MA S TER 
 NEGA TIVE 
 
 NO . 92 -80599 
 
MICROFILMED 1992 
 COLUMBIA UNIVERSITY LIBRARIES/NEW YORK 
 
 as part of the 
 "Foundations of Western Civilization Preservation Project" 
 
 Funded by the 
 NATIONAL ENDOWMENT FOR THE HUMANITIES 
 
 Reproductions may not be made without permission from 
 
 Columbia University Library 
 
COPYRIGHT STATEMENT 
 
 The copyright law of the United States - Title 17, United 
 States Code - concerns the making of photocopies or other 
 reproductions of copyrighted material... 
 
 Columbia University Library reserves the right to refuse to 
 accept a copy order if, in its judgement, fulfillment of the order 
 would involve violation of the copyright law. 
 
AUTHOR: 
 
 RADFORD, ROBERT S 
 
 TITLE: 
 
 STUDIES IN LATIN 
 ACCENT AND METRIC 
 
 PLACE: 
 
 L^' 'J 
 
 DA TE : 
 
 [1 904] 
 
COLUMBIA UNIVERSITY LIBRARIES 
 PRESERVATION DEPARTMENT 
 
 Master Negative # 
 
 ».t 
 
 BIBLIOGRAPHIC MICROFORM TARHFT 
 
 Original Material as Filmed - Existing Bibliographic Record 
 
 Restrictions on Use: 
 
 BKS/PROD BoakES FUL/BIB MYCG92-B32022 
 R&izord 1 of G -- Record added today 
 
 4- 
 
 I D 5 
 
 PC 
 
 DCF:? 
 XMTs? 
 
 RTYP:a 
 CSC:? 
 GPCs? 
 REP:? 
 
 ST :: p 
 HOD: ' 
 B 1 
 CPX 
 
 ,-^ 
 
 m 
 
 H 
 
 m\ 
 
 RRs 
 
 040 
 
 100 
 
 245 
 
 260 
 300 
 LDG 
 
 FRMs 
 BIMR: 
 F I C r ? 
 
 f-;b I : ? 
 
 COL : 
 
 Acquisitions 
 
 NYCG-PT 
 
 MS 5 
 
 NYCB92-B32022 
 
 96 6 B BLT:am 
 5 XX L senq 
 ss FD: 1904/ 
 
 nm>-\ xzYmc 
 
 1 Rad ford , Rober t S , 
 
 10 Studies in Latin /*iccent m^d Metri c« f hi:mi croform:] . 
 
 C s - 1. n , "= I bs - it * , == I c 1 90A "\ . 
 
 r33.] -64 p. 
 
 ORIG 
 
 G5--07-92 
 
 Fl... 
 ATC: 
 COM 
 
 AD 
 UD 
 
 5 0!!>-Q7~92 
 1:05-07-92 
 
 '•■> '-N •-> 
 
 1 1... C : ? 
 EHL. : 
 
 /-V /^ "^ 
 tit 
 
 Us? 
 
 GEM 
 
 BSE: 
 
 TECHNICAL MICROFORM DATA 
 
 TT 
 
 FILM SIZE:__3_? 
 
 IMAGE PLACEMENT: lA @ IB IIB 
 
 REDUCTION RATIO: ^LLK_. 
 
 DATE FILMED:__6_-_7_-_i2^ INITIALS ^_J^Sj. 
 
 HLMED BY: RESEARCH PUBLICATIONS. INC WOODDRIDGE. CT ' 
 
c 
 
 Association for information and Image {Management 
 
 1100 Wayne Avenue, Suite 1100 
 Silver Spring. Maryland 20910 
 
 301/587-8202 
 
 Centimeter 
 
 12 3 4 
 
 iijiliijiliiiiliiiiliiiiliiiiljiiiljiiiliiy 
 
 wm 
 
 Inches 
 
 m I I 
 
 5 
 
 iliiii 
 
 8 
 
 liiiiliiiiliiiiliiiil 
 
 iiiiiiii 
 
 ' ' I 
 
 9 10 11 12 13 14 15 mm 
 
 llllllllllllllllllllllllllllllllllllllllllllllllllllllllllllUIII 
 
 T 
 
 TTT 
 
 1.0 
 
 1^ 2.8 
 
 1^4 II .y 
 Sf us, 
 
 lib 
 fa ^, 
 
 IttbUk 
 
 1.4 
 
 25 
 2.2 
 
 2.0 
 1.8 
 
 1.6 
 
 I.I 
 
 1.25 
 
 P^ 
 
 n 1 I I 
 
 MRNUFFICTURED TO fillM STPNDRRDS 
 BY APPLIED IMfiGE, INC. 
 
Extracted from the Trahsacti.ms of the America,, Philoloi;kal 
 
 Association, Vol. xxxv, 1904. 
 
 h^"^^ i^c^ i/^j 
 
 / 
 
 III. — Studies in Latin Accent and Metiic. 
 By Trok. R()15KR-r s. RADFORD, 
 
 ELMJKA COLLKGK. 
 
 I. Introduction. 
 
 In a former paper ( Tnui.. Am. Plnl. Assoc, 190,) I sot.^ht 
 to show that in tribrach word-groups, the first word of whk:h 
 IS a monosyllable, the accent receded upon the monosyllable 
 
 (A J. I N„ 98 ff.) that this is regularly the case also in 
 many dactyhc word-groups of the same kind, such as AV 
 
 Jacis, scd scio, a patrc. 
 
 The question at once arises, -What is the explanation of 
 this striking phenomenon ? Why does the accent recede so 
 cons antly in these groups ? Although monosvllables are 
 usually connected closely in pronunciation with the followin.^ 
 word. ,t ,s evident that this fact alone is insufficient as an 
 explanation, unless it can also be shown that the great 
 maj,,nty of these groups follow some usual arrangement of 
 words, some usual and preferred word-order. This proof 
 I have attempted to supply by pointing out {A.J.P XXV 
 256 ff.) that groups like hlc /wm„, liic dies follow the I -Fur' 
 traditional word-order by which the demonstrative pronoun 
 .mmechately precedes the substantive, groups like,;.!/,/.,.., 
 -P<o, fans the traditional order by which the object immedi: 
 ately precedes the verb, groups like sai rrus, scd ego the 
 raditional order by which the sentence-introducing coniunc 
 tmn inimechately precedes the subject, and so on. The prin- 
 
 Phr s s ,!;, ' •".°""^.y"=^b''-"' - "'-y frequent and usual 
 ^ene ""■'^=''■'^^ "''*'= ?'"'^ "Sis, guodfacis, etc., that a 
 .enera type ,s eventually established to which the accent 
 
 ther. • '"'■"''' '=°.T''''"'''"°"^ ''^ "^^de to conform. Finally, 
 here .3 some evidence to show that even in the case of poly! 
 } liable words a traditional order, such as that of objec and 
 
<J'^ 
 
 Robert S. Radford, 
 
 [1904 
 
 verb, sometimes admits in Plautus the recessive accent, e.^. 
 
 sgmm veiim. 
 
 In tie present paper I wish to show the influence of the 
 traditional word-order still further by treating the accent of 
 the conjunctions and pronouns in the Latin sentence. The 
 numerous peculiarities of accent and metre which these parts 
 Hi speech cxWbit, liave long been a cause of perplexity to 
 Plautine scbolars, and, with the notable exception of the treat- 
 ment of the accentuations iiium, isium, ipsthn by Skutsch, 
 have usually been inadequately explained. I shall further 
 discuss the accent in the traditional order adjective + noun, 
 and finally, since recent accentoal studies have considerably 
 modified and, in some cases, seriously unsettled the views of 
 Latin scholars in this field, I shall attempt to state concisely 
 the general relation which appears to exist between word and 
 verse-accent in Latin verse. 
 
 IL Accent of Conjunctions. 
 The grammarians repeatedly assert that the conjunction 
 
 qtiand& is often accented qitando (Scholl, Dc ace, p. 173 f.)- 
 In accordance with their favorite method they wish to make 
 the difference of accent depend, it is true, upon a difference 
 of meaning, and no two of them can agree just what this 
 difference of meaning shall be. Nevertheless the original 
 statement appears tO'^ be essentially correct, and is confirmed 
 by the independent statement of Donatus (Ter. Eun. Ill, i, 
 47) upon the accents siquando, n^quando. Similar testi- 
 monies respecting the oxytonesis of other conjunctions (/>////r, 
 fmmmm, miiem.,fii^sfqmm)Mt collected by Scholl, /.A, 194 ff., 
 17s. A monosyllabic sentence-introducing conjunction, as 
 I have shown elsewhere {A J. P. XXV, 259), regularly coa- 
 lesces with the verb, r»/. sfseio, si-vah, and, by extension 
 of' this usage, sometimes with other parts of speech. We 
 find three cases ol a similar proclisis of qnand§ki the critical 
 feet of Plautine verse, viz. Bm, 224 quando volt ; Cap. 86 
 
 <|uani<S fts; M€u, 664 naando* quid.^ In all these cases 
 
 1 A miimeral placed in the line denotes a fjot of trochaic verse; below the line, 
 a foot of iambic verse. 
 
 Vol. x-xxv.] Studies in Latin Accent and Metric. 35 
 
 quando^ coalesces with a following monosyllable, just as the 
 prepositions in the frequent combinations /r.^/.V;;,. praetd^ 
 spent, and the like. When a preposition or conjunction coa- 
 lesces with an iambic word, as in propter mare iRu u^ 
 quando sciet (C./. 406), quando lubet (4 times ^x'the result 
 
 ;;;? ZLT^r^' ^T"""' ^^^ ^'^"^^^^ '^ -gulany needed 
 for making the verse-close, and this is the true explanation 
 
 ot Its rare occurrence in the critical feet. Moreover, the 
 prochsis of quando is not invariable; for we find Am. 1007 
 quando satis instead of quando satis ; cf Cap ^'^e 2 
 
 With respect to the other polysyllabic conjunctions, an 
 examination of the critical feet yields a similar result v" 
 the prochsis of the conjunctions is admissible but not invari' 
 
 WHnn ' T ^""^ ^" '" '""^'^^^"^ ""^^^ ^^ -^Tt^^^e con. 
 r r^ ^R-rM'""^'" ^^ A^../..;.(7)v. 913 tamqua^m 
 me, where Ritschl writes unnecessarily tarn quam me • As 
 427 tamquam si); one of etst ( 7)v. 52; etsi scelestus est) • 
 one of ../.;./ (Syr. Sent. 39 siciit fax, Ribb. ; Meyer with one 
 Ms. : ita lit fax); three of nnnio ( Tri. 427 immo quas ; P/^or 
 644 inim6 malum hercle ; Poe. 151 imnio mihi); three of 
 etnrm (A/er. 728 etiam vis; Ad 279 etiam maneo otiosus ; 
 Syr. Sent. 557 etiam queri); two of siqui^ian {Mi. 624 siqui- 
 de*m te; St. 616 siquide^m mea); one of quidni (Mi cc. 
 quKln. fateare); one of quam obrem (Phor. 715 quam obr^m 
 dem); one perhaps oi postquam {Am. 806 postqua^m cenaVi- 
 nnis); two perhaps of .7///^ (/^^r. 162 tibi quii super est cf 
 
 » For this study I have used chiefly the dissertations of O Bru-^man D. 
 
 I87V kZZ i^ ] ^^' ' ^' ''^"^f'^^o ap. PL septenario, Leipzig, 
 
 •073, *>^^^\^r,De trochauis stpftnarii^ riauf ri-ill,. iR^n. t> r \ • rs 
 
 Wnwt^^^.^* • J .. '"'^'"'^ iieriin, 1SS2; W. Meyer, Beobachtunz des 
 
 /'"W,>, Strassburg ,g8c- Ahlb' r / /; . ^' ^'^'^'^^'^ ^' P^'^^^^- -ne 
 
 Lund ,nL. Ri,,Hl'^ / '^^^^^'^' ^'' ^'''''^''^^'^''^'^'^^ 
 
Robert S. Radford, 
 
 [1904 
 
 Podiaski, /./., p. 12; Ad 523 nfsi quia« prop^st; cf. TrL 938 
 nisi qiiiaa lubet).^ Also, in the case of apparently broken 
 dactyls with atque (Tri. 935 a*tque cunila; Men. 508; ML 
 958; cf. St, 701 du^nique se exornat), the assumption of 
 syncope is not necessary, but the proclisis of the conjunction 
 iS" an alternative explanation ; cf . the similar example with 
 skmt, p. 45. From this survey I conclude that the oxytonesis 
 of the conjunctions, which is asserted by the grammarians, is 
 confirmed by the evidence of dramatic verse.^ 
 
 III. Accent of Pronouns. 
 
 As early as the time of Quintilian the ancient grammarians 
 had noted the exceptional oxytonesis existing in the pronouns 
 and derived adverbs (in adverbiis fere solis ac pronominibus, 
 Quint. I, 5, 26), and similar statements are often made later, 
 usually with an explanation drawn from the fatal difftrcntiae 
 €Mmsa method (Scholl, /./., 170 ff.)- ^« t"^^ ^^ concerns 
 itlmm^ istmM, f^si/iii, a thoroughgoing oxytonesis of these pro- 
 nouns is indicated by the derivative Romance forms, and has 
 been justly vindicated for the Plautine period by Skutsch, 
 Ftfrii'/i^p. 130 ff.^^ Skutsch has also assumed that the oxy- 
 tonesis first arose through the proclisis of the pronoun in 
 such traditional orders as i/him-fairem, OMm-vidrt, and was 
 subsequently extended beyond its original limits. This ac- 
 coiinl is undoubtedly correct, so far as it goes, but the full 
 logical consequences of the proclisis of t'iie are not developed 
 by Skutsch. The phenomena to be discussed are curious 
 enough, and if each one of them be studied separately, as 
 
 1 Examples like quid-igitu«r (3Ie, 91 1; Tri. 333), with preceding mono- 
 syllable, are not incluctei in tlie above. For the frequent accentuation of i£^/ur, 
 eJepolj also aiiter, aiia upon the ultima in proceleusmatici, v. Ahlberg, Procei, 1, 
 
 %%. 
 
