FROM THE LIBRARY OF

REV. LOUIS FITZGERALD BENSON, D. D.

BEQUEATHED BY HIM TO

THE LIBRARY OF

PRINCETON THEOLOGICAL SEMINARY

Section /

Digitized by the Internet Archive

in 2012 with funding from

Princeton Theological Seminary Library

http://archive.org/details/stressaccOOdubo

Columbia ©ni&etwtig

STUDIES IN CLASSICAL PHILOLOGY

THE STRESS ACCENT IN LATIN POETRY

'?&$*&'

/

THE STRESS ACC IN LATIN POETF

BY

yy

ELIZABETH HICKMAN DU BOIS

THE COLUMBIA UNIVERSITY PRESS

THE MACMILLAN COMPANY, Agents

LONDON: MACMILLAN & CO., Ltd.

1906

All rights reserved

Copyright, 1906, By THE MACMILLAN COMPANY.

Set up and electrotyped. Published August, 1906.

Norfoooti ^rcss

J. S. Cushing & Co. — Berwick & Smith Co.

Norwood, Mass., U.S.A.

PREFATORY NOTE

This monograph contains a condensed and care- ful summing up of the most authoritative evidence with regard to a stress accent in Latin. On the basis of the doctrine here set forth, Miss du Bois has formulated an ingenious and very plausible theory of the Saturnian Verse, and has sought to establish an explanation of the purely quantitative Latin poetry which shall reconcile the opposing views as to an apparent clash between word accent and verse accent. I regard her discussion as a valuable contribution to the literature of this highly controversial subject.

HARRY THURSTON PECK.

Columbia University, July i, 1906.

THE STRESS ACCENT IN LATIN POETRY

WORD ACCENT

Accent is the prominence of one syllable of a word over the other syllables. It is the essential part of a word,1 its cachet. Because the Romance languages have preserved the accentuation of the Latin, they are, as Gaston Paris says, "deslangues filles" and "des langues sceurs," while, though many French words have been borrowed by Eng- lish and German, because the Teutonic accent has been substituted for the Latin, the whole physiognomy of the word is changed.2 This predominance of one syllable of a word over the others is accomplished by pronouncing it at a higher pitch and with increased stress of the voice, the two factors varying in importance both, abso- lutely, from one language to another (often be- tween different dialects of the same language)

1 anima vocis, Diomed. p. 430, 29 K ; Pompeius, p. 126, 27 K.

2 Gaston Paris, £tude stir le Role de r Accent latin dans la Langue francaise, Paris and Leipzig, 1862, p. 9 ss.

B I

2 THE STRESS ACCENT IN LATIN POETRY

and, relatively, between different modes of utter- ance.1 So the musical accent of ancient Greek and of ancient Sanskrit is essentially one of pitch, differences of stress playing a subordinate and all but negligible part. English and German, on the other hand, have a stress accent, though differ- ences in pitch are still important. In English, the word really, for example, by variations in pitch may be made to express a wide range of feeling from mild interest to profound contempt.

There is another factor in Greek and Latin which helps to make the accented syllable promi- nent, though it is not in itself sufficient to consti- tute such prominence ; and that is quantity. Professor C. E. Bennett2 maintains that Latin, in the Classical Period, at least, was " absolutely unstressed." He writes: "May not a syllable be primarily prominent by virtue of its quantity? That is, in a word like amavit, for example, may not the rule of the grammarians, that such a word was accented on the penult, simply mean that they felt the quantity of the long penult as making that syllable prominent, without any stress on the one hand or any elevation of pitch on the other ? And in words like latuit homines y etc., may not the rule that these words were accented on the ante- penult simply mean that, in consequence of the short penult, that syllable did not possess any

