, No. 41, ii. ; viii. ; Gehocpireov ibid., vii. One exception is found in a late inscription, Berl. Phil. TqJpoKaEos 35 ; r7a-tcKpoeeo Woch., 1886, No. 42, col. 18, 2 ; 1323, where Neo~vLto9 becomes Norvto'or. The same inscription is characterized by the v-movable, an evidence of the late period to which it belongs. L (in accordance with § 7, I) in coo-t, 11. E+6± changes to (for wo-c, subjunctive) COLL. 60, 31. On Deecke's bpov~oi (i.e. 4pov&~o-t) 68, 4, see below, § 20, 2. I6o Sounds and Inflections of the Cyprian Dialect. 3I 12. L+ in the Bronze Tablet does not contract, but becomes jc (in accordance with § 18, I, c), viz. in AtLBerl. Phil. TWoch., 6 wr7Xj3'TCOLL. 60, 6. 1886, No. 41, ix. remains uncon- L tracted after the disappearance of the F. Elsewhere L+ contracts to L, viz. in T'72o-pL (for T7O'OJ-pLL) COLL. 45; 'Or-ipt 72. Hall (sour. Am. Or. Soc., xi., p. 216, 222) now reads mT 'Ovao-pLp and 'Ovaalipt in these inscriptions, which however does not affect the question of contraction. 13. o+o contracts to ,, frequent in the gen. sing. of -o- stems; e.g. apypo (for *apydpoo) COLL. 60, 6; Ttpo&4Guo 23, 3 ; 7O 29 ; 31, etpass; icareodpFcov (for *caTEFo 14. X6e COLL. III ; 133 ; Xoov 88, I; taFEo p ' coov) 6o, I. oToa tcov o103 are all too uncertain to be taken into consideration in this connection. 15. Compensative Lengthening. The extent to which this prevailed in Cyprian is uncertain, owing to the nature of the syllabary, which does not distinguish the long and short vowels. The question of compensative lengthening presents itself chiefly in the development of the group -avs and -ovs, and here the problem is still further complicated by the fact that the nasal is regularly omitted in Cyprian before a consonant in the same word (see § 23, I). Hence the syllables -a *se " can stand for -as, -as or even -avs. So also -o 'se ' may stand for -os, -cs or -ovs. Under these circumstances it is perhaps simplest to follow the model of the closely related Arcadian and write d(v)Oprov, acc. plu., COLL. 6o, 3 (cf. Arcadian Tse o-vvto-Trapvov, COLL. 1222, 51); so cda7ro9 60, 30; icao"-yvluroo 31; t'o(v)o' *ro 6 0, 3, IO, I I, 23, 30; 60, 3, II ; iX[a iVov 60, 3; fut. ind. E o(v)oa60, 60, 31 (cf. Arcadian cpovov', raperdTaov-a COLL. 1222, 5, 15) ; acc. plu. of -astems, vrs 28, 29; 71; 7ro-e 6o, 60, 28, 29, 30; FP~7Ta 60, 28, 29. If 8ty odolv (for 3 tydctOov, see § 20, 2), Deecke's reading in 1 COLL. 69, were certain and the inscription really a hexameter, 161 Charles E. Bennett, 32 w we should thereby be forced to admit that o by compensative lengthening produces in Cyprian, at least in this instance, since the metre requires a long syllable at that point in the verse. But Deecke's transcription of the inscription is unnatural and unsatisfactory, so that his text does not afford the basis for valid conclusions. See p. 3. There is less doubt in case of the frequent e" mi'; e.g. COLL. I, I ; 16, 20, etpass. This form, which might be taken COLL. 307), in accordance with the for d'L/ (cf. Lesbian A'ppe, Cyprian mode of writing doubled consonants singly (see § 24, 2), is shown to be ' i by the bilinguis, COLL. 65, which, by the side of the Cyprian syllabic signs ka ' ru' xe j e "mi ', has KAPVT EMI. The possibility that this latter may be for d~p' is not absolutely excluded, since even inscriptions written in Greek characters, particularly in the Old Alphabet, sometimes have , X, v, etc. for 1~, XX, vv. Like ?),L is 'AlvLija, COLL. 60, 18. Cf Attic 'AedwtaS. 16. Elision, Crasis, Apheresis, Synizesis, Diwresis. 1. Certain cases of elision are few; viz. d ' t COLL. 59, 3; KeTI w a-dr 'H&aX ov 59, I; wep' 'HdXtov 60, 27; and probably L' pleTrv 126, I. The first of these a&0' is not to be regarded as for arrv ct (on a7rr as the Cyprian form of arrd, see § 9, 2), since the elision of v is inadmissible. The oracle in Herod. vii., 220, ?7 /ea where the final acTU v of epL'KvoSvr aoryT aV pdco-t Hepetpo-ct, is elided according to Kiihner (Aus- fiihrliche Grammatik, I., p. 189), cannot be cited in support of such elision, since the reading is justly suspected. Cf. Stein ad loc. &b' is rather to be taken for a&r JS and referred for its origin to the period before final o in Cyprian Wb became v. Once formed, the phrase 4c' Jt continued as a stereotyped expression even after airo' became ' The nature of the elided vowel in KeT1rWv 162 ". d 'HsaXtwv Sounds and Inections of the Cyprian Dialect. 33 COLL. 59, I, cannot be determined, as the origin of the word is uncertain. (See § 34, 2.) 7rep' for crept in the third of the cases cited, rep' 'HdcXtov is poetical; but the reading seems certain. Cf.Pindar, Pyth. iii., 42; 7rep' arca ; Nem. xi., 51 Irepd 6 0. If with Allen (On Greek Versijfication in Inscriptions, p. I 50) we take COLL. 71 as a metrical inscription (see Allen, p. 46), we shall then have elision of the final Eof W or, although it is written in the text. This practice of writing the elided vowel is common even in inscriptions written in Greek characters. See examples collected by Allen, p. 127 ff. (aX(X)' ~7vx ' /cpCOLL. 68, 3, and 7ror' EFeola- 68, i, are omitted as too uncertain. See p. 2. The character of the Cyprian syllabary did not admit of the expression of elision except in case of words so closely connected in sense as to be written as one, like a "po i'= &a' a wot, etc. But most words were not so written. Inscriptions which show care even separate the words by a divisor. Thus words ending in -Ice-, -Se, -o-e, etc., would have to be written -ke -te', -se', even if by elision the final E disappeared, since the Cyprian had no way of expressing a final consonant without a following E. Hence the expression of elision in case of words written separately like cepcre (o " te ke "re ") ; ke r (8e -re e" ve "xa") was a practical impossibility. So also in case of final -a. 7wd(v)T' iXev, for instance, could be written only pa te" j e"ke ne ", which would give 7rd(v)Te TroTe erFpea (po "te" " This fact accounts perhaps for the apparent retention of final short vowels in cases where they might be expected to suffer elision. 2. Crasis is apparently certain in iral for Ti Err COLL. 37. In place of Deecke's T'TOo-ipt (for 7^ 'Oo-lp) COLL. 45, I, Hall (Jour.Am. Or. Soc., xi., p. 216) after a re-examination of the inscription in New York now reads 7o 'Ovao-lp. Yet it is difficult to reconcile this reading with Hall's original facsimile (Plate viii., 34) as reproduced by Schmidt (Sammlung Kyprischer Inschriften, xx., 6). 