 8 Interesting also is the regular proclisis of the adverb in association with the 
 adjective, e.g. And, 120 ade6-modesto, adeo-vendsto, cf. the strict observance 
 
 of Ijiehmann's law seen in idm-ciiS {AJ.K No. icx)); Ep, 380 aliquanto 
 lubentids. 
 
 » A correct view was also taken by Conradt, De vers. Terent. struct.^ Berlin, 
 
 1S70, p^ ^. 
 
 Vol. XXXV.] Studies in Latin Accent ami Metric. 37 
 
 has hitherto been the case, they are of a kind to baffle the 
 most practised investigator, but I hope to show that they 
 form a simple and harmonious whole when studied to-ether 
 Hence in order to exhibit the accentual and metrical peculi- 
 arities of the pronouns in full, I shall first take the pronoun 
 cited by Quintilian, viz. qudUs or .jualh, and illustrate its 
 uses theoretically. A pronoun like qualis occurs as a pro- 
 clitic chiefly in two traditional word-orders, viz. pronoun -(- 
 noun, as qnalis lumto, qiialcs sciies, and object or subject + 
 verb, as quales vidct, qualis crat ,• more rarely in simple e.x 
 tensions or variations of the orders just named. By sayin- 
 that qualis is a proclitic in such cases, I mean that it coalesces 
 in pronunciation and in accentuation with the followino- w„rd 
 and that the pause which falls after most words (and'^which 
 may be termed the word-end), very largely vanishes in the 
 case of the pronoun ; in short, pronominal combinations like 
 those just named are commonly treated in Latin as quadri- 
 syllable words, and accented qualis-hovw, qualcs-sccs, quatcs- 
 vtdet, qualis-cmt. The consequences that flow from this 
 cardinal fact, that no full word-end falls after the Latin pro- 
 nouns and pronominal adverbs, are fivefold. 
 
 A. Although the accentuation of a spondaic word upon 
 the ultima is not permitted in the critical feet, pronominal 
 combinations like quaU's saus, quales videt enter these feet 
 freely. 
 
 B. Although the accentuation of a trochaic word upon the 
 ultmia is very strictly forbidden in the critical feet, pronominal 
 combmations like qualis /wmo, qualis crat, unde vans are 
 admitted freely. 
 
 C. Although it Is forbidden in all feet, except the first foot 
 of a colon, 1 to separate by a word-end the two shorts com- 
 posmg the thesis of a trochaic dactyl, yet dactyls like qudlts 
 hmto, gndlis crdt, undc vniis are admitted freely, since the 
 division of the two shorts is for the most part only apparent; 
 cf. the case of prepositions, as m />r6ptcr dmonin, iutcristds. 
 
 D. The Lachmann-Ritschl law, which forbids a foot to be 
 
 » I shall use henceforth the term • tirst foot ' freely of the first foot of a colon. 
 te. of either the first or the fifth foot. 
 
3$ 
 
 Robert S. Radford, 
 
 [1904 
 
 filled by a dactylic word-form, is waived in favor of all the 
 
 dactylic pronoiliiiial forms, vk. kmeeine, isiucine, sicine, etc., 
 iiifuSf and largely also in favor of the similarly used adjec- 
 tival forms, 0mma, omnibus, omniUm, 
 
 E. The initial syllable of ambo, omnis, ccquis, ciuSy huiiis, 
 etc., Jis well as of ////, «//, ipsi (cf. also etsi, ergo), being 
 BOW iinaccciited, is freely shorteeed after a Brevis Brevians 
 (for examples of this well-known use, cf. Klotz, Gnindz., p. 
 46 f. ; A hi berg, De corrept. Plant., p. 69 ff.). 
 
 The evidence in support of these uses will be presented in 
 
 A. OxYTONESis OF SPONDAIC Proxouns. — Exclusivc of the 
 very niimerous cases of /////;//, istiun, ipsum (Skutsch, Forsch., 
 p, 132 If.), also olim {And, 221), kuius {Foe, 389), etc., we find 
 In,, tie critical feet thirty-seven examples of the oxytonesis of 
 the spondaic and anapestic pronouns kaecin ( i ), quant us (4), 
 mquis^ (4), numquis (3), siqnis (4), qnisquam (i), aliquis (3), 
 aiim {2% n/Ins, muiins {2\ idem (2), nostcr, voster {i\ omnis 
 (4X mmb0 {l\ perhaps also vobis (2). To this number we 
 should add eight exanii>les of oxytone pronouns with res, 
 
 Hmecin: Ad. 379 haecin[e] flagitia, cf. 408; cf. An. 746 
 
 ,fsta€i%i. te oritiiSne ; As, 932 fstosci%ie patrem ; — qnantus : 
 Man. 1013 quantum mall; Per. 517 quantu*ni tu ; Cap. 51 
 quanti sunt; Ad. 700 quantu-'^m potest (a frequent phrase 
 
 which Fl. always needs for the verse-close (eleven times), 
 except once (Am. 971), hut which Ter. places more freely); 
 very exceptional is Poe. 534 qua^ntum ve^lis, corrected in ed. 
 eiai. to quantum Yis; — eeqms: Per. 108 ecquid meministi; 
 
 Ps, 482 ecquam scis ; An. l§ ecqui' maiorem ; P//or. 474 
 
 ecquid spei(iamb. oct); — mmrnqnis: Cnr, 516 numquid vis; 
 
 ^ 2 
 
 Hec, 272 numquid vis (iamb, sep.); And. 235 numquid nam 
 
 t 4 
 
 haec \&th^\—siqtns: And. 258 siqui% nunc m^ roget; Ad, 
 941 siquid te miius oret; Ci. in siquiM tibi opus est; Fp. 
 
 1 I assume throughout that the first lyllabic of uqmM, qmiqmid, nequid, num- 
 quU, siquid, etc., is long; /&r precisify m tHi exisiime of nempe is disprmed by 
 the non'occurrmei tf mmp{i) wiik eiisim, w fAe existence of ecquid is disproved 
 ty the non'occurreme pfecqu(a)^ <Yf ii(/«), etc. 
 
 Vol. XXXV.] St;^aics in Latin Accent and Metric. 39 
 
 449 siquM vis. cf. Am. 453 ; Terentianus Maurus 2288 siqu.'s 
 vel,-t (.ncomplete collection). Only in the first foot of a colo'n 
 do we find And. 333 s.quid poHes ; Poc. 1200 qu.'cquid sapit 
 -qmsquam: An. ;6 quicquam meliust mihi; cf. J/cr I02.' 
 neu quisqua^m posthac ; cf. V./. 346;~./i,nis: And 957 
 ahqu.s fors[,tan] md putct ; Han. 752 aliquot dies; cf Ad 
 509 in-aliqu6d magnum malum, cf. Rn. 575' ,'„.aliquo* tibi 
 and, for examples of aliqu.'d boni (mali) in non-critical feet,' 
 V. Lodge, Lex. PL, p. 93 ;_.,//,,,, Tri 458 aliud vis; And 
 189 aliam vitam affert ;- „«//„.. „//,,, p,,. ^^^ „^,„,^ ^^^ 
 est; cf. Cap.c^x nullum pen'culumst (quadrisyllabic word in 
 verse-close) ; Men. 594 ullu«m teneri vkli ; cf. Ep. 497 ulla 
 pec.inia;_/^^,„.. Am. 447 ide^m sum; cf. 808 m-eode^m 
 lecto;-„„^/,-;-,.^^/^;.,yV.^;-. 609 nostcr Chromes; cf Am 221 
 n6s nostras more (cretic); En. 4.8 di vcstram fidem : hdmi- 
 nem (a frequent phrase, which is elsewhere always utilized for 
 he verse-close (sixteen times) in PI. and Ter., cf. Ikugman 
 '•'- P- 30); cf. Ci. 550 filiani nostram sustollere (quadrisyl- 
 lable word); cf. Cap. ij; cf. also Am. 356 horu^nc servos 
 sum;-^w«„.. Eu. 1092 omne«samarent (subject); Mo 192 
 di deaeque omn^s me pessum.'s e.xemplis mterf/ciant (often 
 corrected, cf. Mohr. /./., p. ,9); Hau. 26 omnes vos c5rat6s 
 V0I6; cf. Am. ,013 apud omni«s aed.'s sacrds ;\dd .;„;,/„; 
 m. .;„„.. res {Cas. 506; Hee. ,94; Ad 364; Lucil. XXVIII, 
 10, nee. 738; 483) and, in the case of other pronouns, tan- 
 tarn rev^iTrt. 682), alias res (Hec. 826), ipsa re {And 359; 
 Han. 266), etc. A general or indefinite ' enclisis of res; such 
 as ,s sometimes assumed (Klotz, Gmnd,., p. 324, and Bnr- 
 Stan s JalnrslK, 1883, p. 427), seems improbable ;-^;«^^ • Ad 
 131 ambos curare; cf. Rn. 1104 hasce amba^s ut; — perhaps 
 //^/and vobis: As. 654 tibf-dab6 (double iamb, in colon-close); 
 cf. Per. 847 vobis dabo (iamb, sep., ace. to Ritschl, cf. Mohr, 
 U, p. 18, but anap. oct., ace. to recent edd.); Per. 855 vobis 
 do (iamb. dim. cat., cf. Mohr, /./., p. 24). ' 
 
 I I 
 
Robert S. Radford. 
 
 [1904 
 
 A similar oxytonesis may be proved for the iambic pro- 
 
 iifllliiliml forms «#iw, €0, meum^ meo, tmnty utnmi, etc. Thus 
 tbcse forms arc often (nine times) found in the third foot of the 
 senariiLs(BrugmaE, /./., p.lolf.), as Tri. 794 eas resignatas 
 
 siW ; €L 568; Tm, 85; TrL arg. 6\ ML 484; Turp. comjr. 
 130; J»/. 442(e.im rem); Jrii. 656 (meo); Ba. 344 iitruni 
 
 veliro (an apparent double iamb., which Ritschl, ProL 
 
 ccxviii, would emend). Similarly we find eum in apparent 
 double iamb, verse-closes, as Am. 991 eunvsequor ; Men. 880 
 atque eam-meae'; perhaps also Naev. trag.fr. 13 mea-manu 
 
 I nioriire, where R,^ needlessly corrects : mea moriaris manii. 
 B-., Oxytonesis of Trochaic Peonouns. — We find in the 
 critical feet thirty-five examples of the oxytonesis of the trochaic 
 pronouns ilk. Hid (8), iiik (4), iste (i), tmde (2), cf. tute, 
 fmtdm (SX ecgmis, §msqmis, siquis (13), mstr^ {l\ omnis ( i ?), 
 
 qmisqtiedy 
 
 Examples, in the critical feet, of ilii quidem (Ba. 103 ; 
 Men 540; M&. 375; Si. 561 ; Pkor, 754) and iiiic homo {Ep. 
 