1 Cf. Eduard Sievers in Paul's Grundriss, 1897, 1 Bd» 2 Lief- p. 304 ss. 2 A, J, F, vol, xix. p. 362 et ss.

WORD ACCENT 3

prominence, and hence after the establishment in Latin of the three-syllable law, the syllable next preceding became the conspicuous one ? " Take the word amavit ; the penult is an " open " syllable (to quote his own terminology 2) with a long vowel ; it is, therefore, a long syllable. The ultima is a "closed" syllable, "and a closed syllable is pho- netically long."2 There is therefore no difference in quantity between the penult and the ultima, so that it is difficult to see how the former could be "quantitatively prominent." Further, in the word homines, because of the short penult, " the sylla- ble next preceding becomes the conspicuous one." How ? Both penult and antepenult are short. The only long syllable in the word — the only one, therefore, which can be said to possess " quantita- tive prominence" — is the ultima; so that, follow- ing his own rule, the word should be accented on the ultima. Latin possesses a very large number of long, i.e. " quantitatively prominent," syllables, so much so that Plautus and Terence were obliged to shorten many such syllables by the law of Brevis Brevians, and Ennius and his successors still more. In an iambic word like mbdo, for example, what influence was at work to cause the

^Appendix to Bennett's Latin Gra?Jimar, Boston, 1895, p. 32. But cf. Pompeius, p. 112, 26 K.

2 The only exception would be where, with no break in the sense, the following word began with a vowel. Before a pause and, as he expressly states, at the end of a line, such final syllables are long. — Op. cit. p. 375 note.

4 THE STRESS ACCENT IN LATIN POETRY

shortening of the final d ? If quantity alone was responsible, why did not a short unaccented syl- lable produce the same result? But iurigo (in Plautus) becomes, later, not iurigo, but iurgo. Nor is this syncope of the unaccented syllable, which can be due only to stress, confined to ante-classical times, when, according to Professor Bennett, the language may not have been " absolutely un- stressed." Augustus stigmatized calidus for cal- dus as a piece of affectation, " non quia id non sit latinum, sed quia sit odiosum," 1 while It. caldo shows that caldus was the form in late Latin. It is, in fact, precisely the " quantitatively monoto- nous " character of Latin that makes some other principle of accentuation imperatively necessary. But such a thesis as that of Professor Bennett cannot be seriously maintained for any age or any language.

Behind the lyric and epic in Greece, as every- where else, there must have been rhythmical songs of the people, but so imperceptibly does this Volks- poesie shade off into the Volksthiimliche Poesie of later and more cultivated times, so industriously is every motif made a subject of art, and, withal, so national and democratic is the whole body of Greek poetry, that the first rude songs of daily life and of worship — at least in their original form — stood small chance of being preserved.2 It is a tempta-

1 Quint, i. 6, 19.

2 Smyth, Greek Melic Poets, London, 1900, p. 488 et ss.

WORD ACCEXT 5

tion, however, with Christ1 and others, to see the influence of stress in the Lesbian Mill-Song, quoted by Plutarch:2—

aAei, fxv\a, aAei

KCLL IIlTTaKOS jap aXcL

[xeyd\a<s MvriAavas /3acriAeva>v,

where the last line, at least, seems to match the rhythmical movement of the hand as it turns the mill. Keller3 adds the saying of the children of Attica when they first saw the birds in spring. It is from the Scholiast on Aristophanes' Birds,

1.54:-

86s to ctkcAos rrj 7rcToa kclI TreaovvTaL to. opvea.

He reads : —

805 TO (TKcAoS TT] 7T€Tpa KOL 7T€oWvTCU TLOpVtd.

He also quotes the Tortoise-Game from Pollux, ix. 125, where he sees in the long i of rt (w. 1 and 3)

the influence of stress : —

\€Xl^€Xo)V7} Tl TTOLCLS €V TO) fxicTU) ',

Zpux fj.apvop.aL Kal KpoKtfv ^ItXrjcriav. 6 8' tKyovos crov tl ttolojv oAcoActo ; XevKwv a<fi lttttcov eis ^aAacrcrav aAaro.

The lengthening of a short vowel in an accented syllable and the shortening of a long unaccented

1 Metrik der Griechen und Romer, Leipzig, 1879, p. 374.

2 Sept. Sap. Conv. 14 (157 E).

3 Der Satumische Vers ah rhythmisch erwiesen, Prague, 1883, p. 81 et s.

6 THE STRESS ACCENT IN LATIN POETRY

vowel is a sure sign of the growing power of stress, which, like every strong influence in the history of humanity, works its way up from below. Setting aside the fishmonger's rdpcov /3o\cov and /ctco /3o\a)v for rerrdpcov 6f3o\a>v and oktco 6/3o\cov in Amphis' comedy, IlXaVo? and the occasional suppression of a short vowel on Attic vases, e.g. iTrotTjcrv 'AOrjvrjcrv, — instances which may seem to be entirely sporadic only through the losses of centuries, — there is con- siderable evidence for the confusion of long and short vowels as early as the second and first cen- turies before Christ.