163 34 Charles E. Bennett, Meister, Berl. Phil. Woch., 1885, No. 51, col. 1604, reads 103 as ri-raco, i.e. 7 'Tac "des Ohrenkranken," and 104 as T4o7irwoT(, i.e. T'7arrro "des Tauben." Both these COLL. conjectures are extremely doubtful. 3. Aphaeresis of L is to be assumed in o' ('v) 73, i.e. of i(v) Deecke also assumes aphaeresis of a in T73 COLL. 6, 31. Oe tL aX(X)'; but a is written, and the reading labors under too many difficulties to be accepted as correct. So Hall's rd(v)Se 'v 'Arr6X(X)wv (Jour.Am. Or. Soc., xi., p. 223) = COLL. 76, cannot be regarded as certain in view of Cesnola's plate (Cyprus, Plate II., Io). Deecke's obydp r drrLc-rat COLL. 68 is best explained by aphaeresis of E (see Allen, Versification in Greek Inscriptions, p. 74) rather than by assuming a hiatus and shortening of the L before o,, though the latter is proposed by Deecke (Bezz. Beitr., vi., p. 80). 4. Synizesis is maintained by Deecke for 0eo COLL. 68, 4, and Geoly 68, 2, with reason, if these words really begin hexameters, which is not certain. Synizesis in doov in COLL. 88 is uncertain. 5. Diaeresis is claimed by Deecke, COLL. 68, 3, in o 'vo ", which he takes for ob. This seems hardly possible. Diaeresis in such a word would be surprising under any circumstances. The parasitic F (see § 17, 2) is not elsewhere found after o, and even if it were, we should expect the last syllable of the word to appear as vu ", not as vo .. The fact that no character has yet been found for vu "cannot be held to support Deecke's view. If the sound had existed, the character for it would have existed also. On a similar view advanced by Deecke for Aajar~oao COLL. 58; Aaja4ac9v 32, I, see § 18, 2. "Ajapot 31, 1 ; CONSONANTS. 17. F. 1. Initial F is regularly retained in Cyprian, always in the Bronze Tablet. The instances are Favag COLL. 18, I; 59, 2; 164 Sounds and Inflections of the Cyprian Dialect. 35 68, I; abbreviated to Fa" 154; Favo-(o-)ac 38, 4; 39, 2; 40, 76, 2; F 7 r'ja ('ros) 60, 26; Fe'reL 59, I; 6o, I; Foy (i.e. o-Fol) 59, 3; 60, 29; FOlKco& 60,6; FOlVO 73, I; FP~7Ta, (cf. Fprpa) 60, 28, 29; Favo-(o-)a9 Bezz. Beitr., xi., p. 315; I; Fecicva 316. The only exceptions are avdo-(o-)ac COLL. 33, I ; ETe 76, I; Aro. (doubtful) 150. The absence of initial F in XeL COLL. 60, 9,furnishes clear evidence, in view of its retention elsewhere in the same inscription, that the word has no etymological connection with Lat. vallis, as still maintained by Curtius, Grundziige der Griechischen Etymologie,5 p. 360. Medial F is also regularly retained, always in the Bronze COLL. 60, 31; aXFo 60, 9, Tablet. The instances are: aiFel 18, 21 (cf the Hesychian gloss lkXova 'iiTroL, Kv'rpcoc, where ov is used to represent the bilabial character of Cyprian F (see § 13, I); paao-tiro9 39, I; 46; 47; 59, I; 60, 6, 8, 17; 153; 154; 176; 177; 178; 179; AFe/ilEles 60, 21; oFpva 60, 5, 15; EFp6a 71 ; Fp~rdo-arv60, 14; 'EreFd(v)8pm 46; 67; erep- yeo-la9 71 eVFpfTard/-ar 60, 19; lepFo9 I, KevevFov 60, 4; wv IcaTEFOpK (i.e. iceveFov; see NLtcKOKXFcl 27, 183; I; 40; NICo Xe'pOs 7aoiFouLo9 193 ; § 'HBaXLFece 60, 2; edpFo(v) (cf Hom. (F)dp-o9) 60, I; 13, I) 20, 2; veFoo~7-aTa 179; OLFWo 59, 2; 60, 14; 'OvaCIFOLo9 27; 183 ; T LooFo'p I43 ; Ttco- rKXEeo9 36; 64; 'ApaTroFdvaf Berl. Phil. Woch., 1886, No. , Berl. Phil. Woch., 1884, No. 21 ; 41, xii.; 'Apco-roKX6Fr TteoFdvacro ibid.; 'Ova-tFooo Berl. Phil. Woch., 1887, No. 52, col. 1644; NKioKX F Sr Bezz. Beitr., xi., p. 315; P. 316; (cf.Lat. veho) Studia Nicolaitana, p. 67; IlFa (= d'io-a; cf./ 1 (f)ov, Lat. vivus) Prellwitz's reading of COLL. 134 (see Bezz. Beit., ix., p. -172). 'AparroKodfv, Deecke's reading in COLL. 45, 1, is not cerEFe e tain. Hall (Jour. Am. Or. Soc., xi., p. 216) after a fresh examination of the inscription in New York reads 'Ap'rdywov; but this does not seem possible, judging from the copy of the inscription given by Schmidt (Sammlung Kyprischer I schriften, xx., 6 a). &c aco COLL. 70 is extremely doubtful. 165 Charles E. Bennett, 36 EdEire, 68, I, has F, but the exact form of the word is uncertain; see § 12, 2. e;baETre 56 is impossible as a Cyprian word; but the F is certain. LXPA(X) Berl. Phil. Woch., lcaFL Tc LX() ticaFoc COLL. 29; 2 Woch., 1887, No. 12, 1886, No. 41, ii.; EabL&FoS Berl. Pil. col. 380, are Phoenician names. and 'AxaLFo6 communicated by Sayce in Berl. Phil. Z6 Woclz., 1884, No. 21, are doubtful, and need confirmation. Exceptions to the retention of a primitive medial F are more frequent than in case of initial F. We find the follow- Oprs pao-tXio a (cf p ao-L17Foc) COLL. 17, I ; 38, I; 40, 2; etFLw) 73, I ; 'Ereo154, 155, a, b; 156; 193; AZL (cf ALF &d4a (cf 'ETEFd(v)83p) I35; OeoKcXoc (cf NLKcoKcaEFqS) 126, I; iepo (cf leprFos) 38, 3; TtpoicXko 35 ; ALt Berl. Phil. Woch., ing: 6 1886, No. 41, ix..; /3 ao-ago, Bezz. Beitr., xi., p. 316; Nea-? 76. Noow/vo Berl. Phil. Woch., I886, No. 42, col. 1323, is a contracted form (see § 14, io) for Neopivtwos; cf vFoUo-7rac COLL. 59, 2. On 'HaXt jt COLL. 60, 31 instead of 'H3aXL~Ft, see § 18, 4. Several inscriptions seem to belong to a transition period and exhibit some forms with F and some without; e.g. in ()a ; in 39 TLOXadpLFov, COLL. 38 pa-Xioc, epo0s, but Fav /aolrXFo9, Favao-(o)aSc, but ijepc0j0; in 40 NKcochaEfs, Favao(o-)ae, but /3ao-tXoS0; identical with COLL. 40 is the inscrip- tion given by Deecke in Bezz. Beitr., xi., p. 316. 2. In addition to the F above mentioned the Cyprian has developed a semi-vocalic v between v or ev and a following a or , which it also expresses by F. The instances are: u8vFVOL (for 30-avot, root see § 9, I) COLL. 60, 6; E'Farypm 153; 8v-; 154; abbreviated E Fa . I55 b; 157; EFa" I155 a; 163; EpdF(v)OeFo 161; EbFd(v))eo' 159; Epda(v)O6, Ei~XOv 171, 172; EFflXOo(V)TroS 165; 167; 168; 3 The preposition viat , 6o, 10, 22, KcarTo-cevFaoe 31, 3. I56; 162; 169; 28, is also probably to be explained in the same way; see under Prepositions, § 33, 5. 2 evbFpyeJita, which Meyer (Gr. Gr., § 157) refers to this 166 Sounds and Inflections of the Cyprian Dialect. 