 666; 3f€M. 992; MFi. 334; Ru. 1297) are cited by Luchs, 
 Cmam, fms. II, 4, and Hinms, VI, 279, respectively ; istic 
 kmi0 {Ep. 488) and nmiims imno (Ba. 808) occur in the extant 
 literature only in the first foot. Further, the well-known 
 
 Latin word-orders, by which the demonstrative pronouns and 
 
 * Ib coiwectfop witli the last cjcample, wfcicli is explained by the regular order 
 of the proBomis (p. 41), I wish to deny emphatically that the prec«'cl,ing elision, 
 m In aifmii), jostiiies or explains the double iamb., in the irratit)nal manner 
 assumed by Kkitz, GrmttA,^ p. 245 f. In such assumed cases the tluulile iaml). is 
 iustitied only when «i m^mmj^iiaSiif wkuk/arms pari of a worU-group, is really not 
 ei tiled ai ail, rtvfi ill iMem, as //rr. 495 cnio-ab» ? adcs; Av. 290 se-amet potest. 
 These cases »e them cpiite similar to the well-known dehortari (Poe. 674), de 
 h6rdeo (As, 706), ciim fstac (Cm, 6l»), om which v. Klotz, Grmulz.^ p. I39f«. 
 and Skutsch, Btrl. I%il IFo^k^ 1894* p. 139 1; cf. also examples in I.ucilius and 
 the Augustan poets, as Hor. 5. 1, 9, 38 sf m£-am£s; 2, 2, 28 num adest, etc. 
 (L. M tiller, ^. Jf .« 371 ff.). Other examples of the double iamb, in Klotz, /./, as 
 /Vf. 447 (luando amor iubet, «ft probably to be referred to some form of the 
 traditional order (subject + «fb).— The view which I have here defended against 
 KU»tz is, as I have since notedf allQ the view of Luchs himself, who writes of just 
 such cases {Studtm. Stud. 1, 18) : 111 h» enim versibus pes paenultimus non 
 continetur iambo, sed anapaesto, quod cinn hiatu semper pronuntiandum est : se 
 am//, qui ha bet. 
 
 Vol. XXXV.] Studies in Latin Accent and Metric. 41 
 
 adverbs immediately precede possessive pronouns and oblique 
 cases of personal pronouns (Kampf, />;-,;;/, vz/.A/wz/.r/., pp. 27 
 30 ff.,35), give rise to the accents ille-mi/ii, unde-tibi, etc.; r./ 
 Ad. 139 iste tuos; Cas.^^i unde* meac ; Ep. 179 ilki mihi(cl 
 Skutsch, Eorseh. 136), and cf. Men 451 ; Ad 457 (fiVst foot)- 
 cf. ^7. r33 ille2meus(cf. Skutsch, /./., 118); cf. also Gr/. 461* 
 ipse^ sibi.i In association with the verb, ille'^ rcprehendit, 
 7/7. 624, happens not to fall within the critical feet, but 
 Commodianus, who observes the accent in the close of his 
 hexameters, writes Tnst. I, 35, 15: unde-^ licet I'lle (Hanssen 
 Dissert. Phil. Argent. V, 24),^ cf. the adverb undelibet. The 
 developed oxytonesis is seen in Tnt. 30Q ille« nieretriculis ; 
 note further that no example oi unde or ///./ rcdis (dissyllabic 
 verb) occurs in the dramatists, but /;//./.' red cam (trisyllabic 
 verb) occurs rei)eatedly (Tri. 937; /:-;,. i, ; j/,, 865; Hec. 
 177)' Examples of oxytonesis in the first foot are very 
 frequent, e.g. Cas. 432 il rl!6 trepidabat, Poe. 620 et ilM 
 chlamydatus (cf. Seyffert, Biirsians Jahresb., 1894, p. 282). 
 St. 24 neque ille sibi mereat (cf. Skutsch. /./., 118, n. 2),' 
 where the proceleusmatici, as usual, follow closely the gram- 
 matical accent 3; very doubtful, however, is .SV. 175 quia ind6 
 iam a pausillo (cd, miu.: quia inde iam a); Ps. 503 illijd 
 erat; Lucil. XXIX, 43 M. turn illiid iiTi<^mvd- doubtful \sAs. 
 123 nam illiid ego (so ed mai. ' ego illud Mss.); Cas. 932 
 inde foras ; Poe. 1055 inde sum oriundus; —tute tibi Cap. 371, 
 ^^- 563, Car. 9, hence also Per. 573 tute*^ tibi ; — for still other 
 examples, cf. Ahlberg, Be corrept. PL, p. 50, n. i, and Luchs, /./. 
 Haecin d.x\d haecine : Hec. 771 haecine east; ib. 2S2 haecine 
 ego vftam (resulting from the traditional word-order, which 
 attaches other pronouns to the sentence-introducing pronouns, 
 
 ^ This accent fs not, however, wholly invariable, ,.g. Eu. Si', fstuc mHii, and in 
 the first foot :/>.;.. 355; ib.2; O. 561 unde tjbi; v. oth.r examples in Kampf,/./.. 29. 
 
 ^ *This maypossii.ly be the true explanation of the accents diinde, pkrinde, 
 txtnde prescribed l,y the grammarians (Scholl, /./.. p. 192), /... de'inde, etc.; so 
 perhaps St. 545 deiiiiHe senex, rather than deinde senex. 
 
 ^ Hence I cannot accept the view of Ahlberg, Procel. I, 34, no. 
 
 ii 
 
Robert S. Radford, 
 
 [1904 
 
 €f. KSlUff, ill, pp. 31. 3^); Cf. Phor, 1013 haeci^ne erat ea. 
 Hence it is apparently largely due to accident, or rather to 
 the natural position of the sentence-introducing pronouns in 
 the beginning of the verse, that examples of this kind occur 
 
 and Ad. 128 sicfne agis (cf. Wagis, hoc-agis, sic-agis); Haii. 
 
 203 bttodne erat; Em. 771 haiicine ego. Similarly it is ap- 
 parently accidental that fmncine ego occurs in PI. only in anap. 
 verse {Ru. iSS. 189)- We may perhaps, if we wish, assume 
 alwayt for Fl. the syncopated forms luucin, etc., though this 
 is going much beyoad the available evidence (cf. Schrader, /./., 
 p. 10 ff., and the numerous examples there cited), but in any 
 case there can be no serious objection, on the score of accentua- 
 tion, til e«mplfis Bke Mo. 26 hocinemodo (cf. hocmodo); 27 
 hocine-boni; TrtL 7i9hici%etu eras; Am. SHhoci'ne placet; 
 As. 08 hiclne preti, etc. (cf. also Spengel on Ad. 183). 
 
 Ecqnis^ nmmquis^ qmisquis^ dquis : Foe. 364 ecquiM ais; 
 cf. Cm. 913 and 914 nisi quidquiM erat; — always siquid 
 agis: Ep. 196; Mi. 215; Per. 659; St. 715; 7^7 \ Tri. 981 ; 
 — cf. P'k&r, 5S3 siquiM opis; Ait. 193 siquiM opust; As. 117 
 siquid opus; Ad, 87/ CCqiiiM ego. Often in other feet, 
 especially the first, as An, 653 quicquiM habes; Poe, 505 
 qpiccpiM agit; Cas. 4S6 ecquid amas; En. 475 numqufd 
 habes; — quicquiM erat, etc.: Pen 46; 47; ^'''- 5^ ; 1308, 
 cf. nullus erat, A'//. 1253 ;—lf<rif. 439 siquiM ego; Eu. 523 
 ecquis 4m; As. SSf and il«i. 41^ sicit ego; cf. ML 1206 
 quo moMo ego. In short, if we take a trisyllabic word (verb 
 or pronowii) like imni and a dissyllabic word like erit, we 
 shall always find in the one case tlie accent quidquid inerit 
 {Ru. 1 1 34), in the other the accent quidqufd erit (quidquid 
 erft). This accentuation long continued m use, e.g. Sen. 
 Ocd. 263 quidquid ego; Rufinus 341 G. quidquid agis (quoted 
 by Ribb., Com. Frgm? oXx)} In % former discussion ( Tr&m. 
 
 1 Ribb • accents quite needlessly quidquid pla*cct, Caecil./r. 148; recomposi- 
 tion is perhaps first attested fur the Christian poetry, ^^. Et qufdquid micat 
 sideris (Konigsfeld, Lat. Hymn. I, 32). 
 
 Vol. XXXV.] Studies in Latin Aecent and Metric. 
 
 43 
 
 Am. Phil. Assoc. XXXIV, 84) I assumed with Ritschl and other 
 critics the divisibility of compounds like ecquid into their 
 constituent parts, i.e. ec quid ego, but this is a desperate 
 remedy, as Maurenbrecher, Hiatus, p. 31, n. 3, and Ahlberg, 
 Proccl. I, 81, have already observed, and is improbable for 
 the period of PI. or Sen. At the most, the pronunciation 
 might be derived from an earlier formative period, when the 
 two monosyllables w^ere still independent, i.e. ec-quid-ego, 
 hence ecqui'dego ; but even this assumption is unnecessary, 
 and no reason exists for separating the treatment of ecquid 
 and quidquid from that of other pronouns. 
 
 Nostra^ omnis Q\quisquc : St. 741 nostra* placet; perhaps 
 Tri. 329 meumst, omne'' meum autem tuomst, — so Scholl in 
 ed. mai., but the ed. min. retains the reading of A, etc., and 
 scans meumst, omne meli'^mst autcm tuom ; Ep. 214 suis 
 quaeque® | amatoribus (r^. mai. quacque ibi); so the phrase 
 suo quique loco (also suo quicque loco, St. 62 ; cf. Prehn, 
 Pronom. Indcf., p. 11) is always accented sucS quique loco, 
 viz. Mo. 254 (troch.2), St. 62 (tr.^), Poe. 11 78 (anap.^), Titin. 
 com.fr. 130 (tr.'*, acc. to Miiller, PI. Pr., p. 61, but Ribb.^ 
 scans as anap., quique loco); cf. St. 693 suom quemque^ 
 decet, and also Au. 732 quoi tanta^ mala.^ 
 
 C. Apparent Division of Trochaic Dactyl. — Ex- 
 clusive of the first foot, we find thirty-four examples of a 
 divided dactyl, such as quisquis homo, with illic, ille, unde, 
 inde {io\ cf. hasce {2), haccin, sicin, istficin, tute (4), qualis^ 
 omuis, quisquis, ecquis, numquis, siquis, quisque (18), and 
 apparently many more with huius, eius, quoius? 
 
 ^ In addition to the pronouns, the proclisis and oxytonesis of esse (cf. Skutsch, 
 Forsch. 136, n. I) and hercle have been established in certain word-orders; thus 
 Ba. S3 voles esse* tibi; .SV. 716 vi'des esse'' tibi, and, in Ten, cf. Hec. 398 esse^: 
 
 scio; Au. 40 hercle tibi; hence the frequent shortening seen in these words 
 
 4 
 after a Brevis Brevians. Also in the case of pred. adj. -f copulative verb pro- 
 clisis has perhaps been known, viz. Poe. 1194 pulcra videre; Ru. 218 serva forem, 
 
 6 6 
 
 or servS forem (so B). Some of the other cases of oxytone trochees quoted by 
 
 Klotz, Gritnih., p. 239 ff., may also be genuine. 
 
 2 The admissibility of the broken dactyl in general has been discussed espe- 
 cially by Muller, Pros, Xnchtr., p. 12 ff.; Leo, Forsch., p. 236 ff.; Ahlberg, Procel. 
 I, 14 ff.; Maurenbrecher, Hiatus, p. 26 ff. 
 
f ■' 
 
 Robert 5. Radford. 
 
 [1904 
 
 Examples of ilk quidim (twice within the verse: Ep, 
 673; Mo. 1081) and Ulk homS (once within the verse: 
 Jffi. 593; sixteen times in first foot) are cited by Luchs, 
 
 Cmtm. Fms. II, 4, and Hermes, VI, 278, respectively; cf. 
 also Skutscli, Forsc/i. 115, Since the sentence-introducing 
 pronouns are placed especially in the verse-beginning, it 
 Is probably due oely to accident that we find examples of 
 tic siinilar combinations isU qmidim (Men 945 X ^ semx 
 {Men 44<5; Si 559). ^/'^'^^ ^^^^^ i^^^- JO/o). i'ccjm's homo 
 {Mo. 354); cf. sMm komo (St. 58; 442; for the combina- 
 tion ¥. Asnius, De apfositiomis collocat., p. 18) only in the 
 ifst foot. Again, while I accept with Skutsch the weaken- 
 ing of the final syllable of f/Zr, mde (also of hocin, quid- 
 quid, etc.), i" sequences where it is regularly unaccented, 
 I do not admit any need of this explanation (Skutsch, 
 £4 no f., Si) in cases like Bm. 281 file mihi (cf. ilia 
 milii, ilM meus, etc., above); Am. 660 i^lle revortitur; 
 Hmdi. 107 immo \V& fui't senex inportunus (cf. also Ahl- 
 
 berg, Pm^/1 I, no); Cas. 903 mde volo; St. 6j i%de voca- 
 tote; Foe. 902 i%ide surriiptus; Per. 150 unde surrupta, 
 not to mention the numerous cases occurring in the first 
 loot, as iSnde lub6t {Ep. 144), ille iubebit (Mi. 1192), or 
 Jlle miserrumum (il//. 713). Even hasce tabellas (Ba. 787), 
 
 basce coronas (i4ii. 38s) admit of some defence ; for, although 
 
 Pi never allows the particle <e in these pronouns to have the 
 IttI ■iraltte of a mora and so to form the whole thesis (Schmidt, 
 Depronom. demons tn f^rm. Pimitinis, p. 16 ff.), he may well 
 have allowed this particle to form part of the swift thesis of 
 
 the *€fclic' dactyl; cf. Tri, 186 hascc mihi (ace. to A and 
 
 f^. fHXil.). 
 