Kretschmer, in an excellent article,1 which by the way is entirely misrepresented by Vendryes,2 has collected, from papyri and inscriptions, a number of instances of this confusion. In con- clusion he writes : " Die oben zusammengestell- ten belege aus papyri und inschriften zeigen noch kein durchgehendes abhangigkeitsverhaltniss zwischen vocalquantitat und betonung. Es finden sich schreibungen wie yiyoirco, fcarcoxjj, oopoicos, coparcu, 7r/3a)e<7Tft)TO?, €%&)*> st. e%ov, fxei&v st. fiei^ov und veorepov, irapaTV^pv st. -Tv%d)v. Aber in der mehrzahl der falle sind betonte kiirzen als lang oder unbetonte langen als kurz bezeichnet : man vergleiche Ma/eeSo^o?, &Wo?, TrpaHceincu, <w7Tft)?, /3oa>?, fieyaXcoSo^ov, wvoixa, iBcodrj, BcaBco^a), irpoare-

1 Kuhrts Zeit. xxx. p. 591 et ss.

2 Recherches sur VHistoire et les Effets de VIntensite initiate en Latin, Paris, 1902, p. 34.

WORD ACCENT 7

Ta%a)T(ov, ^o?, reOrjafjiat, ivvr)a, avSpet, andererseits TrpoaoTTOV) eSo/ca, evtyovov, eyvov, fiaprvpov, ixeOoiropt- z>o'?, aireWdyrjV^ KaTaarpovvve^ <J>tXoz^o?, 'Apiaro- vlhas, faXocfrpovos st. -(£/doW>?, etc. Thatsache ist also, dass die vulgare aussprache bereits im 2. jahrh. v. Chr. lange und kiirze zusammenfallen Hess. Mit der aufhebung der quantitatsunterschiede fiel aber eine der wichtigsten voraussetzungen fur die urspriingliche musikalische betonung fort; denn der unterschied von acut und circumflex sowie das ganze sogen. dreisilbengesetz sind durch die ver- schiedenheit der quantitaten bedingt. Hieraus folgt, dass die betonung der griechischen volks- sprache schon in vorchristlicher zeit eine nicht unwesentlicheveranderung erfahren haben muss." Finally, Westphal 2 shows that in the later Greek times there arose a kind of didactic poetry whose appeal was directly to the people, through fables told in choliambic verse. It is, however, of the utmost importance to note that, while the ancient verses of Hipponax and Aeschrion were based solely on quantity, this new verse required that in the last foot, word- and verse-ictus should al- ways coincide. Unfortunately we do not know the date of Babrius, who first used this verse. It has been variously given all the way from the third century before Christ to the third century after Christ. Crusius,2 after giving the arguments

1 Allgemeine Metrik, Berlin, 1892, p. 242 et ss.

2 De Babrii Aetate (Leipzig, 1879).

8 THE STRESS ACCENT IN LATIN POETRY

(which often rest upon very slight grounds) both for an early and for a late date, decides in favour of the time of Alexander Severus. In the Byzan- tine Period, the choliamb of Babrius had lost all trace of prosody and had become a verse of twelve syllables in which it was only required that the last ictus coincide with the accent of the word —

Ancient choliamb w_w__w_w__^ ^»

Choliamb of Babrius w_^_w_w_w_Z.vy- Byzantine choliamb vywwwvywwwww^w-

Many of the so-called Political Verses {cttlxoi ttoXltlkoi) of the Byzantine writers employ this measure of twelve syllables, as, for instance, the following lines of Tzetzes : —

TrpoXoyos eort ^XPL X°P0^ euro Sou* e7re«roSiov icrriv, ws /cat 7rpoe<j>r)Vf A.oyos fxera^v 7r\r)V fxeXwv xppuv 8vo.