37 category, does not belong here, but the F is part of the root. (Cf. Cyprian EFpe a COLL. 71.) oaa -XeFo(v)T709, 59, I, also mentioned by Meyer in the same connection, should be omitted. The syllabic text gives only pa "si 'le u, i.e. /ao-thev-; -FO(v)ro7 is conjectural. The development of this parasitic F occurs regularly between every v and a following E or a. As an exception must be noted vev~dtLevos (i.e. rw-r-ev acevo9; on u-, see § 33, 4) COLL. 45, 2. Deecke's reading here has been questioned by Hall (Jour. Am. Or. Soc., xi., p. 216), who suggests that the sign for u" here is a mistake of the stone-cutter for mi ", lacking simply the lower transverse stroke of the sign for that syllable. This mi " he takes for pc(v), with omission of the final v (see § 23, 2, 5), regarding it as the pronoun of the Ist person; see § 31, 2. But the omission of final v before a vowel is inadmissible; see § 23, 4. Hence I believe Deecke's reading is to be sustained. 2 Meyer, Gr. Gr., § 239, suggests that vev~dapevo may be for Fev adevor, comparing the Hesychian glosses vio-vL (i.e. FIeost)" c roXr; vdX~ (i.e. FXdl)"*o-icvX . But I see nothing to support this hypothesis, and should be inclined to attribute the glosses to a later stage of the dialect. The above phenomenon of the development of a parasitic F is not confined to the Cyprian, but occurs also in other dialects; e.g. Boeotian BaceiFaL COLL. 458; Corcyraean capto-rvFov7ra Roehl, Inscriptiones Graecae Antiquissimae, 343. 3. The F of TtpoxdptFo9 COLL. 39, 1; 193 and Kvrrpoicpd7 rFOv 26 is difficult to explain. These words are both -L-stems, and as such their genitives should be T t,toxdptov, Kvwrpotcpd- r That cannot have developed regularly between L and o seems clear. Spitzer (Laut. Ar. Dial., p. 51) suggests the 70oS. following explanation. It is to be assumed that the inter- vocalic F in Cyprian gradually lost its sound and disappeared as in other Greek dialects. Evidences of this have been given above in such forms as Sao-LgXo Ep~os, as against the earlier /ao-tjFov, lteprFo, etc. That the forms without F are in general the younger there can be no reason to doubt. 167 Charles E. 38 Bennett, Spitzer assumes that the spelling with F was retained in these and similar words, as an archaism, even after the had lost its sound. Cf in Latin the retention of C as G in Cajus, Cnaeus, long after C had assumed the sound of K. So in Cyprian he believes that written, even after after this analogy paorLXeiFo etc. continued to be Xao-aLXo, to be spoken, and that began TLoxydpFOS arose, and KvWrpoKcp TLFo though Ttloxdpto and KvrpoicpdtLos were spoken, the F being superfluous. This view of Spitzer has much to commend it, especially the fact that one of the inscriptions in which TLtFoxdptFoS occurs (COLL. 39) belongs clearly to the transition period when F was beginning to disappear (at least in the vicinity of Paphos), as is evinced by the form iep oS beside Favdo (o)as and pao-Xr$Fow. Cf. also COLL. 38 and 40. This period of uncertainty in the employment of F would furnish just the conditions for the rise of forms like Kvrpoicpireov and TwtoXaPtFOS. IIp'rFo Ben. Phi. W'och., 1887, No. 12, col. 379, if correct, is to be explained in the same way. So also the second F of EdeF(v)OeFoS (for Evde(v)0eos, nom. EvFa(v)Oys) COLL. 161, were the reading at all certain. 18. j. 1. Between L and a following a, E or L, a semi-vocalic L has frequently been developed, which is generally written J. This is often called the parasitic j. The instances are the following : a) j between L and a. 'AXa(.)rprjdra COLL. 60, 8; 'AvLyt3a 6o, I8; 4 (v)8pt- jc(v)rav (cf.Att. av ptac) 59, 2; avoo-ja 6o, 29; 'Apto-Ttrav 74, I; FE7rja 20, I; aTeeXya (Ion. a Tevea) 60, 23 ; Ajalct (Ion. Frra) 60, 26; Ieprljjtav 60, 20 ; i'Jao-Oa 6o, 3 ; Ia7 pav 6o, 3; MaXavijat 60, 17; MLXcJdjwrvos 59, I; IIag as 15, I; Haqgja(v) 69; rem8tat 6o, 18; ITao 7ja 18; rTao jav 17, 1; I68 Sounds and Inflections of the Cyprian Dialect. 39 Trpxvwja 60, 9, I8, 22 ; Icare'jav 60, 27 ; 'ApLa7lja Ber. Phil. Woch., i886, No. 52, xx. It will be seen by the above examples that this change took place as well after the . which developed from an original E (see § 7, 1)as after primitive L; cf. FE7a, r pxvtja, etc. In the Bronze Tablet this j has developed without exception between every L and a. Elsewhere we find exceptions. Thus IIalas COLL. I, I ; 2, 1; 4; 5; 6; 7; 8; 9; io; II; 3 (cf IIablia(v) 69; IIalfja 15, 1); FoXylat 61; Azdle ,t(9) (cf. AdjaiE et(s) 74, I) 00oo; c(v)ptcd Berl. Phil. Woch., 1887, No. 12, col. 380; (v)>pcd(v) rav Berl. Phil. Woch., I886, No. 42, col. 1323. b) 1 between L and E. 12; HIIala epEvs I, COLL. 40, I; 33, ic9EpIs I; ' °EpY9 39, 3; i)epE~s Bezz. Beitr., xi., p. 3 16. This change does not hold for the Bronze Tablet, viz. in iepjtjav 60, 20. Other exceptions are lEp31Fov I, 2; iepgos 38, 3; Kvpted I193; lepo9 Bezz. Beitr., xi., p. 315. •j does not develop between and l. Thus we have'H&aXtjFeC COLL. 60, 2; 'HaXtijjt 60, 31 ; KET1gFes 6o, I, where we might have expected KeTtjq-, 'HSa fj7 -. Deecke writes these words 'HSaXdsFE9, 'H&atIejt, KET1eFES. With that reading we should simply have additional illustrations of the absence of j between L andE in the Bronze Tablet. c) j between 1 and L.6 The only example is 7rbo Xtjt COLL. 60, 6. On the other hand Berl. Phil. Woch., I886, No. 41, ix. The Pamphylian exhibits precisely the same development of a parasitic semi-vowel between L and a following vowel. This it writes (with Greek letters) as L; e.g. Fe'TLca (i.e. FeTja, Ion. ~-ea) COLL. 1267, 5 ; &tdc ibid.; itapoo-t, 'Eo TF/(v)&8ve (= 'Aoriv8tow) 1259. Between L and o, or L and w, j never develops in Cyprian; e.g. AZdv COLL. 73, I; 'Abpo&8to-i 86, 4; 'wot- 6o, 31. The , ztl assumption therefore of Spitzer (Laut. Ark. Dial., p. 51) of the forms Tt oxpptjoe 39, I ; 4povjool 68, 4, is without foundation; and Meister's conjecture of vd'i3Jov as new reading 169 Ckarles E. Bennett, 40 of COLL. 41, 3 (Berl.Phil. Wock., 1887, No. 52, col. 1644) is very improbable. 2. Besides this parasitic j we also find j in the proper names AjaT~to-ao COLL. 58, which Deecke suggests may be for AaTrlo-ao. But this is purely conjectural. Aajarf&a and "Ajapos COLL. 31, I ; 32, I, which Deecke previously took in the same way, are now read by him Bezz. Beitr., xi., p. 319, as Tdcippa and 3. Deecke's reading iepdjtjav in COLL. 60, 20 makes difficulty by the presence of the first j. This might possibly be taken as indicating merely that the E and L were spoken separately, i.e. as Et. But EL elsewhere in the same inscription is not so written, viz. in ~XeL, line 9; FEreL, line I ; and it seems to me better on the whole to write icpjjjav and to consider the as developed from E, just as in case of the Doric adjec2 tives in -qcos for -ELos (see Meyer, Gr. Gr., .