 Haecim. siein, isiueim, tmte: And. 236 hocinest officium ; 
 
 Per. 42 si'cine h«ic ti; Em. 830 istucine interminata; Tti, 
 
 386 tu^te concilies (unnecessary is tut' with Skutsch, 
 Forsch. 151); numerous examples for the first foot, as 
 Phor. 10 1 2 lm#cine erint; En. 804 sPclne agis (cf. hae- 
 cfne erat, etc., above, p. 42); Poe. 512 siWne oportet, 
 
 Vol. XXXV.] Studies in Latin Accent and Metric, 
 
 45 
 
 are quoted by M tiller, Pros., p. 441 ff., and Schrader, /./., 
 p. 17 f.^ 
 
 Qualis, oninis, qnisquis, ecqiiis, siquis, quisquc : Hec. 766 
 qua^is sim amicus (the insertion, in accordance with Wack- 
 ernagel's law, of an ' enclitic ' in the second position does 
 not affect the connection); Cap. 536 omnis in incerto; Am. 
 
 309 quiVjuis homo hue; Ps. 713 qui-dquid opi'ist ; Men. 772 
 sed quicquid id est (bacch.; needlessly corrected in ed. min.)\ 
 
 Mi. 311 qui^cquid est; Tri. 218 qui'dquid audi'tum ; Ps. 740 
 
 2 
 
 e^'cquid habet ; — ecquis hoc aperit ostium as verse-close : Ba. 
 
 •A 4 
 
 582, Am. 1020, Cap. 830; — Ba. 581 ecquis [his] in aedibust; 
 
 Trn. 839 si^quis eiim ; St. 182 si'quis me essum ; And. 258 
 
 4 
 
 si^'quis nunc me ; An. 340 si'quid uti' ; Ep. 729 si^quid impru- 
 
 4 
 
 dens; As. 326 qui'Mque derogito. Often in the first foot, as 
 Tri. 655 and Rn. 1100 omnia ego istaec ; Ru. 1359 omnia 
 ut; /At. 287 omne quod ; Rit. 1121 quiMquid ibist ; Ep. 677 
 and ffafi. 961 qui'dquid ego; Ep. 293 niimquid ego ibi ; Poe. 
 506 sicut ego hos ; St. 576 nequid adveniens ; — ccquid agis 
 Ah. 636, Ep. 688, Ci. 643 ; — (Ccquid amas Tni. 542, Poe. 327; 
 — Ci. 67 si'quid est quod ; Ep. 647 si'quid erit ; etc.^ (cf. 
 Ahlberg, /./., p. 80). 
 
 ^ T have purposely not quoted examples like A/i. 61 hicine Achflles; P/ior. 992 
 
 hicine ut ti'ln. The long quantity of the first syllabic of hicine (pronoun) is very 
 
 4 
 doul»tful, and is not proved, as Miiller, //., thinks, by the iamb. sep. Ad. 709 hie 
 
 nrSn amandus? ILine non-gestandus, j///rt' no diacrens ne:essarily falls after a* 
 
 t 4 
 
 mmiosyflahli^ as I shall show more fully elsewhere. On the i ther hand, the 
 
 quantity in hdeine^ kmeinet s'cine is often attested. 
 
 ■-' Among other cases of the broken dactyl whi :h appear to be justified by the 
 
 regular word-order, I wnuld specify the following: i) the phrases composed of a 
 
 pred. alj. -f incedo, like Mi. 897 ornatus incedit (also As. 405; Ba. 1069; A/er. 
 
 600; Poe.$Tj) ; cf. Mcr. 887 ami'cus advenio. This combination has the same value 
 
 as adv. or adj. + copula, as Poe. 922 intus-ero odio; cf. also Cap. 321 unicus-sdm. 
 
 2) Common alliterative phrases, or coml)inations of related words, viz. Ps. 704 
 
 trma triplicia; Mer. 385 ami'cus amfcis, similarly J//. 660; Pomponius /;-. 145; 
 
 cf, also Pa. 401 commodiis I'ncommodds. Cf. the not very different view which 
 
 Leo takes of these combinations, Forsch,^ p. 244, and observe that in etymological 
 
 phrases like amicus amicis tJie order is fixed, i.e. * nominativus casus obliquos 
 
 praecedit ' (Kellerhoff, Student. Stud. II, 58 f.). 3) Perhaps adj. or gen. + noun, 
 
 viz. Men. 268 magnus amator (P; A mag.) ; Accius fr. 501 horrida honestitudo; 
 
46 
 
 Robert S. Radford, 
 
 [1904 
 
 HuiMSp eimSf ^Mifius: These forms, instead of being always 
 taken m^ mmkmjlUhm^. ov^t probably often, to receive their 
 regular scansion as dissyllables in the very numerous cases 
 like Cap. 887 quo'^ius erat tunc ; Si, 545 quo'^ius erat tibicina ; 
 Rii, 52 ^ius erant; 1204 ^ius amatiost; see the examples col- 
 
 iected by Ahlberg (De correpL PL, pp. 84-90), whose conclu- 
 sions probably require some modification at this point.^ 
 
 D. Admission or Dactylic Pronominal Forms. — Ex- 
 clttsive of tic irat foot, mt, ind about twenty-nine examples 
 of a foot illcd by tie dactylic pronominal forms haccine, 
 isfaecme, iiiaecim (7), illius (20), omnia, omnibus (2). 
 
 Haecine, istaecine^ iliaccine: Foe. 11 66 haccine meae sunt; 
 
 Ms*%l istdcine picto; 847 ista^cine caiisa; Rti. no isticine 
 v&; Tru, $37 ho^cine mfhi; 606 istiicine mihi; //a/s. 751 
 illancine mulierera. The examples quoted above (p. 44) of 
 
 Mmm, etc., with elision, i.e. IwScinesI officium, show clearly 
 tiat tie pfobleni presented by Mctm^ without elision is not 
 solved by the assumption which is sometimes made (Klotz, 
 Grftfids,, p. 308; Ahlberg, Procei. I, 20, 112), that the final 
 syllabic of this form was always syncopated in pronunciation ; 
 besides, tils assimiftion of constant syncope after a short 
 vowel appears to be unwarranted (Schrader, /./., ro, 14). For 
 the freciuent occurrence of haccine and similar forms in the 
 irst foot, see Scirader, /./., la 
 
 JiMm^i Matt* JlSfllius inimum; Ad. 261 illius 6pera; 722 
 fllius adulcscentis ; 572 iUIius hominis, and more generally 
 
 Inn. Ir. /r. 40 virginalis modestia; Amf. 857 tri'stis severitas; As. 509 maHris 
 imperium; Si. 43a fritris ancillulam. In most of the common phrases so far 
 €|iote4 the elision of final s before an initial vowel, which Leo advocates {Forsck, 
 229 ft.), may also afford an explanation* 4) Iii| + auxiliary verb. viz. A. ltS2 
 frelicebit (cf. ilicet); A'm. 112 esse decet; Ep. 573 ferre iubcs. Cf. esse-vuli 
 used as equivalent of a trisyllabic word in verse-close not only often in PI, but 
 abo in Syr. Sent. A 51, and dare-vis placed in the pentameter-close by so perfect 
 a master of technique as Martial (7, 75, 2); cf. also dare-nhcit, Syr. Sent. 46, the 
 
 i.tily divided anapest, accordinf to M«yer, ill, jk a|| irlilcli Publilius admits in 
 the third foot. 
 
 1 The view that huius^ etus, etc., may also be pyrrhics, advocated, so far as I 
 know, only by Exon. Hermatfuna^ XI, 208 ff., does not seem probable. 
 
 Vol. XXXV.] Studies in Latin Accent and Metric. 
 
 47 
 
 Ep. 717 i^llius invenisse; 447 illius illae ; for other examples. 
 
 3 ^ 
 
 see Leo, /./., 290 ff., Ritschl, Opusc. II, 67S ff., and Ahlberg, 
 
 De corrcpt. PL, p. 91 ff. The proclisis of ilk is, in fact, so 
 far developed that there is no diaeresis in the iamb, sep., Mi. 
 1 23 1 qiiiroquam illiun-multae sibi expetessunt, and this is the 
 real explanation of this much-cliscussed verse. The scansion 
 illius which I have here assumed has been successfully vin- 
 dicated for PL, in my judgement, by Ritschl and Leo,i although 
 the scansion ill'is must also be admitted for a few examples 
 like Phor. 648 mi'ttam illius ineptias. - Omnia, omnibus, om- 
 nium: Tru. 447 omnia qui; Am. 55 omnibus isdem. Often 
 in the first foot, viz. Mi. 1 148 ; Poe. 834 ; 905 ; St. 1 14 ; 336 ; 
 526; 684; 7>'i. 933; Hcc. 380; Ad. 971. Several of these 
 examples are of the type omnia genera {Poe. 834) ; in the 
 case of dissyllabic words, however, accents like omnia mea 
 {Han. 575 ; Phor. 248), omnia bona {Han. 942) possibly 
 correspond sometimes to the actual pronunciation. 
 
 While proclitic tendencies exist in all the pronouns, they 
 have not necessarily been developed in all to the same extent. 
 The proclisis is most nearly complete in ilU\ iste, ipse, and an 
 accent illes/nex, istuc-dgo is probably nearly as exceptional as 
 altrims^cus (Ps. 357) or amabilis (St. 737); thus illic homo 
 occurs twice {Ep. 45 = troch.«; 671 = troch.^) against twenty- 
 four cases of illic homo and illic homo. PL and Ter. have 
 only istuC'agd,—t\Y\CQ within the verse {Tri. 819; Eu. 349), 
 four times in verse-close {As. 358; Ba. 708; Man. 346; 558); 
 scid is as frequent in the dramatists as scio, yet we find only 
 once tantundem scid {Per. 517) against eight cases of ilium- 
 scio, illuc-scio, etc. {Am. 922 ; Jl/eu. 246; Per. 161 ; Poe. 1028 ; 
 As. 869; Mi. 236; IVu. 811 ; St. 474). Many combinations, 
 however, vary, cf. Eu. 536 I'stuc ita, and while PL regularly 
 accents tut/ . tibi (four times; on the order, cf. Seyffert, 
 Phtlol. XXV, 459 f.), tute ti^bi seems the correct scansion in 
 " ^en. II I b. 
 
 We have seen in this investigation that the proclisis and 
 
 1 Against Luchs, Studem. Stud. I, 319 ff. 
 
4'0 
 
 Robert S. Radford. 
 
 [1904 
 
 oxytonesis of the pronouns as a class, which the ancient 
 grammarians strongly attest, is confirmed in every possible 
 
 way % ■ the' eiiience of dramatic ¥ers€/ and we have found 
 at tie same time, as I hope, an explanation for several 
 phenomena of the verse which have hitherto remained 
 imexplained. Since the study of the pronouns is the main 
 object of this paper, my treatment of the topics which 
 remain will be brief. 
 
 IV. Accent of Adjectives. 
 
 Although the L-Eur. word-order, adjective or genitive 4- 
 
 substaEtive, is very imperfectly preserved in Latin, we might, 
 perhaps, expect to find that it had exerted some influence 
 upon the Latin accent The critical feet offer, however, very 
 few direct evidences of recession at this point, and we find 
 Iianify more than half-a-dozen certain examples of a recessive 
 accent like recti via in the critical feet Yet the rare occur- 
 rence of such an accent may be exi)lained as due in part to 
 the structure of the verse. Thus, to take illustrations from 
 th$' .siructii're of the scnarins, tie combination of spondee 
 and iambus which is found in recta via, needs, as a rule, to 
 be utilized for the verse-close, and this fact of itself would 
 largely prevent its occurrence in the critical feet. Again, 
 the combination cannot be placed without elision in the 
 second foot, since it would then produce an inadmissible con- 
 flict in the third, i.r. rectd via, nor can it be placed without 
 
 elisioB in the fourth, since it would then introduce an iambus 
 into the ifth foot, i^e, recid'vid; it follows that it can, as a 
 
 rule, be placed only in the third foot, and we find in fact 
 
 1 The procHsk of the clcmoiistrattve and relative pronouns in Latin, Umbrian, 
 and Oscan is also indicated at times by the orthography of the Inscrr., viz. mmrim, 
 eare, eaires^ Ammmertm (Corssen, II, 879), eajiveka (lat. eas iuvencas), erer- 
 nomneper (pro dill nomine), pamncemto (quae incensa), cf. Corssen, II, 919 ; 
 V. Planta, I, 599. For ih* conjunctions also proclisis is in. Heated in Umbrian hy 
 writing ape (= Lat. mbiy in m*aninj^), and /»««•, I.at. -cum/e, at times uith the 
 fallowing word, as apepesomira, cf. Corssen, ibid. Especially oflen is the proclisis 
 of the pronouns indicated in connection with modus, e.g. huiusmodi, huiuscemodiy 
 eiusmodi, etc.; for fuodammodo, omnimodo, {nullomodo,) v. the lexicons, and fof 
 ullomodo, V. Cod. Bern, ^i, Keil, SuppL 180, 9. 
 