It would seem, therefore, that the history of Greek poetry shows the same successive phases as that of Latin. Rhythmical at first, in all probabil- ity, though the finer poetical sense of the Greeks may not have allowed the suppression of the thesis, so frequent in Teutonic popular poetry,1 it had become quantitative long before the period of the Homeric epic, and for more than a thousand years had so remained. Then, through the influ- ence of the people, its musical accent became less nuance', the fine distinctions of pitch gave way to

1 Usener, Altgriechischer Versbau, Bonn, 1887, p. 78 et ss.

WORD ACCENT 9

the heavier, more palpable differences of stress, and along with stress as a dominant principle came in a poetry which ignored quantity altogether and only required that in the last foot of each line (in the longer lines, of each hemistich), word- and verse-accent should coincide.

Just as all the dialects of Greek have a common system of accentuation, and all the dialects of Ger- man, so, Corssen1 thinks, have all the old Italic dialects. For Oscan and Umbrian, at least, it seems clear that the accent {i.e. stress) fell at one time on the initial syllable of the word. This is proved by the same phenomena as in Latin, namely: i. Syncope of the vowel, which under the later Penultimate Law would bear the accent

(a) in the antepenult; as, Osc. Anagtiai from Anketidi or *Angetiai (Lat. Angitiai); Osc.-Umb. nessimo- perhaps from *nezdismmo- or *nedhism- mo- ; Vo. atahus perhaps from * ad-tetahnst (like Lat. attigi from *ad-tetigi, reccidi from *rececidi)\

(b) in a long penult, the Oscan proper name Opsci, from *Opisei, Osc. minstrels (mistreis) from *mini- streis (Lat. minister). Syncope in these positions is more widespread in Oscan and Umbrian than in Latin. 2. Weakening of the vowel in the same positions, which is rare and doubtful ; for example, Umb. pre hub ia y Lat. praehibeat.

Whether this initial accent was preserved in

1 Uber Auss. Vok. u. Beton. der lat. Sprache, Leipzig, 1870, ii. p. 907 ss.

10 THE STRESS ACCENT IN LATIN POETRY

Oscan-Umbrian or replaced by the three-syllable law, as in Latin, cannot be determined with cer- tainty. Brugmann,1 on account of the widespread loss of the vowel in final syllables, is in favour of the former view; Corssen, with whom von Planta is inclined to agree, prefers the latter. Von Planta writes,2 " Wenn auch nicht alle angef iihrten Argu- mente von gleichem Werth sind, so scheinen sie doch ausreichend um es entschieden wahrschein- lich zu machen, dass im Osk.-Umbrischen in his- torischer Zeit die jiingere lateinische Betonung herrschte. Uber der Zeitpunct der Aenderung lasst sich nur so viel sagen dass er spater fiel als die Syncope in osk. A?iagtias, Vezkei, umbr. mersto-, etc., und als die Schwachung in umbr. /r^- hubia (osk. Mamerttiais ?). Dass die Aenderung uritalisch gewesen sei, ist aus verschiedenen Griin- den unwahrscheinlich."

As to the nature of the free shifting accent, claimed by many to have preceded the stress accent on the initial syllable in Latin, we are ignorant. Vendryes3 claims that it was a pitch accent like that of the parent Indo-European, but he adds, " Ce ton n'a eu aucune influence sur la constitution et le developpement de la phonetique latine." Conway, Wharton, Collitz, and others think it was a stress accent, and see in certain vowel

1 Grundriss, i. p. 553.

2 Gram, der Oskisch-umbrischen Dialekte, Strassburg, 1892, p. 596. 3 Op. cit. p. 99.

WORD ACCENT II

changes, for example the a in quatuor {Gr. rerrape^) and in magnus(ln&.-¥.u.r. *meg-nos, Gr. /xeya?), traces of its influence. This earliest accent was, however, replaced, as Corssen proved, by a stress accent resting on the first syllable of each word. In- stances of Syncope under the Initial Accent Law are, ancidus for ambi-quolus (Gr. a/jL<j)L-7ro\o<;), naufragus for ndvif vagus, selibva for *semilibva, undecim for *oinidecem ; vettuli for ve-tetuli, veppevi for ve-pepevi, veccidi for ve-cecidi; of Vowel Weaken- ing, infvingo from in and fvango ; concldo from cum and caedo ; tviennium from tvi- (tves) and annus.