§ 67) ; e.g. Cretan 7rpvTavtov CIG. 2554, 51; Delphian laptca CIG. 1688, 14; and apXo~. , the Ionic substantives Lavretla (i.e. aXJO'iit, pavTrytit. Cf also Boeotian ,tavrwnta) COLL. 494, 2. Cyprian iepjjav is identical with these formations except that it retains the j, which in the other dialects disappears in the preceding q; or, we may assume that a new i has developed between and following e. 4. 'H~aXtnje (Deecke writes -jL) COLL. 60, 31 is still more perplexing. We should have expected here 'H&aXiF, dat. sing. of 'HBabevs; cf. 'H&aciFe COLL. 60, 2; KeTcFES. 60, i. The form 'HaXLrti cannot be derived from 'H3aXd Ft by any phonetic process, nor can I see any plausible explanation of its origin by association or analogy. 5. Change from to j before a vowel has been assumed by Deecke in case of the diphthongs EL, OL, UL do-~a for doeia in j4b (for (i.e. do-cit) COLL. 41, 3; Sojae (for 8oc&a) 41, 3; 7 bv1 1) 126, 3. The two former of these examples are no longer maintained by Deecke (see Bezz. Beitr., xi., p. 317), and the last one, 4)0'7, not by any means certain in its is reading. I believe therefore that we are not as yet justified in claim- L 170 Sounds and Inflections of the Cyprian Dialect. 41 ing this change of Lto for the Cyprian. Yet the change is probable enough in itself and must have occurred in other dialects as preliminary to such forms as Arcadian 7roeV'ro (for wrojivrw, (for i.e. 7rotEvTO), COLL. 1222, 9; Lesbian l&Kcdo 8calios) i.e. 304, A, 44. 'ipc, § I, 2. see 6. On japc for iapd, i.e. icKaom, 19. In Cyprian, t corresponds not only to t of the other dia- (?) lects, 'viz. in Ov (cf Att. f'do) COLL. 60, IO, 23, 28; ,e'o 150, but also sometimes to y of the other dialects, viz. in 'earth,' COLL. 60, 8, 17, 24 and raO6s (= tyaOs) 0 (= ya) 37, 3; 59, 4. These two latter forms probably developed a parasitic 4 after the original y, and this rye then regularly became g. This a-Oat (for *,y7echange has an analogon in the word edo-ao-0at) given by Hesychius as dialectic form of yeo-ao-Oat. Cf. the Arcadian 'XXELV (for *y XeLv)). /dXXLew, Hesych. 20. 1. Final e disappears in a few instances, viz. 'Ovaowpo jaOcieet 7&t COLL. 75, I; 'A... (for 'Ovaao-opo9 'A...) 7wcb)74, I; ALcdOe c(s) Fa... I00; Kca(s) ('and ') (for Aja eT , &(v)rl 60, 5; iad(s) uev 71 I; 7a(,) Favdo(o-)a 3 8 , 4; 7a(9) bXjpOv 60, 5 (twice); FAX(X)ica(s) 'OvaoL~daXa 120, I; in composition wro-e o'xevov (for roo--eXd evov, i.e. 7rpoo-exo'devov) 60, 19, 21 ; "EvFd(v)O(s) 1 I63 ; "Av(v)a(s) 'Acd(v)a 147; Studia Nicolaitana, p. 68; o8 d( 6piv COLL. P'L(X)lica(s) or 60, 12, 25 may be either for e i4( 8pd g (with omitted -s) ae o may be the article used as relative. The above data do not warrant us in drawing any positive conclusion as to the law of this change. Omitting 8 ie 8ptfy as capable of other interpretation, the remaining twelve instances present six cases of the disappearance of -sbefore 171 Charles E. Bennett, 42 vowels, five before consonants, and one where no other sound follows. We can hardly infer from this that the disappearance of -s took place through the medium of its change to the rough breathing, since in that event we should expect it to be confined to those cases where the following word had 2 an initial vowel. Meyer (Gr. Gr., § 305) in judging of the assumes that the pecuBoeotian proper names in -EL for -ELS liarity originated before initial vowels, and was subsequently extended in its use. The same may be true for the Cyprian. But since this peculiarity is confined almost exclusively to proper names and in them is found in the whole field of Greek inscriptions, it may be better to assume a weak pronunciation of final -s in this class of words. This, however, would leave Cyprian Tra, cd, and 7ro- unexplained. It should be noted that, while ica and ra (for Icdc and ras) are found in the above-mentioned instances, the full forms iccd and ra, are frequent, e.g. a avcdo(o-)a COLL. 33, I ; rTa cbXOha 59, 60, 6. Ta 'Ereo&dLa" 7r*Ot COLL. 135, which 3; Icas' Deecke (ad loc.) suggests may be for either Ta 'Ercosd4avs or &at 'Ereo8adLa, is better taken with Dittenberger as Ta, 'EreoSdc4a, rZ0, in which 7-a is the regular Cyprian form of the Homeric 7) 'take,' and 'Eeomnia is vocative. Cf t 347 KucXhoq, T7, 7rTi olvov. This would add another illustration of the influence (already beyond question) of the Homeric diction upon the Cyprian vocabulary. Cf. Deecke-Siegismund in Curtius' Studien, vii., p. 262; Smyth, Onz Poetical lIVords in Cyprian Prose, Am. Your. Phil., viii., 4. ebXOXa CQLL. 27, 2, which might also possibly be taken for a genitive or dative (see § 25, 5), is, I believe, best taken as Cf. &pa 'Avoho COLL. 97; capa Ad Berl. Phil. a nominative. Woch., I886, No. 41, ix. Deecke's IcK 7rOdr COLL. 68, I; [Ica] OvaTo~ 68, 2; and o-l() (for T7i) 126, I, are doubtful. 2. Deecke (Bezz. Beitr., vi., p. 8I; p. 147) seeks to estab- lish the loss of intervocalic v-, or at least its change to the rough breathing in two instances, viz. povnwi (for Opove'moa) (for *&tc(,Cosv, i.e. *td'vrTOtv) COLL. 68, 4 and &stcolv 172 69. Sounds and Inflections of the yprian Dialect. 43 This change is well assured for other dialects, e.g. Laconian ev ic)ais for Edv 5 o-acat Cauer, Delectus 2, 17, 15 ; 'o-Tpacto for CAr,1o-lo-pa-ro9 22, 8. IcaLvta, i.e. Kcao-ivr-r A -yr- The Hesychian glosses ; o-aa/ia, i.e. o-o-cay ; aov rcTragov, and others, given as Cyprian, also point to the change in question, at least for some period of the Cyprian dialect. But the reading in the two instances claimed by Deecke cannot be regarded as certain, especially in view of the numerous difficulties of the context; see pp. 2, 3. Moreover, the prevailing usage of the dialect in all other cases is to retain the - arising secondarily by assibilation of T, such as we have in cpov&co-L and 8Lyco-oLs, e.g. E/3ao-s 31, 2; I'o(v)o-c 60, 31 (the, (v) not absolutely certain). /ao-X ed Cf also cao-ly7rlTos 60, 14; I17, I. It is, therefore, impossible from existing inscriptions to admit the existence of any such change of a- to the spiritus, as is insisted upon by Deecke. The glosses given by Hesychius are doubtless to be referred to a much later period than that to which our inscriptions belong. Cf. the parallel case of the Laconian glosses exhibiting