 Vol. xxxv.J Studies in Latin Accent and Metric. 49 
 
 several examples of its use in this place, viz. Hcc 177 prinios 
 dies; Han. 61 pro deum atque hominiim fideni ; cf. Jri. 425 
 mille drachumarum (numeral). In addition: Syr. Sent. 388 
 non est turpis cicatrix; Am. 481 decumo post mense • Hec. 
 198 pro doum atque hominum fidem(oct.j; /:>. 249 hominu^m 
 me vis; cf. Ba.g68 uno mendacio (oct.); cf. P/u?r. 867 sus- 
 penso2 gradu. Cases of adj. + quadrisyllabic noun in the 
 verse-close, as //ec. 462 una^ sententia ; Am. 840 sedatu^m 
 cupidinem; 841 cognatu^m concordiam ; As. 298, — are in- 
 conclusive. 
 
 That the non-occurrence of the type is partly due to the 
 verse-structure seems confirmed by the notable fact that the 
 accent trigintd-minas, etc. (numeral -f noun), which is at- 
 tested by nearly all the Romance languages, also does not 
 occur a single time in the definitely critical feet,i and we 
 find only one example in these feet of the quantitati\-e type 
 fratrJm-maim (And. 540 gnat'm tuam et; cf. Rn. 1341), al- 
 though the occasional existence of this accent seems indicated 
 by verse-closes Vikcpa/nm-hu-'im ( JAv/. 750; Mcr. 972 ; Titin. 
 //'. 65 ; Asin. 64). Very serious difficulties, however, still 
 remain ; for the combination of iambic or cretic adjective -h 
 noun, i.e. mald-manu, dcxtcrd-manu, is always avoided in the 
 verse-close, and we find instead dextera retinens manu, Cap. 
 4i2, and the fllce; the only exception is the phrase l?ond(n) 
 fidd Tru, 586; Mo. 670 (v. Luchs, Studem. Stud. I, 21, and 
 Kohler, /./., 31). We must conclude then upon the whole 
 that, in the case of adjective combinations, there is no suflfi- 
 cient evidence for the existence of an accent maid manu, 
 except in the case of a few special phrases. 
 
 V. Relation of Word and Verse-accent. 
 
 The results reached in the preceding study afford a fresh 
 proof of the frequent agreement which exists between word 
 and verse-accent in the verse of Plautus. It is important, 
 
 ' All the examples cited by Skutsch, Forsch. 163, involve the second arsis of 
 the septinarius. 
 
5© 
 
 Robert S. Radford. 
 
 [1904 
 
 Vol. XXXV.] Studies in Latin Accent and Metric. 
 
 51 
 
 however, not to exaggerate the extent of this agreement and 
 not to misinterpret its meaning. Thus at the present day 
 we sometinies hear the belief confidently expressed that a 
 complete or almost complete agreement of word and verse^ 
 accent will eventually be proved for Latin dramatic verse, 
 and further, that all those kinds of verse which, like the 
 dactylic hexameter, show serious disagreement, are artificial 
 verse-forms m Latin. Such extreme views have little or no 
 basis in fact, lor both the agreement between word and verse- 
 accent in Plautus is far from being complete on any scientific 
 hypothesis, and the quantitative poetry of the Romans is a 
 thoroughly genuine and national product. 
 
 Importance of Quantity. Problem of Ictus. — It is 
 difficult for the modern student — who has rarely, if ever, 
 heard an exact quantitative pronunciation — to understand 
 the aniquc importance which attaches to quantity in the 
 pronunciation of the classical languages; and this general 
 difficulty is vastly increased, so far as concerns the pronunci- 
 ation of Latin, by the numerous special problems which Latin 
 here presents. Hence many scholars have sought an escape 
 from this diiictilty by supposing that a strict observance of 
 quantity formed no part of the original Latin language, but 
 was adopted later by the educated Romans in consequence 
 of Greek influence. The actual predominance of quantity, 
 however, in the genuine Roman pronunciation — or, at loast, 
 in some widely prevalent form of this pronunciation — is 
 indisputable. Not only has Latin inherited its system of 
 ilttaiititf from I.-Eur, in an essentially unchanged form (Ber- 
 gaigne et Henry, Manuel Vidique, p. 38), but under certain 
 conditions Latin quantity is invariably accompanied by an 
 important secondary product, viz. intensity ; for any regular 
 alternation of long and short syllables in Latin prose or verse, 
 provided: aciftf ill fnwciation of the quantities be employed, 
 produces nminralfy * a rhythmic beat or ictus {impressia, Cic. 
 
 I The same view is taken by Venclryes, Vintemite initiate^ Paris, 1902, p. 65, 
 
 ma<l. accoriiiiif to V., by Mcllkt, who writes "par suite de la pronunciation 
 fi#^rr//# del mots**; cf. also Uppgrcm, il/r/riWir Komposiiion J. 7V/-f«s, Lund, 
 
 1901, 107 C 
 
 de Or. Ill, 48, lSS\ pejriissicy, ib. Ill, 186; quasi pulsus, 
 Quintil. IX, 4, 136), which approaches the value of a stress- 
 accent,^ or, to state the fact more briefly, variations in quantity 
 
 1 In connection w ith Professor Bennett's attempt to show that ' ictus ' was not 
 identihed with eUvatio voeis 'before the tifth century' (././/•. XIX, 368 ft'.; cf. 
 also Veiidryej*, /./., 66), I cannot refrain from callin<j attention to a iiuestion of 
 chronology involved in this claim. To obtain such a result, it is not onlv neces- 
 sary to explain away many early testimonies, hut also to assijjn a very late date 
 to Terentianus Maurus, who says very clearly with respect to arsis and thesis, v. 
 '345= P^'"'^ "^"^ attollit sonorem, parte reli(iua deprimit; v. 2249: (iicccsse est) 
 scandendo et illic {sc. in secundo IccoJ puncre adsuetani nioram (T. is discussing 
 the principal ictuses of the trimeter). Whde IVs date is not certaiidy known, 
 accordinjr to the judgment of all the best recent authorities he is ])r(.i)al)ly to be 
 assigned to the close of the j^^ow^/ century (SchuU/, //,; w,,, XXII, 275 f.; W crth, 
 JahrlK Sp. XXIII, 2^5 ff., and preface to dissertalicn, Leii)/ig, 1S96; Teuffel- 
 Schvvabe, i\'6m. Lit. 11,945; Schan^, horn. Lit. Ill, 26); in a.l.lition, his state- 
 ments are usually drawn from good early sources. In ( irct k als.., references to the 
 existence of a metrical ictus are not so rare as Professor Ooodell appears to think 
 in his recent book {Greek Metric^ 156 ff.). Thus, besides Christ's examples of 
 Kpouttv {iMttr.f p. 50) and the freqiieet marking of the crriyfiai in musical 
 schemes (Gleditsch\ p. 322), Plutarch, Dew. 20, describes I'hilip after the l)attle 
 of 1 haeronea reciting the tetrameter verse Arj/jxxrO^vijs \en6(rd^vovs, kt\., dividing 
 it into feet and beating time to it (rrpds ir65a biaipwv nai vTroKpovuu); the author, 
 w€pi IVoi's, c. 41, 2, states that the rhythmical clausulae of the orators often have 
 the effect of dance-music upon their auditors, who sometimes cannot refrain from 
 staiiii»ing their feet in time with the s[>eaker (uvoKpoveiv rots \4yov(ri. kuI . . . 
 TpoairodiSouai ttjv ^dffiv). Similarly an ictus in the oratorical cola ami clausulae 
 is constantly affirmed by the ancients (Cic. </t' Or. Ill, 47, 182; Or. 18,^9; cf. 
 (Juintil. IX, 4, 31), and is assumed by all modern writers in this lield. "To'the 
 ancient testimonies upon the verte-accent cited by Ilendrickson, .]././'. XX, 198, 
 a<ld the very late Vergilius Grammat. Fac, K., Suppl., 190, 12, who mentions an 
 .iccent r^^« 'secundum rationem metrorum.' At times the ancients do not ajv 
 pear to distinguish shari)ly between word and verse-accent, cf. Auson. Ep. XXII, 
 47; Cassiodor. Var. 9, 21, 3, and the striking use of Tiivo% in Cireek to denote ' 
 iKith word-accent and rhythmical cadence. Hence I am disposed to justify, 
 again$t the objections cf Crusius, /.///. ( rntralhL 1891, 7, 213, the use which Klotz, 
 iiritmiiz. 269, 348, makes of Dionysius' avWaSr} S^vtovos {de Com p. Verb., c. 1 1) in 
 iiuestions relating to the ictu^, and even suggest for comparison in part Dionysius' 
 other statement {ite Adniir. Vi Dem. c. 48), that Demosthenes has produced won- 
 •lerful 'melody' through the arrangement of 'sharps' and 'graves' in his cola. 
 Since the ictus, in all poetical recitation, largely takes the place of the tone, such 
 a lack of distinction, though inexact, is not wholly unnatural. Similarly, in his 
 •hscussion of volilcres ( I, 5, 28), Quintilian perhaps employs syllaba acuta 
 'inexactly' of the ictus ; such/St lea.st is the view of some metricians (^^/llavet, 
 Mttr.^ § 49' ; Scholl, De acL, p. 26\ although the passage seems t<j m/to admit 
 
 jjtf. Humphreys, Trans. Am. Phil. Assoc.Alllj 109). 
 
 of another interpretation 
 
 f^^^, if^ /V4> A»I Aa 6itL*il 
 
 fU 
 
 u — ^ 7? 
 
 
 
52 
 
 Robert S. Radford. 
 
 [1904 
 
 Vol. XXXV.] Studies in Latin Accent aftd Metric. 
 
 53 
 
 litii 
 
 I 
 
 
 naturally produce variations in intensity. Hence Latin pos- 
 sesses, in all carefully enunciated quantitative sequences; a 
 
 aaliiral scries of stress accents, and this series assumes many 
 
 Cf. alio Ciissiwior. Var. 2, 40: natoralis rhythmus animatae voci cognoscitur 
 itlribtttw : c|ttt tunc mths pukhre custodit, $i apte taceat, conuruenter loquatur 
 el per m€€eMtm viam musicB ixclibus composita voce gradiatur ; where taceat and 
 hqmmimr are doubtless to be explaiiied from Aiist. Q. de Aim., p. 31, 17 M. 
 Iiwiir Rui #rf»iPt if^oif Kai ifp€fuap. III a Jdition to the pa.ssages usually (ju jtetJ, 
 a clear teslimony upon the rhythmical accent in prose and verse is founl in Sen., 
 i/f Bret'. Fii. 12: <niid illi qui in conponemttis, a«dien<iis, diccndis canlicis uperati 
 sunt : dam vocem caias rectain curwin naiara et optimum et simplicissimum 
 fecit infiexu piodulationk inertissiinac torquent ? quorum digiti alitjuo.l intra 
 se carmen mclicnte* semper sonant, qaofum, cum ad res scrias adhibiti sunt, 
 exaudita tacita moJulatio. That the interpretaii.m of Diotiysius given al>ove is 
 correct, and that he refers lo the * fcttts Vin the Uemoslhenic melodies, ai»i>ears 
 tO' be coiififmcd by the very similar passages in which Lucian speaks of • the fieai 
 and Mtemimil mJody ollhe Demosthenic rhythms' *the /w/of the Demosthenic 
 periods,' viJ!. />#«. ^mmm.. 32 d S* *Atti«oI {nfjTopet TratSta napaiidWeiv ry tovtov 
 (m. Anfme0ivovt) icpdT«» Kttl t6v«ji xai X^^£wi/ cjJpuM^ ; i^'- '5 ^^^^^9^ 5^ « KpoTOs 
 Twr AfifimBtPm&p Wymif <lir%«i' itpalvtro tov WKrepiKov Xvxvov- Cf. AnikoL 
 Fiammi. 226, 5 4|*#i «^ <toI puenioio Kara Kporov (vdeov Tx^js py\<Tai(Td(a (descrip- 
 tion of a 'Song accompanied by the pipe); acj'ording, to l»assow's Lex. Kporos u 
 here**der Klang eincr Rede, eiiies Ge*ang;s"; in oratory, rb ^irCKporov tCcv 
 XA^wf is very aptly descril>ed by Herodes Alticua in rhiU>str. 539 Boiss. Some- 
 what late is the aw of latipoKporos, • iaml>ic-beaten,' in the meanhtg • iamliio,* 
 •in iambic rhythm,' vii. Niceph, in Wak, RhtM, i, 443 {Idfi^oKpdrois Uyo-.^ etprtKtv, 
 •in iaml>ic verses*); Planad. ik 5, 450 {& pvfffids ianu 6 Toidt fjxoi roO Uyov, 
 iaft4i6KpoTm t^bp # iXtyetm). So far as concerns the relation of word and verse- 
 accent, this relation in latin is undoubtedly largely one of similarity in kind ; 
 cf. Schiill, Z>ir <«¥., «3. This fact is not only clear from the statement of Anni- 
 aniis ap. Gcllius VI, 7, 4, and other similar passages named above, but is made 
 evea more obvioas by the frequent employment of precisely the same terms 
 itt reference to both, viz. mmMarit mmiuiaiio, modermrt^ flectere^ etc. Thus in 
 the well-known passage Cic, Or, i8» 58 mmimhri is em[>h»yed of the wor d-a c c ent : 
 ipsa natura, qaasi meduiareimr homlnum orationem, in omni verbo jwsuit acutam 
 vocem ; ct Mart. Cap. 111* p. 65, 19 Eyss. On the other hand Quintilian (I, 10, 
 
 , jt^ 22) employs m&duiM^ of the «beat' which accompanies rhythm: jtvBp^t . . . 
 