Sometime before the dawn of the Literary Era (Stolz conjectures the fifth century of the city1) the Initial Accent in Latin yielded to the law of the Last Three Syllables. Vendryes, who holds that the former was a stress accent and the latter a pitch accent pure and simple, makes no attempt to explain the manner of the change. Lindsay,2 who regards both as essentially stress accents, thinks that the change began in long words like sapientia, tempestatibus, which, in order to be pro- nounced at all, must have had a secondary as well as a main accent, and that the change from the older accentuation to the Penultimate Law of the Historic Period, consisted merely in substituting the main accent for the secondary, and the secon- dary for the main ; sdpihitia becoming sapie'ntia,

1 Lateinische Grammatik in Iwan Miiller's Handbuch, II. s. 321.

2 The Latin Language > Oxford, 1894, p. 158 ss.

12 THE STRESS ACCENT IN LATIN POETRY

te'mpestatibzis, thnpestdtibus, etc. In one particu- lar this change appears to have been still incom- plete at the time of the Early Drama, words like facilius, balineum {y w \j ^) having the metrical ictus on the first, not on the second syllable, in the plays of Plautus and Terence. So also (I think) capitibus in Naevius' line: —

Noctu Troiad{e) exibant capitibus opertis.

The very fact that the place of the Latin accent was so circumscribed points to an essential differ- ence between it and the pitch accent of ancient Greek. Except in a few words which have dropped or contracted their last syllable like acids, ilhic, tanton (tantonc) the accent never falls on the ultima, but is determined rigorously by the quan- tity of the penult, even Greek loan words like Epirus, tyrdnnus, submitting to its heavy, hand. On the contrary, all the pitch accents that we know have a far wider scope. In ancient Sanskrit the accent may fall on the seventh syllable from the end. Greek has a recessive accent, which is only provisionally established for Latin in con- ventional word groups.1 In Chinese, the only pitch language of modern times, the tones, of which there are five (some say four or seven), seem to play all about a word combination like veritable

1 Radford in A. J. P. vol. xxv. " On the Recession of the Latin Accent in Connection with Monosyllabic Words and the Traditional Word-Order." (Three articles, pp. 147, 256, 406.)

WORD ACCENT 1 3

will-o'-the-wisps, often changing the entire mean- ing of a sentence.1

Further, in Late and Vulgar Latin, even a short penult attracted the accent, as is abundantly proved by the evidence of the Romance lan- guages.2 (i) In a syllable not initial the second of two vowels in hiatus attracts the accent; thus the accentuation mulierem in Vulgar Latin is at- tested, not only by the Romance forms, Eng- mulcr, Old Fr. moulier, Prov. mother, Roum. muli- ere. Span, mujer, It. mogliera ; but by the precept of a late grammarian,3 "mulierem in antepaenul- timo nemo debet acuere, sed in paenultimo potius," and by the usage of Christian poets of the third and fourth centuries. (2) A mute followed by r at the beginning of the last syllable attracted the accent to the penult, the result, in all probability, of the practice among Latin poets of allowing a mute and a liquid to "make position."4 Lat. tent- brae is attested by Span, tinieblas ; colobra, by Fr. coulenvre, Span, ctilebra, etc. (3) In com- pound verbs the accent shifted to the stem-vowel of the verb. Lat. recipit is shown by It. riceve, Fr. recoit, Span, recebe ; demorat, by It. dimora, Old Fr. demtiere, Fr. demeure, Prov. demorat etc. (4) The evidence from the Romance numerals, it is true, seems to point in the opposite direction,