rhotacism cited by Hesychius, /ovayop (i.e. Sovayds), y&vop (i.e. yJovos). Yet this change is not attested by a single pre-Christian inscription. See Mtillensiefen, De Titulorum Laconicorum Dialecto, p. 54 f. 3. Meister in the Berl. Phil. Woch., 1885, No. 51, col. 1604 (cf. Baunack, Die Inschrift von Gortyn, p. 23), seeks by circuitous combinations to explain another word on the above principle of ' for a-, viz. ro in rot -r racC&, von derm Ohren- kranken, his proposed reading of COLL. IO3. This preposition ?rol he explains as follows. From primitive Greek rorl arose in the Arcado-Cyprian dialect *voo-/ by assibilation. Before vowels this appeared as w r, e.g. Arcadian rroo80s COLL. 1222, 9. After the separation of the Arcadian and Cyprian, iro in Arcadian excluded its sister form *7roo-, and we accordingly find 7ro alone; e.g. wr ro&t 1222, 54; 6 wroo-caTvSXa 5 fl 1222, 38. But roo-i and wro were both retained in Cyprian. The latter occurs in iron OdpFo(v) COLL. 60o, 19; gros 'cvr 6o,- 19; wre IIao-ayopav 6o, 21; *roo-1, 173 Charles E. Bennett, 44 however, first lost its intervocalic a- (for which Meister compares 4pove'wl already discussed above), becoming 7ro~' and then Trol. It is this latter form which Meister reads in ro TcOTacoJ COLL. 103. This Trol, he considers, became still further reduced to wro- in 'ro-exdevov for vro-eXcevov COLL. 60, 19 (cf. Arcadian 7roe'v7ro for (- roL-e'mTo). This would remove the necessity of assuming loss of in this word, as explained above, 1. Against this view of Meister's is to be urged I) Assibilation of T in case of 7roT'i, though naturally to be expected, is not attested by any Greek dialect. We find 7roTr in Homer; *7roo- is unknown. 2) If the form %*ro had originated from 7rorT we should expect it to remain *joo, since the arising in this way is not wont to disappear. Cf. eflcoc (primitive form FicaTL), fao-t(s (from *Tst). a 3) The example, which Meister cites to illustrate the disappearance of o- arising from T before L, viz. povdowi, we have already seen above (2) is quite doubtful and opposed to the clear laws of the dialect. Argive -ot, which Meister cites 2 (relying evidently upon Cauer, Delectus , 62, 9 and Etym. Mag. 678, 44) is not sufficiently assured. Locrian 7ro Tdov, which Bechtel defends in COLL. 1479, 14, is taken by Allen (De dialecto Locrensiunm, p. 67 = Studien, iii., p. 271) and Roehl, Inscrzitiones Graecae Antiquissirnae,322, b, 5, as a mistake of 6 the stone-cutter for r? rT in which ro&-is by apocope for v, 2 wrorl. Cf. Meyer, Gr. Gr., § 299, note. Allen compares Locrian KAITO, which he takes for dTr T, COLL. 1478, 46. 4) As to the origin of wroeXodEvov from wroteXoytevov by the disappearance of the L (through the medium of r), such a change should be accepted cautiously, even were the existence of Trol proven. We have no instances of the Cyprian treatment of the r (j) developing from the second part of diphthongs (aL, EL,OL, v) unless perchance 0dj9 COLL. 126, 3 be such an instance. That certainly would not make for Meister's view, but would lead us rather to expect wrojexdFevov. 174 Sounds and Inflections of the Cyprian Dialect. 45 21. Indogermanic q 1 apparently develops irregularly as ir (in- stead of T) before e and L in several words :-1. 7reio-eL (Idg. root qei-), COLL. 6o, 12, 25, corresponds to Attic TreiOeL (on this and not ticeL as the correct form, see above, § 12, ad in.), fut. ind. 'shall pay.' Attic r-lcoeL repre- sents the regular development of q. Cyprian 7rela-el has undoubtedly borrowed its w from other formations of the same root, where w. was phonetically justified, e.g. *rd-Trot-a (perfect), 7roLi7 'pay.' Cf. Thessalian a r7rec-daov (i.e. aroTELc'aTO) COLL. 1332, 28, where the same irregularity occurs. 2. wrecaEpwov, COLL. 59, 2, i.e. re()ac poov (see § 23, I, 2) gen. sing. of re(p) q-d4epov, 'five days' period' (cf. Att. 7revO t1epov) points to 'wpre (Idg. penqe) as the form of the numeral for 'five' in Cyprian as well as in Lesbian. Here also the (for T) owes its origin to the influence of other w primitive formations from the stem penq-, e.g. -et7dras, where the .- before. a was regular. 3. In orto-L (= o0-rts; see § 22, 2) v7rt- an adverbial is formation from the pronominal root qi-, which like W7ro-EL (see above, 1), ought regularly to appear as - L-. The - is to be explained as borrowed from forms such as owcos, droTrepoc etc., where .r for Indogermanic q before o and o is regular. 22. Assibilation of T before L. 1. This occurs as in Attic in the verbal ending -(v),cL (see above) for -v.r, and elsewhere. The examples are Ego(v)o. 3 for Igovrc (Att. Ifovo-c) COLL. 60, 31; 't(v)o- 60, 31; &#/ aa(Att. c3aao-tc; see § 24, I) 31, 2; 32, I; Er& 73, I and a 7rr0oTt 26, 2. (v)r 60, 5, et pass. retain the , as in all dialects. 1 Following Brugmann's use of this character in his Grundriss der Vergleichenden Grammatik. 175 46 Charles E. Bennett, cdrt has been assumed by Deecke, as the full form of the elided /cKr' 'and' in KaT' 'H3aXlov COLL. 59, I. If this is correct, the form would belong with ETC and J(v)Ti. At all events we are not justified in assuming that the form icds 'and' originated from c~d7r by the latter's becoming 'icide, whence (before vowels) KcdS. So also wred COLL. 60, I9, 20, 21 is not to be explained as the ante-vocalic form of *rood (for 7roTt), since 7roTt so far as known never assibilates its T. The s of ros' must be explained in some other way; see § 33, 3. rroT', which Deecke reads in COLL. 68, I, by elision for ro-i, is perfectly consistent with the existence of rwo' in Cypritn (see § 33, 3), but the context is so doubtful that small probability attaches to this form. 2. The indefinite o-1 (for Tss) occurs COLL. 60, IO, 23; and S'rtcam 6o, 29. This is irregular, since initial T.before L is not assibilated; yet the form is certain. Possibly, fT, an as enclitic, was so closely connected with the preceding word as to be felt as a part of it. In this way the T became intervocalic and so changed to a-. This is the explanation of Meyer, Gr. Gr.2 , § 299, and in support of it may be cited Att. arra, which developed from the primitive nom. pl. neuter of 71, viz. 'Td, in such phrases as Xp4/aTrd 7-a. The two words in such instances were so closely connected as to be treated like one. Hence XphLad raabecame regularly p awa. a This was felt as xprl.aTr' arTa, so that TrTa arose as an independent word. The only objection that can be urged against this explanation of o-l? is that Hesychius gives us o-1 as an interrogative pronoun in the gloss a /o'Xle* 71 OEXets. 