 I moiiulaihne constat, and shortly afterwards (25) explains the meaning of mtniiP- 
 ^^^^iatio more deJinitely l)y the terms imienim zmis, remiissm^ ^xm (c£ Hendrickson, 
 .-/,/./'. XX, 206). lioth uses are found together in Victorinus VI, 206, 12 K.: 
 met rum est ratio cam mpdmlaiime, rhythmus sine ratione metrica modit/aiio, 
 plerumcjue tamen invenies rationem metricam in rhythmo, non ariiticii oliserva- 
 tione servata, icd mm et ipsa m^duiatimi ducente. Similarly Gellius (XIII, 6, i) 
 uses moderamenta of the word-accents, bat Caesius Bassus (ap. Rufin. VI, 555 K.) 
 moderare of the verse-accent : nisi percussione ita moderaveris. Again, fiecten^ 
 ftxuh ttc., aw constantly employed of, the ictus (^e.g. Quintil. 1, 8, i quo flew . . . 
 
 I c 
 
 >' 
 
 /different forms corresponding with the various forms of metre 
 employed. So much is perfectly clear, but since the Latin 
 word-accent appears to have possessed from an early time 
 an element of intensity (probably varying greatly in strength 
 according to the qtiantitative relations, i.e. moderately strong 
 inccil{t)d.'is, weak in urbani ; cf. L. Miiller, R.Ml^ 468), it is 
 evident that, unless we are able to explain away the internal 
 evidence which the language affords in favor of an accent of 
 intensity, we must admit the eidstence in Latin of a second 
 series of quasi-stress accents. 
 
 SoNG-iJKE Recitation of Ancient Poetry. Effect upon 
 Quantity and Word-accent. — The difficult problem pre- 
 sented by the conflict wi|ich often arises in Latin poetry 
 between these two series of accents has given rise to many 
 discussions. Instead of attempting in this place a general 
 treatment of the subject in the usual way, I wish to limit my 
 inquiry to several definite and special questions, viz. What 
 was the form of delivery adopted by the ancients in the reci- 
 tation of their poetry } What was the probable effect of this 
 delivery both upon the quantities and upon the word-accents.** 
 Although the Greek word-accent was, down to the beginning 
 of our era, primarily a musical accent, and although its rela- 
 tion to the rhythmic accent has often been discussed in a 
 general way (r.^<^. WestphaF, III, 31 f.; Christ^, 55; Gleditsch^, 
 81 f.), I shall, for obvious reasons, include the recitation of 
 Greek poetry in the present discussion, and make free use of 
 the statements of Greek writers. The whole subject may be 
 treated under two principal heads : 
 
 I. Character of Poetical Recitation. — In that part of their 
 poetry which was sung, the Greeks, according to the well- 
 
 dicendum), but Diomedes (T, 456, 18 K.) observes that whenever the inflections 
 ijlcxus) t>f the voice come into play, the • accents ' come into play : tenor in 
 ftexibus vocis servatulus est'; nam quaedam acuto tenore . . . desiderant enun- 
 tiari. With reference to (Ireek usage, one must speak with extreme caution, and 
 there is no express identification of the oxytone syllable with ictus to be found 
 earlier than the mediaeval fraL^nnents of Pletho (fifteenth century); yet, as we 
 have seen, Lucian joins Kphro% and t^ws, and a certain degree of similarity is 
 indicated even for the classical period by the twofold use of t6tos already 
 mentioned. See further, AJ.P. XXV, 420 (421), n. 5, 
 

 Robert S, Radford. 
 
 [1904 
 
 Vol. XXXV. J Stitdiis in Latin Accent and Metric. 
 
 55 
 
 IcBowii stJUement of Dionysius Hal (^^^ Comp, Verb,, c. 11), 
 subordinated the usual proiittuciatbn to the melody, and 
 entirely disregarded the word-accent.^ This statement is 
 cumiiioely thought to apply exclusively to song, and to have 
 
 B0 tearing upon the poetry which was spoken or declamied; 
 but such a view can scarcely be correct, since all poetry was. 
 in a certain sense, sung among the ancients, and * declama- 
 tion' appears to have differed in degree, and not in kind, 
 fforo ' song/ This fact is often overlooked, and has scarcely 
 received from metricians the .attention which it deserves ; for 
 while the technical employment of song, recitative and decla- 
 mation in the delivery of ancient poetry has often been care- 
 fully discussed/' one will seek in vain in our standard metrical 
 works for any account of the real character of ancient 'decla- 
 mation'; valuable notices of the subject are to be found, 
 however, in several works of a more general character, viz. 
 Friedlander, Sitteugesch. l\V\ 337 U Teuffcl-Schwabe, Roman 
 lit. I, I 34* 3; Grasberger, Erziehung u. Unterncht im 
 
 1 Cf. the descriptioB wWcli Ileliodorus (3, 3) B'^«s ^^ ^"^^ rendering «f a 
 meWy : tm^m W n I^M«f«« ^m%^ rm% x^poiu icai oJJru, avt.(iaiyw. 6 krotos 
 rm ^fmro^ irpds rd liAot #^pi-<?Mi*T<»i ^^ '^r\.; cf. 3. 2. There is evidently no 
 fetenliow of the word-accent here. On the other hand .Schiill, /}e n-tT., p. 18, 
 wholly misapplies the very iaie scholiast on Dionys. Thrax , p. 830 IJk., m his 
 attempt to prove that the Creek accent w«i sometimes observed in song. The 
 Kholiast bch>Bgs. to a time when short syllables were lengthened by the word- 
 accent (e^. attXoF ifip\ and when, as he himself says, musical notes {Kpovaaara) 
 and wofd-accents were often ma le to agree. For the earliest examples of this 
 tendency, v. M«iifO^,. if&des' #/ Cnei Mmie^ p. 90. 
 
 ' E^. W. Christ, £Mg .fymiaMUfff im p-, u. r'Sm. Drama, Miinch. 1875; 
 M./nr\ 676; Zieliilski, CtkderMfti il 11//. X'^m., Lpz. 1885, 288-314. I" 
 general, Christ's views upon these questions seem to me much less correct than 
 those of Zielirtski ; according to the latter, the Greek tragic trimeter was com- 
 m.>nly rendered through rccitattve, the purely comic trimeter through the V AJ? 
 X^^is.' So far at conceflit this last statement, its correctness depends upon the 
 meaning given to ft\^ K which Im itself is an ambiguous term. If the al>sence 
 of musical accompaniment alowe Is meant, no exception cm be taken to the use 
 of this term ; but if it be meant that th« fecitation of the comic trimeter if •mere 
 speech' Of *mere declamativm* in the nmlern sense (cf. ifiXi? 4>tavii, Dion. H. 
 tt^ Comp. It), and that it is unsupported by the ir \(£<r/xa or w^rrXaatxivi vvbKpUftf, 
 such a conclusion appears to me to be quite false ; cf. Quintil. II, 10, 13 ; XI, 3, 
 183, and see esp. Klot/, Gnmds, 381. The Latin equivalent of ^l^CKn ^<#»4. >» 
 still another sense, is assa vox, Non. pp. 76-77 M. 
 
 kiass. kit. 279 ff., 384 ff. ; Norden, Kunstprosa, I , 55 ff. ; cf. 
 Ly, *' Connection between GreeR i\lusic and Poetry," in 
 Studies in Honor of Professor Gilderslecve, p. 2^5 ff. ; O. Jahn,^2.-2i,y 
 Hermes y II, 418 ff. A few ancient testimonies, several of 5^ ^^J, 
 which are not quoted in any of the works just named, but are ' 
 
 of the same general character as these quotations, may be men- 
 tioned here: Quintilian (I, 8, 2) shows at some length that the , ^ 
 reading of poetry/In the schools should be different from the/ 
 reading of prose {non qiiidcm prosac simili.\ and should be|^ '■»fi 
 
 some resemblance to song (idrmrn); cf. id. I, 10, 29; Plin. 
 £/f. IX, 34, 2 ; this was the regular custom of the schools in 
 the reading of Vergil, Euripides, and other poets, cf. Macrob. 
 Sal. I, 24, 6 (cane re); Plut. de Fort. Alex, 5, p. 328 d {aheiv); 
 Cassiodor. Var. 9, 21, 3 {decantare)\ Auson. Rp. XX 11, 47 P. 
 (fiexn et aeuminc vocis).^ Aristides Quint., who belongs to 
 the fifth century a.d., but who often draws from authoritative 
 early sources, expressly names a movement of the voice in- 
 termediate between song and speech, which is to be used in 
 the reading of the poets {dc Mns., p. 7, 23 M. /xeVr; hk {kl- 
 PT]ai^\ 7j ra^ rmv TroiTjpaTmv apayvcoaet^ TroiovfieOa)^ \ this state- 
 ment of Aristides is discussed in the Trans. Am. Phil. Assoc. 
 XXX, 48, by Dr. C. W. L. Johnson, who, after referring to the 
 fact that the points of pitch must have been more stationary 
 in this movement than in conversational utterance, reaches 
 the conclusion that "the metrical quantity of the syllables 
 must have been made more evident than is possible in the 
 case of the spoken sentence " ; simikirly Nicomachus states 
 that if, in any form of utterance, the quantities of the several 
 tones are distinctly rendered and also the intervals, the voice 
 passes naturally from speech to song {Euc/iirid., p. 4 M. el 
 
 * Christ, Parakat.^ pp. 175, 222, cites still other passages as proof "fur den 
 Gehrauch «les latcinischen cantare im Sinne eines bios rhythmischcn Vortrags, 
 nicht vollstiindij^cn (iesanjjes" ; he is quite mistaken, however, in wishing to limit 
 the term * sin^'' in this rhythmical sense to Roman usage. 
 
 ^ We find this statement of Arist. fully reproduced in Mart. Cap. IX, p. 353 
 Eyss. (quoted by Jahn, /./., 421) : continuum (genus vocis) est velut iuge conlo- 
 quium, divisum quod in modulatione servamus. est et medium, quod in utroque 
 permixtum ac nc(|ue alterius continuum modum servat nee alterius frequenti 
 divisione praeciditur, quo pronuntiandi mt>do carmina cuncta recitantur. 
 
 
5^ 
 
 Robert S. Radford, 
 
 [1904 
 
 Vol XXXV.] Studies in Latin Aecent and Metric. 
 