1 Kleczkowski, Cours de Chinois, Paris, 1876, i. p. 29 ss.

2 Meyer-Liibke, Gram. Rom. Sprach., Leipzig, 1890, i. p. 489 ss.

3 Anecdot. Helv. ciii. i But cf. Serv. ad Aen. i. 384.

14 THE STRESS ACCENT IN LATIN POETRY

namely to a Vulgar Latin viginti, triginta, quadrd- ginta, etc. Triginta, according to Consentius1 (fifth century), is one of the barbarisms, " quae in usu cotidie loquentium animadvertere possumus." But while, according to Meyer-Liibke, it is pos- sible to derive It. venti from viginti, and even veinte, treinta from viginti, triginta on the sup- position that the i was close (though not possibly quarante from quadraginta), it seems to me more likely that there was a still later change in Vulgar Latin, so that while the earlier Romance forms are derived from viginti, triginta, etc., the Italian are derived from shortened forms which were accented on the penult. There is some evidence for this view in late inscriptions, for example, on a fifth- century inscription 2 quarranta is written for quad- raginta (It. quardnto), and an epitaph in hexameters has vinti, for viginti (It. venti).3

The phenomena of syncope and vowel reduction, characteristic of all periods of the language, are the main support of the stress theory. These have been very ably treated by Lindsay in his chapter on Accentuation4 and need only be briefly sum- marized here.

A. Syncope (i) Pretonic : artena (Gr. apv- Taiva), perstromah (Gr. Trepia-rpcopia); enclitic or

1 p. 391 K. 2 A. L. L. v. 106.

3 Wilm. 569, cf. C. I. L. viii. 8573 : {Et menses septem diebus cum vinti duobus). 4 Op. cit. p. 148 et ss.

5 Lucil. (i. 41 M. and Lowe, Prodr. p. 347).

WORD ACCENT 1 5

subordinate words which drop final e before an initial consonant, e.g. nempe, proinde, deinde, atque, neque becoming *nemp (so scanned by Plautus and Terence), prom, dein, ae, nee; benficiiwi, malftcium, (calefacere, ealefacere, then) calfacere, olfacere, mins- tcrium or tnisterium ; aet for aevit in aetemus, aetatem, etc., then in aetas ; frigdarius1 beside frigidus, caldarms beside calidus, portorium beside portitor, postridie beside posteri, altrinsecus be- side altcri ; si audes (Plaut.) in the Class. Period, sodes. (2) Post-tonic : barca, lamna 2 (in Vul. Lat. lamia), lardum, iurgo (still inrigo in Plautus), usurpo for *tt,suripo ; nouns and adjectives in -atis, denoting the country of one's birth, as nostras, Arpinds, etc. ; u, i, in hiatus, larna, a trisyllable in Plautus, is later a dissyllable, so gratiis later gratis ; occasionally, ardus 3 for aridns, aspris for asperis ;4 soldus? possum for pote-sum of earlier writers. In Vulgar Latin wave after wave of syn- cope, as is shown by the Romance derivatives, changed the whole appearance of the language ; e.g. slave names like Marpor^ for Marcipor, etc. ; mattns 7 for madidus, virdis, fridam for frigidam

1 Lucil. (viii. 12 M.).

2 Hor. Od. ii. 2, 2 (inimice lamnae).

3 Plaut. Aid. 297; Pers. 266.

4 Verg. Aen. ii. 379. Cf. aspritudo, aspretiun, aspredo, It. aspro.

5 Lex Mzinicipalis of Julius Caesar (C I. Z. i. 206, 114, 115); Hor. Sat. ii. 5, 65 and i. 2, 113.

6 C. I. L. i. 1076.

7 Petron. § 41.

1 6 THE STRESS ACCENT IN LATIN POETRY

(on an inscription of Pompeii) ; 1 calda is read in Cato2 and the proper name Cald(iis) is found on coins as early as 109 B.C.;3 domnus for dominus, and also the proper names Domnus, Domna, Gr. Aofjuvos 4 ; so saeculum was restored to its orig- inal form saeclum, etc., veclus took the place of vetulus, anglus of angulus, stablum of stabulum, vaplo (Ms. baplo) of vapulo? etc.