8 Deecke's 't in COLL. 68, 3, is to be rejected. The reading is uncertain, and the form highly improbable by the side of 0~-. 3. Deecke reads 7reTt COLL. 68, 3, as vocative of 7roT, 'lord.' The word occurs, however, in 26 as 7do-Ta, with regular assibilation of the T. The fact that we always find rroan in other dialects would certainly tend under any cir176 Sounds and Inflections of the Cyprian Dialect. 47 in cumstances to discredit 'rrdrtL Cyprian, especially as T in this dialect regularly suffers assibilation; but the assumption in the that wrorL existed beside irrdo-, same dialect is entirely untenable. Another fact which makes against Deecke's reading is that the word never has the sense of 'lord' in Greek, a sense which he attaches to it in the present instance. 23. Loss of Nasals. 1. Before a consonant in the same word the nasals v, , y were always dgpped. This is generally indicated by putting the omitted nasal in parenthesis. The instances are the following : I) Omission of v. d(v)8pjd(v)rav (Att. avpLdvra) COLL. 59, 2; a(v)Opch7ros 60, 3; &(v)71 60, 5,I5, 'Ad(v)a (?) 17; I47; A(v)rTiatpov 28; drod(v)ra 60o, 9, 19, 22; 'ETeFd(v)8p 46; 47; E6Fd(v)Ol 163; EbFd(v)OeFov 161; 162; EbFXlo(v)roc 165; 167; 168; 169; 16(v)7a 60, 23; 'Ovdo-a(v)ro 30; 7rrd(v)ra 60, 19, 22 ; 68, 4; TaXd(v)Tov 60, 7; rd(v)&e 60, 26; 76, 2; 88, I; r(v)8e 59, 2; 60, 13, 25; 72, I; cIa(v)raro-'o 81; 83; 'ApwTrdoa(v)Tro 10, &(v)8pLcd(v)rav Berl. Phil. Woch., 1886, No. 42, col. 1323; la(v)8pua Berl. Phil. Woch., 1887, No. 12, col. 380. 2) Omission of . re(p)ae,dpcv COLL. 59, 2; 'AXa(,)rpejdrat (cf the present "Alambra, twenty minutes' ride west of Dali," Cesnola, Cyprus, p. 87). new reading Meister's conjecture of of COLL. 41 (Berl. Phil. in his (t,)fld[vrL] Woch., 1887, No. 52, col. 1644) is quite uncertain. 3) Omission of y. Probable in "O(y)ia(v)Tro COLL. 60, 9; IIa(ry)cpa- 62, 2. In Pamphylian inscriptions v disappears similarly before , or a in the same word; T under such circumstances changes to 8, e.g. re8eicaiKca (i.e. rrevTecaLecKa) COLL. 1267, 5 ; Fi yo&L (i.e. dEdyovrT) 1267, 16, 20; yevwa ('Ao'rrev8tor) 1259. 177 8 1267, 20; 'EaFFe'tFvc Charles E. Bennett, 48 2. Certain short words ending in a nasal, and closely connected in thought with the following word, omit the nasal, as in the interior of a word. These words are the forms of the article PLV Tdv, (= ftev, i.e. I) r7v, T-Jv; the preposition iv; and the pronoun Fee; see r(v) Xy&pov XpaVdo The instances are evov COLL. 60, I8; r7(v) xpavodevov 6o, 9; 60, 18 ; 60, 21; r7(v) 7T(v) 7rocxo',evov 60, I9, 21 ; To(v) 7T(v) Apv- rci7aov 6o, 20; as relative in To(v) /AtFeiOeLt tLaov 60, 19; r-(v) 2) § 31, I). Tov: E&tde(v) 126, 2 is improbable; see § 31, 4; Tca: Ta(v) rrTdtv 60, I ; Ta(v) d8XTov 60, 26; T7(v) Odo' 60, 27; TA(v) 814baro(v) 69; Ta(v) Feticva 76, 2; 3) Tov: Two(v) rralov 60, I I, 30; T^(v) Kcaoyv/pTOv 6o, 14; 4) iv: t(v) TXat 17, 2; 27, 2; 28 ; 31, 4; 37, 3 ; 59, 4; 72, 2 ; Berl. Phil. Woch., 1886, No. 42, col. 1323; 1887, No. 12, col. 380; 1(v) T&7L COLL. 60, I, 3, 8, 9; 1(v) at 60, 8; 1(v) Maxavijat 60, 17; 1(v) iA(p)t 3 os 60, 20; of ('v) T7 the L; see § 16, 3) 60o, 31; (with aphaeresis of 5) 1-"t' pt'(v) /lce Kar I, 2; 2, 2; probably also in o-l(v) Tvxa 120, 4. The above words always lose the v before a consonant, without exception. The forms Tov, Tiv, etc., occur only before vowels. The omission of v has also been claimed by Meyer (Gr. Gr.2, § 113, note) for jv (i.e. &dv). Meyer would read j(v) Ice in COLL. 60, IO, 23. But the existence of "v has already been shown to be improbable (see § 14, 6), and Deecke's reading (= el) is sufficiently justified by the occurrence of ' in Cretan. 3. Loss of final v before an initial consonant in other cases than those above mentioned is to be accepted with 1 Tdv COLL. 60, and liXFo(v) Tro 60, 21 seem caution. Odpro(v) certain. But other instances given by Deecke are doubtful, 178 19 Sounds and Ifections of the Cyprian Dialect. 49 viz. vab(v) rd(v)6e COLL. 41, 2, which he no longer maintains (see p. 3), and Ta(v) 3 baro(v) 8i0ao(v) IIaolbja(v) ye COLL. 69. This latter instance might possibly seem worthy of acceptance did we thereby gain a reading which commended itself in other respects, which is not the case. The individual words of the passage are several of them strange, and the sense which Deecke attaches to them (Bezz. Beitr., vi., p. 146 f.) is forced and unnatural. Equally improbable is Hall's at(v)Opc rwo(v) Oea (Jour.Am. Or. Soc., xi., p. 220), which he reads COLL. 68, 3. in place of Deecke's i(v)Opo7re Oe 4. In COLL. 126, 2, Deecke would even maintain the disappearance of final v before a vowel in &0'e(v) "AS4. This is also assumed by Hall (Jour. Am. Or. Soc., xi., p. 216) in his reading dp(v) e d~4evo9 in place of Deecke's ievcdI4evos (i.e. rwev7dpcevos'; see § 17, 2) COLL. 45; further in Kvveow(v) QbdwLo(v) do-ela (Jour. Am. Or. Soc., xi., p. 226) COLL. 87; 'Avr o (ibid., p. 225) COLL. 83. But in none of these cases does any probability attach to the reading. 24. Other Peculiarities. 1. Triple consonance occurs in '8Paa-tv COLL. 32. So also is the preposition &( used invariably before initial consonants, 7Ta 60, 6, 24. iE ats 60, 5, I1, 24; 2. Doubled consonants are regularly written singly, viz. in avda(-)as COLL. 33, 1; Favao-(o)at 38, 4; 39, 2; 40; 'A/(/p)Gi e: 61 ; "Av(v)as 147; 'AirdX(X)wov 31, 3 ; 32, 2; 59, 3 ; 72, ltX()lXia(s) 120, I; 2; 74, 2; 75, 3; 77; 78; 120, 2; jLp(p)t8So (?) 60, 20; 29; IIdot-r(7r)o9 194; FLX(X)kiaFo Bezz. Beitr., xi., p. 315; at(,) o7rdrwop (?) 85; Fava(o ()ac p. 316; LX(X)i/caFL Berl. Phil. Woch., 1886, No. 41, ii. ; doubtful is 4X(X)' COLL. 68, 3; Mavao-(o-)i, Berl. Phil. Woch., i886, No. 42, col. 1323. Instead of 'Arrdo(X)wvt the principles of the Cyprian syllabary would admit 'AnrXovt (cf. Thessalian "AwrXovv, i.e. "ArrXwov COLL. 368; 372; "ArXouvos 345, 22). On the other 179 Charles E. Bennett, 50 hand Cyprian e 'mi" might possibly be taken for 6 ( )c, except for the evidence of KAPV3E EMI the Greek text of the bilinguis COLL. 65 (cf. § 16). This practice of writing doubled consonants singly is not peculiar to the Cyprian, but is found more or less frequently in most archaic inscriptions of every dialect. Cf Syracusan 'AriXov Roehl, Inscriptiones Graecae Antiquissimae, 509; Megarian 'ArdXoaLw ibid., II; Pamphylian 'AreXova COLL. 1267, 30; 7-EtFeo-a 1267, 6 and the list given in Meyer, Gr. Gr.,2 § 287. 