 S7 
 
 fUfi Tit . . . avayivdxTKoyu ye exBqXa fiera^v KaO* exaarov (f)06y- 
 ffm irtlfi Til phtj40% BiiHTdpcov teal fiera^dXXcoi; ttju (fxoptjp aw" 
 &Wim'€k iXhMv, irmoBrm' . . . #€Xcii£€ti' Xeyerai).^ Herodotus 
 repeatedly describes the Fythia as giving her responses (\e- 
 jmp, XP^p) in the hexameter or trimeter 'tune ' (* musical mode/ 
 •caience*: li* Ifa^eTpf toj^, I, 47» etc.; iv Tpifierpo} toW, 
 I, 174*; cf. the similar use. of iimimip), and Plato expressly 
 ascribes both rhythm and musical modulation (apfxovia = Lat. 
 emiemius) to 'spoken * poetry (Re/^. 397 d; 398 d), which he 
 classifies uader fiovaiK^ {ReJ>. 37^ ^ ^- *» /^/'^'^''^- 60 ^ ff.). I 
 may quote also tlic dcinitioa of prose given by Donatus (on 
 Ter. Em. 2. 3» H) ^^ 'that form of speech which is not mod- 
 lilated by song ' (fmrsa &raiio, quam mn inflcxit cantilena), 
 Fafther, although the ordinary recitation of poetry is usually 
 indicated by the expressions 'speak,' *read,' * declaim,' we 
 . find the terms * modulate ' and • sing ' also at times in use 
 ^•i*'ni^(Fricdlander, /./.; Christ^, 681), which imply the ^learr) k.'vj]^^ 
 tmj^v^ tr ■«ir<-/fiiiwi (Petron., c. 59; 68; Gell. XVIII, 5, 2). l^rom 
 these and similar notices we may fairly draw the conclusion that 
 Sf e»^ '^^'^^Rrt of tie elevated poetry of the ancients was intended 
 I for •mere declamation' in the modern sense, that is, intended 
 for the simple crvmxh's fcipnat^ or speaking voice. The case 
 is altugelher slniilar with the chanting of the oratorical 
 rhythms ; here also the rhetor's voice adopted a movement 
 >. intermediate between speech and song, as is expressly stated 
 ■ by Longinus (i?/ir/. I, 312. 14 Sp.), and is well attested for 
 j all, periods of antiquity (Norfcn. U), Finally, the view of 
 
 I 5 Jtist bef.>re Nicotn. ^rritcs: tA 5^ irepop (yivot) rb (rvvex^i, Ka(f 6 ofxiXoOfiiv 
 
 I Tiretf (•ciuaiitlties*) md imKCKpifiims dir' dWi^Xwy Trouicrdai. 
 
 2 So also Joseph. Ami, II, 16, 4 (Ir i^cLfiiTpt^ rdvip) ; Thcmist. On 2, p. 27 c, 
 Suidas, i.r., has the plum! : lIlPf««IP f » Tpo/xdirrevfia iv rbvoi^ i^afi^rpots. Passow's 
 Lex. eKplaiM Hmt in this use as ** die Hebung im Verse u. das dadurch entste- 
 hende lersmms'* ; llius tie «o«il appears to be used in the threefold sense of 
 musical tome (Sent Ewp. 757» 29 Bekk.). word-tone, awl verse-tone. In Latin 
 also wf should probaI>ly follow llahreM and the early editors in reading in Stat. 
 Si/r. \\ 3, 99: lim'i . . . duke vel heroos grcssu truncare tenores. Ho combine the 
 curtaik-.l pentameter with the accents of the hexameter * ; the Ms. reading, truncare 
 leones, which is retained by Vollmer (1898), api>ears to be corrupt and to yield no 
 satisfactory meaning. 
 
 Ctk. 
 
 i 
 
 Christ and Westphal, that, in cases of conflict, the Greek 
 musical accents were fully preserved in recitation, seems im- 
 probable; more likely Is Gleditsch's view (Jlctr.^ 86) that 
 they were preserved only in part (;/;/ ^i^vi.'ssr/// Grade'). 
 
 2. Probable Effects. -- The ancient Greeks and Romans 
 had not passed that stage of development in which a rigor- 
 ously exact rhythm is viewed as the chief essential of poetry 
 (Gummere, Bci^^innini^s of Poetry, pp. 82-115). Hence, as is 
 well known, they required in poetical recitation such a scru- 
 pulous observance of the rhythm as can only be paralleled by 
 the insistence of a modem audience upon correctness of tune 
 in singing (Dion. H. de Comp. 11 vied.; other references in 
 Ilaigh, Attic Tlieatrc\ 309). To this end tlie long and short 
 .syllables which vary greatly in length in ordinary speech and 
 do not exist, according to Westphal, in the form necessary 
 for exact rhythm, must be measured more exactly in terms 
 of the time-unit, the ;)^/3oVo? irpoyro'^^ esi)ecially must a long 
 syllable be given api)r()ximately twice the value of a short 
 one, the ratio which Aristoxenus (p. 18 W.) states existed in 
 .song, but which did not exist in speech according to West- 
 phal (IIF, 8 ff.). The careful enunciation which in this way 
 becomes necessary for the rhythmizing of speech is often 
 mentioned by the ancients. Thus Dionysius Thrax (first cen- 
 tury B.C.) declares that /prosody ' must be observed in read- 
 ing verse, and that the metrical art {rixvi)) of a poem is 
 perceived through 'trained reading according to prosody' 
 (avdyvwai^ ivTpi^r)<i Kara irpoacohlay i), where by * prosody ' is 
 evidently meant such a precise rendering of the quantities, 
 i.r. such an enunciation of the syllables in terms of the ;)^/ooW 
 TrpwTo?, as will produce rhythm 2; cf. also Dion. H. de Comp. 
 U : rj he pvOfiiKT] Kal p,ovaLKrj fiera^dWovaL {ra^ avWafid^i), 
 movaac Kal av^ovaai, kt\. Similarly Dositheus, in his inter- 
 esting account of the 'reading of verse in the schools, states 
 that he was required to recite with an observance of the 
 rhythm as well as of the pauses and clausulae {Colloqnia Scho- 
 
 J Ars, § I, Uhlig, p. 5. 
 
 ^ For the comprehensive meaning; of wootT^pUa in D., cf. Supplem, vetust., 
 I hlig, p. 107, and also Jannaris, A.J.P. XXIII, 77. 
 
58 
 
 Robert S. Radford. 
 
 [1904 
 
 Vol. XXXV.] Stitdivs in Latin Accent and Metric. 
 
 59 
 
 MsU Roder, p. 14: versus adnumerum et distinctum et clau- 
 sulam . . . reddo ad praeceptorem); cl. Qumtilian's remarks 
 upon £7^.ym'^/i^C precision *X IX. 4. 104 and 95, and the frequent 
 refercttces to poetry as •moulding' the pronunciation of the 
 fWHif C^.^ Hor. Ep, II, I, 126 OS . . . fignrat). This * moulded ' 
 pronunciation in its eitrenae form is the well-known irXaaiia, 
 the use of which the ancients often* disapprove for the simpler 
 inctrcs (Quint. I, 8. 2 ; t^), but require in the rendering of 
 the more difficult rhythms (Christ'^ 90. 682; cf. VVestphal, 
 III, i). It is Ctident that the effect of a carefully moulded 
 wpoacphia (cf. under i above) upon the word accent must often 
 have been very great, especially in the case of an accent which 
 is partly expiratory like the Latin, and the variable quantity 
 of many syllables in early Latin seems to result from this 
 fact. Thus the final syllable of domi, under the influence of 
 the word-accent, tended to shorten in popular speech, and is 
 often so used by Fl in the first foot, before the beating of 
 the time was fully begun ; but in those parts of the verse 
 where the time was strictly observed this syllabic has regu- 
 larly the value of two xpo'^o*. ^'•^- ^^''^^* ^"^ *^ ^^^^ pronuncia- 
 tion the expiratory accent must have been practically wiped 
 o«t(cf. V. Henry, Cmtp. Gr.\ Eng. tr., p. ^7 ; L. M tiller, R. AI^ 
 468)^; cf. ^#1////^^/ (where the verse-accent also assists in the 
 
 I Vet by MO metns always ; for we fiiwi in the commentary ascril)e<l to Trobus 
 (Verg. f.fl p. 6) : Acneida ciuoiitani plasmate legi vt.lehat, alt • arma virumque 
 ■mnP.* This whole <l»ciissioii of Probus upon the form of poetical delivery (pro- 
 mmiiaiU) is highly tiiKCstive, in spite of the arbitrary diclum that ih^ plasma 
 m singing deliirery (riin/iire) Is to be employed in a poem (only) from the point 
 where the words emmere or cmrmm first oecilf. It is noteworthy tliat the general 
 teaching of the commentary agrees perfectly with the well-known statement of C*el- 
 lius (VI, 7, 4) that Valerius Probm and his pupils accented ixddvenum in verse. 
 For the meaning of irXda/wi, cf. Spalding on ( )uintil. I. 8. 2. who quotes some 
 ancient authorities, and also the excellent modern definition of Sahnasms : " vocrm 
 mollem et eliquatam, quae per omncs sontis intentionesque varietur/' For numer- 
 ous references to the abuse of this singipg delivery, see Jahn, /./., 4^2. n. I and 2. 
 
 2 At the same time the rhythmical tone falls upon the ultima. In a somewhat 
 similar wav Zieliiiski in his recent valuable work, Bm Clauselgeseiz III Gmm 
 Heden, Leipzig, 1904. reaches the cuitclision that iambic words are oxytone m the 
 oratorical rhythms, e^. firimt: cf. p. 239: -Immerhin ist zu betonen, .lass fur 
 den rednerischen Accent die iambischen und anapastischen VVikter, mi Gegensatj 
 2ur gelaufigen Auflfassung, Oxytona sind." This oxytonesis is of course not the 
 
 shortening) and omittebat. In a similar way our ancient 
 authorities recognize also for prose an easy and colloquial 
 pronunciation, which is called sermo, or oratio ninissa (Auct. 
 ad Her. Ill, 13, 25), but they prescribe for elevated language 
 an emphatic kind of utterance which involves a sustained 
 effort of the voice and is called cofitnitio, or oratio acris{A\\Qt. 
 ad Mcr. ibid.; Cic. Off. I, 37, 132 ; II, 14, 48, etc.); cf. de Or. 
 Ill, 45, 177: non enim sunt alia sermonis, alia contentionis 
 verba, . . . sed ea nos cum iacentia sustulimus e medio, sicut 
 mollissimam ceram ad nostrum arbitrium formamus et finjri- 
 mus. This contentio vocis^ or tense voice, passes readily over 
 into the singing or modulated voice ; cf. Cic. Off. I, 37, 133 : 
 sine contentione vo.x nee languida nee canora. 
 
 It is probable, then, that the limits of variable pronuncia- 
 tion have been sufificiently great in Latin to admit the exist- 
 ence of two distinct forms of pronunciation : i) A formal and 
 dignified pronunciation which is strictly quantitative, and 
 which in its extreme form — the ^learf Kivr}aL^ — usually 
 possesses no appreciable stress-accent.^ 2) An easy and 
 
 normal accent {i.e. Zielirfski's * Vulgaraccent ') ; for the poj^ular speech pronounces 
 diSm", with 'half-long' ultima, and hence, as is well known, the Romans do not 
 admit in their poetry two iambic words in succession in the rhythmized form 
 domh Consequently I greatly prefer the form of statement adopted by Zieliiiski, 
 p. 242: "Die quantitativ bevorzugte Silbe hatte die Tendenz, im Re<lefluss zu 
 einer tonisch bevorzugten zu werden. . . . \'.jllends in der rhythmisch bewegten 
 X^tA^ lag die Versuchung nahe, die accentuirte Kiirze vor dcr Lange zuriicktreten 
 zu lassen." On the other hand, I am not prepared to accept Zieliiiski's conclu- 
 sions upon fmerent as the sole form of the oratorical accent for anapestic words, 
 since he himself hesitates greatly (p. 231) between the accents fdcerent and 
 ficermt. Finally, Zielinski's conjecture (p. 241 f.) that the ' oratorisch-poetischer 
 Atfent'' is identical with the archaic popular accent of the third century B.C., 
 si-rms to me unnecessary; it is rather to be viewed as the accent which the 
 rhylhmizirsg ])rocess and a precise obser\-ance of quantity (m^ctt; Klv^aii) engender 
 in the language of all periods. ( )ne very late ancient grammarian, Vergilius Maro 
 (Kcil, Suppl.^ 190). actually professes to recognize in some form of Roman pro- 
 nunciation the accents ri-ges, amd^ etc., but we can of course give little credence 
 to such a source. 
 
 ^ ^PPS""*^"* '^fetrisch. K'omp. d. Ter., who has a general discussion of this sub- 
 ject, does m>t go so far, yet he writes (p. 113, n. 1) : " Bei den verschiedenen 
 Bewegungen innerhalb der konstanten \'ersbewegung kann doch gar recht diese 
 Stimmhohe . . . wenig oder schvvach herausgehort worden sein, . . . wie solche 
 Acbenljcwegungen beim musikalischen Dirigieren immer eingehalten werden"; 
 
6o 
 
 Robert S, Radford. 
 
 [1904 
 
 CDllaqiiial pronuBciation, in which a weak stress-accent is 
 developed at the expense of quaiitity.i All Roman poetry, 
 since it is at once quantitative and a form of the fiearj K^vrj^i^, 
 is based (with the exception of the unusual quantities in early 
 Latia) solely upon the strict quantitative pronunciation ; it is 
 to be noted, however, that, nilder certain favorable quantita- 
 tive conditions, an accentual pronunciation has been devel- 
 oped in the Latin language as a whole, and has greatly 
 weakciied all simrt syllables immediately preceding or follow- 
 ing the accent, e.g. g/fiim, p/ciSm, ^irire, etc. 
 