3. N-movable is found in a few late inscriptions, viz. om Kcev 'Atdo-o po Berl. Phil. Woch., 1887, No. 12, col. 380; KoKvcevdts and 8vOl7cevv Mavao(o)qj Berl. Phil. Wock., 1886, No. 42, col. 1323; in the two latter cases before an initial consonant. 4. The Cyprian syllabary has no character for the roughbreathing, which is generally supplied in accordance with .the vulgar usage. Initial rrT for rr appears in 7rTo7-cX COLL. 6o, 2, 4, 7, 15, I6, 27; 7r-7OXJL 6o, 6; wrTXv 60, I, all doubtless to be referred 5. to Homeric influence; see § 20, I. Ocepo-e 6. Primitive pae retained in is ance with the regular law. 0epat (i.e. 40'4pac?) COLL. COLL. 32, 2, in accord- Cf on the other hand Arcadian 1222, 8. 7. Hall's lja (Jour.Am. Or. Soc., xi., p. 225) for Ojd, i.e. Oed, his reading in COLL. 85, cannot be admitted. The change a- of a to found in late Laconian (see Millensiefen, De Titulorum Laconicorum Dialecto, p. 56) is not probable for Cyprian; and goddess in this dialect is expressed regularly by Oc's (fem. as well as masc.), e.g. COLL. 60, 27. 8. cvtepvat, COLL. 68, 4, if correctly taken as a collateral form of *cvpepvva, represents the same change of p to r as that seen in icvLeprYrTf for icvpepv4TT, Etymologicum Mag- num, 543, 2, where it is referred to the Aloeit. concerning the form, see § 32, 12. .8o Further Sounds and Inflections of the Cyprian Dialect. 5I INFLECTIONS. DECLENSION. 25. Stems in -a-. 1. On the gen. sing. in -an and -ao of proper names in -as, see above, § 14, 4. 2. Feminine -a- stems have everywhere -as in the gen. sing., e.g. 'AOdvac COLL. 60, 20; vaad-(a)as 33, I; e vxwOa9 59, 3; Favao(o-)at 38, 4; 39, 2; 40, I; 'Ovao-tucvpas 34, I; s I, I et pass. ; TtpoKn;rpaq 23, I; LXoic 22, I ; 'Apuo-ToKv7rpaq Berl. Phil. Woch., 1886, No. 41, vi. No trace is anywhere found of a fem. gen. in -av such as occurs in Arcadian (e.g. oliciav COLL. 1233, 3; fcalav 1222, 12, 25), where it 7rpa9 is borrowed from the masculine. On the occasional omission of -s in the gen. sing., see § 20, I. 3. A peculiar gen. sing. of masc. -a- stems is found in 'At/Tnvdji COLL. 6o, 18; 'Ovaoq-tsXd 120, I; 'Apto-Tarj Berl. Phil. Woch., I886, No. 52, xx. The formation can hardly be Cyprian. It is possibly the result of Doric influence; cf. Cretan 86Lta'o- (for -o), Inscription of Gortyna, v. 35. This expla- nation at all events seems preferable to that of Deecke, who believes that Cyprian -ao could sometimes lose its -o and appear as -a. 4. Eiayo'po (cf. on the other hand 'Ovao-aybpav COLL. 60, 2 I, 22) COLL. 153, 154, is referred by Meyer, Gr. Gr., § 345, to the influence of the Ionic dialect, but such Ionic gens. as Aap/Jra7ypeow ('Ebycepic 'ApXaLOooYLuc, 1884, p. 86) certainly do not speak for an Ionic Ebraypo, although -W(by contraction for -o) does sometimes occur in Ionic, e.g. 'AvvLCcK for 'Avvui~wc Roehl, Inscriptiones Graecae Antiquissimae, 381, c, I I. 'Ad(v)7 COLL. 41, which Meyer also explains in the same manner, is no longer maintained as a reading by Deecke; see p. 3. 181 Charles E. Bennett, 52 5. The dat. sing. ends sometimes in -cu (i.e. a); sometimes in -a. On the relation of these two endings to each other, see above, § 13, 3, a. The instances (with the immediate context) are as follows: '7a 'AOdva T&L COLL. 62; 7xa 74, 3; Xa 120, 4; " a 'AOdvaL 17, 2; 7-ral /ao-tXJFos T7 60, 8, 17; T~ IlIabia, the correct reading of COLL. 9, according to Hall o(v) (Jour. Am. Or. Soc., xi., p. 212). An examination of these examples reveals the fact that the dative in -a is more usually found in those cases where it is accompanied by another dative of the full form in -cL. Cf. the similiar use of -w for see § 26, 3. -w ; Ahrens (Philologus, xxxv., p. 13 f.) considers the forms in -a~ as locative, when they are accompanied by a preposition of place. But the locative in Greek is not elsewhere used as such with prepositions, and there is no ground for recognizing it here. 6. The gen. plu. in -sv (by contraction from -wv; cf Homeric Oewov, X4-Ta wv) occurs in drayopevav COLL. 59, 2. a 7. On the acc. plu. see § 15. 26. Stems in -0-. 1. The gen. sing. in -o (for -oo by contraction; see § 14, 13) COLL. 60, 9, 18; apryvpw 60, 6, 13, 15, is frequent, e.g. 1;XFw 23, 3. 17; Topo8tco 2. A peculiar gen. sing. is found ending in -o instead of -w and occurring interchangeably with the latter. The instances are: 'AtAlEXtIKcov COLL. 59, 3 ; pyTpwv 6o, 7, 25 (cf. Apytcp 60, 6, 15, etpass.) ; Apvwv 60, 19; 'ExeT tipgv 59, I ; Oo 8c;pov 42 ; OeoTl/wv 42 ; KeTLwv 38, 2; 'HaXlwov 59, I ; ,Lto-98v 60, 4, 5, 15; 'Ovaov 21, I (shown to be genitive by the recently discovered "OvatLoc>t Berl. Phil. Woc ., 1886, No. 41, iii.); 'Ovao-TKlWrpov COLL. 60, 2, I I, 30; 'OvaolkXov 60, 24; 7re(/I)capipwv 59, 2; TaXa(v)r w 6o, 7; yxjpwv 60, 5, 15; IxoICdrpov ;6o, I; Ncogviwv Berl. Phil. Woch., 1886, No. 42, col. 323. &v in COLL. 60, II is not to be taken as gen. sing. 182 Sounds and Inflections of the Cyprian Dialect. 53 (as given by Deecke, Index to COLL. I., I) but is rather the a~ 7raaa TSb, Cf gen. plu. modifying the preceding 7rat&ov. 'OvaoXAov 60, 23. The explanation of this genitive formation is exceedingly difficult. The view of Ahrens (Philologus, xxxv., 13) that an original formation in -os has changed its -s to -v, for which Ahrens compares the Dor. ending -Es (e.g. b yoic Kapsc) by the side of the ordinary -J.E (Xyop'ev) hardly needs refutation. The view first advanced by Deecke-Siegismund (Curtius' Studien, vii., p. 232) identifying this formation with that seen in the Arcadian genitive rovi (COLL. 1222, 38), which they took as Trov-i, is hardly correct, since Arcadian Tav[v]l in the same inscription line 53 points to a suffix -vL. The existence of this latter seems also to be confirmed by the Thessalian forms in -vE, 7r-ve COLL. 345, 20, et pass.; rd-ve 345, 23, 45. More plausible than Ahrens's view is that put forward by Deecke (Bezz. Beitr., vi., p. 71). Deecke thinks the ending -wv arose by confounding the gen. sing. in -c with the gen. pl. The v in the latter (see § 23, 2) had an extremely weak sound, according to Deecke, so that the form apparently terminated in -co,at least when followed by an initial consonant. Hence after the analogy of J(v)Opdnrw, i.e. -co(v) as a pendant to a(v)Opirov in the gen. plu., we find also in the gen. sing. J(v)6pcrrov as a pendant to J(v)Opcwrow. Deecke refers to the early Latin accusatives, sed, mid, ted, which are correctly regarded by him as having developed from original s, ;mi, to after the analogy of the duplicate ablative forms sed, se; tgd, mi. Cf. Osthoff, Zur Geschichte des Perfects im te; Indogermaniscen, p. I28; Stolz, Lateinisce Grammatik, mid, § 90. Against this view of Deecke's it must be urged that except in the few words already mentioned above (§ 23, 2, 3) final v does not exhibit a tendency to vanish in Cyprian. Even before consonants it is regularly written, e.g. wrai ov 60, Hence the assumption is not justiI I; Kiao-cyv roV 70, 14. fied that final v in the gen. plu. was characterized by the "ausserste Lautschwache" which Deecke claims, and the con183 54 Charles E. Bennett, clusion drawn from this assumption, that there existed two forms in the gen. plu., one in -w and one in -wv, is therefore equally without foundation. In the absence of these duplicate forms of the gen. plu. it is difficult to see how this could have furnished the motive for the new formation. Others, as Leskien (Berickte der Saicksiscken Gesellschaft der Wissenschaften, 1884, p. Io5) and Brugmann (Griec/zische Grammatik, § 94) suggest an independent ending here, -m or -om, which appears in Eccl. Slavonic. This is improbable. It is not likely that the Cyprian -o-stems had two inherited genitive formations in regular use. The one would have almost inevitably supplanted the other in the ordinary language. Latin famiiids beside stellae is a rarity; dedbus and filidbus have a special reason for existing; whereas these two genitives in -w and -yv exist side by side in the same words in the same inscription. It is therefore more reasonable to view the gen. sing. in -wv as a purely Cyprian development, the result of certain influences or associations which cannot as yet be determined. 3. The dat. sing. ends sometimes in -eL, sometimes in -w. On the relation of these two endings, see above, § 13, 3, c. An examination of the examples given there, reveals the fact that the ending -o is used only in those cases where it is accompanied by another dative of the full form in -wL,or by one ending in -aL or -L. This fact tends to discredit Deecke's reading of "A(c)r7(L) ato-adrWo(L) COLL. 126, 2 and rc2(L) 0 p;rr0w() ibid. 3. Ahrens (Philologus, xxxv., p. 13 f.) considers the forms in -CL as locatives wherever they are accompanied by a preposition, and writes them -oL. Cf his view of the corresponding ending -a in case of the -a- stems; see above, §.25, 5. 4. On the acc. plu. in -oS, see above, § 15. 5. A locative sing. in -ot seems to occur in IIagbo COLL. 56, I and 'H8akeof 62, I; possibly also in 'ApvwcXoi 59, 2. a(P)- 184 Sounds and Inflections of the Cyprian Dialect. 55 27. Stems in -L-.. rTdoXL forms the dat. sing. rLjX, (i.e. 7ro't,; see § 18, I, c) COLL. 60, 2. This represents the primitive formation. So also the contracted 'Oo-lp COLL. 72; 45, 2. See § 14, 12. 28. Nouns in -e1s. Deecke writes the oblique cases of these as -0Fos,-EFL, etc., or with disappearance of F, as -6os, -~, etc. He evidently assumes -Eos to be the primitive formation. This makes a difficulty in explaining the long vowel in the penult of these words in other dialects, e.g. Bceotian Oeo-retor (= 06o-rt)os), COLL. 494, 16; Thessalian /ao-tXeto9 (et = ) 345, 2, I I; Lesbian /aao-X?ac 304, A, 13; Ionic IIXovT~o CIG. 2665, b; Elean fao-cXae (for lao-tXes ; a for as frequent in Elean) COLL. I 152, (for #acx-7eo, laac~Xia, ao-AtXda 3aacrtX,, 3; Att. fao-tXias, by quantitative metathesis; cf Old Attic oliceov, given in a law in Lysias Io, 19). Assuming -eFos, etc., as the q 8ao-ctX&os, original formation, the long vowel in these forms can be explained only by compensative lengthening. But certain as a, few instances of this phenomenon seem to be, e.g. do<6 for *ac (cf. Lesbian a5oc), yet the existence of a uniform law, by which a short vowel, standing before F, is lengthened when F disappears, cannot be established. Such words as oo, IhcXEo, yXvecoc, ods , etc., in fact, veos (for vEFos), WrXoo, are so numerous as to seem rather to disprove it. It is better, therefore, to assume that the original stem of these nouns ended in -. , not -E-. have originally ended in -Ms. The nom. sing. then must This developed regularly to -evs; cf. /3os for *SowD ; vin0s for *vadv (Ionic vyi3i is of secondary origin after vies); Xodyot for *Xory (cf Skrt. gatdis). See Spitzer, Laut. Ark. ,ial., p. 30; cf Meyer, Gr. 2 Gr., § 298. 185 Charles E. Bennett, 56 The oblique cases were originally -Fo9, Deecke is therefore wrong in writing -gFL, etc., and pao-LE'Fo9, KCTt'ee.. The Cyprian is the only dialect which has preserved the primitive formation intact. All other dialects have dropped F. The instances of the formation are pao-ALFos COLL. 39, I; 46; 47; 59, I; 60, 6, 8, 17; 153; 154; 176; 177; 178; 60, 179; 'H~acFES 60, 2; epriFos I,I ; KeT yFES I. As to the Cyprian forms which appear without the F (see § 17, I), they must be considered as retaining q, if the theory advanced below concerning the origin of iep?,j correct; be see § 29, 2. I; The instances are: 40, 2; I54, 155 a, b; 156; 3ao-tXoc 193; COLL. 17, I; 38, tepgot 38, 3; $aotX os~ Bezz. Beitr., xi., p. 3 I6. 29. Othe9*Peculiaritiesof Declension. Peculiar are the accusatives 1aTqpav (= ijarTpa) COLL. 60o, 3; a(v)3pLjd(v)Tav 59, 2; Ber. Phil. Woch., 1886, No. 42, col. 1323. 1. Brugmann (Grundrissder Vergleichenden Grammatik, I., p. 198, and Gr.Gr., § 77) suggests that possibly the -av of these forms is to be regarded as the development of -mm: i.e. the primitive ending -m developed before it the vocalic qz, just as in Sanskrit *pjdm became * fdnm whence p~dam. Cf.also (whence the secondary nom. rrrvta instead of Greek vo'rvTwY It seems much rTvis. cf Skrt. pdtni) from *rd6ovienm. ; * more natural, however, in view of the acc. sing. d-v, Lt-v (for /ke, see § 31, I) to regard the v as borrowed from the accusa- and -o- stems, as if to mark more closely the tive of the -aaccusative character of the form. Cf.Thessalian ic 6 vav (from The late forms vvicTa-v and kv8pa-v COLL. 1332, 40. ic )v) which Brugmann cites admit of no other explanation. An analogous phenomenon is seen in the verb where a primitive 3d plu. imperative ypadrt T first inserts a pluralizing v, producing