 Law of the Relation between Word and Verse- 
 accent. — In any case, whether the Latin accent be wholly 
 iimsica!, or, as seems more probable, in part also expiratory, 
 it is certain that Latin possesses a single colloquial 'cadence ' 
 or • harmony,' viz. the accentual* (Aristotle's 17 Xe/cTi/ei; ap^ovla, 
 Piffi. 6 ; Aristoxenus's Xoywhe^ rt fieXo^, Harm. Elcm. 1,17 W.); 
 with the cniployment, however, of a stricter quantitative pro- 
 niinciation and a special arrangement of the quantities, it 
 acquires a score or more of metrical 'cadences' or 'harmo- 
 Hies,' such as the hexameter cadence (Hdt.'s e^dfierpo^ toVo?), 
 pentaineter cadence, Sapphic cadence, iambic trimeter ca- 
 dence, etc. As Is natural in different spheres of time and 
 tone, the metrical cadences in general are entirely inde- 
 pendent of the accentual cadence, but among them all there 
 is one, the laTObic, whose distinguishing characteristic in both 
 Grcelfc and Lalln, as- we know from many ancient testimonies.^ 
 
 c£ iibotc, p. 54 ff. Perhaps we can only say with certainty that the worcl-acccnts 
 were very greatly weakened; for so g «ocl an authority as Ilanssen writes (Zur 
 hi, fi. roman. Meirik, Valparaiso, 1901, p. 51) : " Es konnen ohne die geringste 
 Schwierigkeitineinem und demselben Vetie zwei rliyttiniiscllc Stromungen neben 
 einatulcr hergAen." Cf, also Christ « 59* Schdll, Ek me., p. 25 ff., and Professor 
 Hale, Proc. Am. f^il Assm. XXIV (l895)« P- ^^^''' ^^ '^ certain, however, that 
 th ; verse-tone is the pfc4o«inml tone in all cases where the words are properly 
 
 rhvthmi/ed. 
 
 1 These ranant prontinaiffons mpMn the well-known fact that Greelc loan- 
 words often assume a double form in Latin, one form showing a retentiim of the 
 Greek accent, as I'hifippus (PI.), tvrannm (Sergius, IV, 528 K.), cf. Apmnis 
 (•Air6\Xwws). the other a retention of the Greek quantity, as Philippus, tyrannm. 
 
 -^ Cf. Cic. Or. 17, 57 : est etiam in dicendo quidam cantus obscurior. 
 
 « Afist. PaeL 4/11,; MkeL 3, 8 ; Demetrius, r. ipfi. 43; Hephaestion, p. 19 
 
 Vol. xixv.] Studies in Latin Accent and Mctnc. 
 
 61 
 
 is to approach more closely than the 'rest to the tone of 
 colloquial speech.* Hence it is not surprising that the poet 
 or poets who arranged the Latin iambic and determined its 
 orio-ina/ fm-m, imposed upon it the law that it should agree 
 with the accentual cadence in the pronunciation of g^'nar, 
 Sp^rire, etc.,^ and doubtless also in the phenomena of the dipo- 
 dic law; further, since the popular iambic cadence tends to 
 be associated more or less closely in the Roman mind with 
 the accentual cadence, it is possible that the popular poets, 
 even after the determination of the original verse-form, 
 tended, in some measure, to prefer agreement to disagree- 
 ment in neutral cases, il#, in those cases where disagreement 
 was not the usage of the verse (in the manner explained 
 below). Such a tendency as this last would, however, be 
 contrary to the usual rule of Latin verse, and its existence 
 must consequently be coasidered doubtful.^ 
 
 Finally, we may state the general law which appears to 
 have governed Latin verse in its relation to accent. The 
 Latin metrical cadences or verse-forms were originally con- 
 stituted entirely without reference to accent, and solely in 
 accordance with metrical laws ; no sooner, however, had they 
 assumed a definite form in this way than they began to 
 
 W. ; Aristid. Q., p. 38, 19 M. ; Cic. Or. 55, 184 ; 56, 189 ; 57, 191 ; 58, 196, 
 etc. ; Hor. A. /'. 81 ; Quintil. IX, 4, 75 f. — Some of these passages arc cited by 
 Christ^, 315. 
 
 ^ Vet even here the wXdafm was employed to some extent, as is expressly 
 iiSLted for f/te comic iambic trimeter by Quintiiian, XI, 3, 183: pronuntiatio ges- 
 ticulalioni!)us molesta et vocis mutationibus resultans ; cf. also Klotz, Grundz. 
 381. 
 
 - See the really excellent discussion of the verse-accents gcnere^ cSrpore^ in 
 Klotz, Grundz. 268-280. 
 
 '^ The most general principle that can be stated here is that, after the determi- 
 natjnn of the norm, any consiileral)le deviation, whether favorable or unfavorable 
 to the accent, was a departure from literary usage and, to that extent, a vulgar 
 usage ; see the striking examples cited by Munro. The literary language is 
 especially careful to avoid too close an approach to the accentual cadence (the 
 ordinary tone), as is clearly implied in Cicero's statement, Or. 55, 184 ; cf. Christ-, 
 p. 59. On the other hand, popular poetry must not be entirely removed from the 
 spoken tone, and the iambic poets have perhaps sought to avoid a succession of 
 disagreements {continuatio oxytonarum vocum) in the neutral parts of the verse; 
 for the usage of Phaedrus, cf. Laugen, Rhein. Mus. XIII, 198. 
 
R&beri 5. Radford. 
 
 [1904 
 
 Vol. XXXV.] Studies in Latin Accent and Mttric, 
 
 ^l 
 
 respond to the iniuence of the accent in accordance with 
 a simple psychological law. In view of the uniformity of 
 tic Latin accentual system, the result of observing the met- 
 rical fttlcs was. U produce at^ ■certain points of almost every 
 cadence i) agreement of accent and ictus in the great 
 majority of cases, 2) disagreement of accent and ictus in 
 the great majority of cases. Wherever this result is brought 
 about, the Roman ear is quick to note the relation which 
 usually exists, and to rec|iire In tlic end, is. in the course of 
 the historical development, that it shall always exist, that is, 
 to require that the agreement or disagreement shall be made 
 invariable; in other words, the Roman ear renumbers the 
 hexaittclcr or tie &ppbic or ti€. trimeter cadence at certain 
 points by tie relation which it bears to the accentual cadence, 
 which is its simplest and most familiar standard of measure- 
 ment In those more numerous parts of the verse, however, 
 wienj. BO' nsnal: relation between accent and ictus was estab- 
 lished, but tie effect of observing the metrical rules was to 
 produce sometimes agreement and sometimes disagreement, 
 tic metrical cadence remained wholly unchanged, wholly 
 tiiMtiiSlsted' 'by tic accent, so long as the Roman ciuantitative 
 system endured. 
 
 What relation, it will be asked, does the explanation just 
 given of the development of Latin verse bear to the views of 
 Ritselt mA Ms iMmcrous followers? So far as concerns 
 cases of agreement in caesurae and in verse-closes, this ex- 
 planation agrees with Ritschl's in recognizing the fact that 
 the agreement has been brought about through the influence 
 of the ,aCCC!ll,: but differs from Ritschl's in holding that agree- 
 ment ii^. not' soiigit /#r m^^ bat: in consequence of fam.iliar 
 association. So far as concerns cases of disagreement in 
 caesurae and in verse-closes, the divergence from Ritschl's 
 view is still greater. For Ritschl, as is well known, started 
 from the assumption that tic Bomiil poets founcl aelnal 
 pleasure in the agreement of letn* and accent, and hence 
 sought to produce this agreement as often as possible. So 
 acute an observer did not, however, fail to recognize that the 
 Romans had mniiilitakabiy songit to^ produce disagreement 
 
 also in many parts of their verse, e.g. while agreement is 
 sought in the second half of the dactyHc hexameter, disagree- 
 ment is required in the first half of the same verse.' Hence, 
 ill a well-known passage of his writings (^^///.r. II, Leipzig^ 
 1868, p. xii). Ritschl sought to explain this latter fact by his 
 famous assertion that the special charm of ancient verse lay 
 in the balancing of agreement and disagreement, in the pro- 
 duction of an ' hannonischc Disharmonie: According to this 
 theory, after having sought agreement in one part of the 
 verse for the pleasure which it gave, the Romans next sought 
 with pleasure {' suchten mit WoJilgcf alien') to produce dis- 
 agreement, in order to contrast the two parts of the verse, 
 and seemingly also (though this is not expressly stated) in 
 order exactly to balance the two opposing forces, quantity 
 and accent. Every one will recognize that Ritschl in this 
 account has truly and graphically described an effect which 
 has been actually produced in many kinds of Latin verse; 
 at the same time the process, through which this effect has 
 come about, is explained in a quite impossible manner, or 
 rather is left in large measure wholly unexplained. For how 
 could the Romans seek agreement with pleasure, and then 
 for the sake of an abstraction, such as ' balance ' or ' contrast,' 
 seek with pleasure — disagreement ? In point of fact, neither 
 aesthetic pleasure nor aesthetic pain plays any part in this 
 process, which is purely a psychological matter of association. 
 It remains to note that the general law of the accentual 
 development of Latin verse, which I have given above in 
 my own language, was first clearly stated by an eminent 
 American metrician and a former student of Ritschl's, Pro- 
 fessor M. W. Humphreys, in a careful study published in the 
 Tram. Am. Phil Assoc. IX (1878), pp. 39-58.^ The article 
 in question deals chiefly with cases of agreement in the 
 dactylic hexameter, but the same principle is expressly applied 
 
 ' And still earlier in the dissertation, Qiiaest. metr. de accentm memento in 
 
 r^rm heroua. Lips. 1874. See also the able paper of II. A. J. Munro, Trans- 
 
 >^'iims 0/ tke Camhndge Philosophical Society, Vol. X (1864), pp. 374-402. 
 
 Juniphreys an.l Munro reach quite independently precisely the same conclusions 
 
 respecting the intluence of the accent upon Latin verse. 
 
o4^' 
 
 Robert S. Radford. 
 
 [1904 
 
 in one passage to the explanation of cases of disagreement 
 (p. 40). Again, Professor Humphreys has limited his rule 
 in its formal statement to the dactylic hexameter, but I cannot 
 doubt that he regarded it as applicable to other Latin verse- 
 forms also.* Since Professor Humphreys* solution of this 
 difficult i)roblem leaves none of the phenomena unexplained, 
 and is in accord, as he has very fully shown, with the histori- 
 cai development of Latin verse, it seems probable that it needs 
 only to be better known ''W supersede in this field the less 
 complete and less carefully constructed theories of Ritschl 
 and of Meyer.^ 
 
 1 Professor Iluninhreys does in fact state his coEclusions in a more general 
 
 fiiriii, jy^tts. Am, I%il, Jssm. Vllf 111, 
 
 * .-idttcmia : 
 
 p. ^_jj, n,, — for siipptseil cases ol the itse of the apex in Latin Imicrr. to 
 denote the ictus, v., Christiansen, IJe apicihtts, \\ 20. 
 
 P. 56, n. 2.— The interpretation which 1 place upon the phrase h l|. rbvo% 
 requires p^trhaps t;,> I)-* more clistiiutly stated. In my jud^'ment 6 c^. rbvo^ is the 
 hexameter /iifn/c, and is ijarallel to the I A'dian modt\ the Phr>gian ///o*//, and the 
 like. The sptiken tone also l)€loiig.-i, as the ancients often recognize in a general 
 way, among the musical modes, but, in the chanting or intoning of verse, it is 
 regularly made sulM»rdniate to the new mode which accompanies the verse. If 
 this interpretation of rhvm is correct, there will be found no real inconsistency in 
 the various rrfcrtnces which I have made t» the meaning t»f the term,; cf. also 
 A,J:P. XX, V, 420 (421), n. 5. 
 
 F. 59 ((k>), n. I. — Compare the somewhat similar vi:*ws of Gocxlell. Chapters 
 in GreeJk J/efrit:, Xew York, icpi, p- 167 f. Several other tojiics mentioned in 
 the preceding discussion are also very clearly tr^«ated l,>y ,Pr»„»fess*,ir Goodell, vi/. 
 the elasticity ».»f sylla!»ic cjuaiitity (pp. loo, 112),, the strict »„>bservance of ihythin 
 
 (p. 125), the use of wMfr/M (p. 129, etc.). IVt.fessor ( loodell is quite ri^ht in 
 
 rejecting (p. 125) Wcstphal's doctrine of "a sharp separation between the rhytlua 
 of song ami that of spoken verse"; on the other han I he appears to rely too 
 much iit' his exposition ufion modern parallels (p. 128), and to take too little 
 account of the very important change which has taken place within historical 
 times in the usual method of poetical delivery, although, to be sure, we find 
 wXdfffia aptly defined (p. 50) as **the more exact observance of rhythm that guts 
 naturally with the singing voice." 
 
 , JS^aAi'^ 
 
 7^ — X" A/' ' ^