B. Weakening of unaccented vowels : Under the Early Accent law, unaccented short vowels were changed to e, before a labial or / to 0 ; so the Mss. of Plautus preserve traces of subegit (from sub and ago, cf. Gr. awaya)) for snbigit? exsolatum for exulaturn? and the Lex Repetundarum of 121 B.C. has forms like detolerit, oppedeis side by side with detirierit, ediderit, etc. ; e is retained before r, e.g. peperit from pario, before a consonant group retnex, but remigis, and after z, ebrietas, parietem, etc., and 6 when not before a labial is retained, eg. invoco, advoco, and even before a labial when i precedes, filiolus. In the final syllable it is invariably re- duced, vicus, older vlcos (Gr. ol/cos), filios on the First Scipio-Inscription, etc. Even diphthongs

1 C. I. z. iv. 1291.

2 ^. R. vi. 1 and 75, also Varro, R. R. i. 13, etc.

3 C. I. I. i. 328.

4 C. I. G. i. 6505, end of second century A.D.

5 Prodi App. 197, 20-22 K.

6 Capt. 814.

7 Merc. 593 (B); Most. 597 (,4), etc.

WORD ACCENT 1 7

were changed, their first element being affected) ai becoming l (through *ei), au becoming u (through *eu). In final syllables, ei, Class. iy repre- sents Ind.-Eur. ai, e.g. tetudl (older ei), just as oi was weakened to ei then to I, foideratei (S. C. Bacch.), Class, foederatl. In the late Republican and Imperial times, possibly on account of the grammatical studies imported from Greece, com- pounds were often restored to their unweakened form (' Recomposition '), and at all periods of the language the analogy of similar forms operated now in favour of and now against vowel re- duction.

C. Shortening of unaccented vowels : By the law of Brevis Brevians the final syllable of a dis- syllabic word was shortened if the preceding syl- lable was short, so in Plautus viodo, ago, Jiabcs, amor, cub at, and even potest. Later the shortening was applied to Cretic words. Horace, for example, admits Pollio, mentio, dixero, and the fourth century grammarians speak of the final o of nouns (nom. sing.), verbs (ist pers. sing. pres. ind.), adverbs, and conjunctions, as universally shortened in the pro- nunciation of their time, except in monosyllables and foreign words.1 So also final syllables ending in -m, -r, -t, -s, and even, occasionally, those long by position are shortened, a result due partly to the inherent weakness of every final syllable, partly, in

1 Charis. p. 16, 5 K ; Diomed. p. 435, 22 K ; Prob. de ult. syll. p. 220, 15 K ; Mar. Vict. p. 28, 23 K ; Priscian, i. p. 409, 16 H. c

1 8 THE STRESS A CCENT IN LA TIN FOE TR Y

the case of words ending in a vowel, to the practice of shortening a long vowel in hiatus.

The view that the Latin was essentially a stress accent is supported by the united testimony of the Romance languages. French must, at one time, have had a very strong stress accent, as is shown by the preponderance of "heavy" syllables (soup- con, maison, amour, planter, attention, commande- ment), and in parts of France stress differences are still strongly marked, as in the north, in Piedmont, and in French Switzerland. Meyer-Liibke sug- gests1 that at a certain epoch a musical element was added to the stress accent, so that (represent- ing the musical accent by A ) a word like soupgon, for instance, would show the series supson, supson, stipson. While the French makes more account of differences in pitch than any other of the Romance languages, the musical element is noticeable in Spanish and Italian, though the whole develop- ment of these languages shows the influence of stress. The same is true a fortiori of Roumanian, in which the syncope of Latin words is carried to a very great extent ; for example, dmeng for Lat. domimca, Sunday, and cal for Lat. caballus, a horse, etc.

The one stumbling-block is the adverse testi- mony of Latin writers on accent, from Varro, with a few exceptions, down to Priscian. Of greatest importance are Varro, Cicero, and Q'uintilian, for

1 Op. cit. p. 500.

WORD ACCENT 1 9

the later grammarians, as a rule, continue to re- peat mechanically the formulas of their predeces- sors, even down to a period when, as is universally acknowledged, stress must have been the dominant principle. This weakens the whole mass of evi- dence from the grammarians. As M. Vendryes rather neatly illustrates, they are like the French schoolmasters who are still teaching the difference between aspirate h and mute //, though the two are precisely alike and have been for more than a century.

All that is said by Varro, Cicero, and Ouintilian, on the subject of accent, interpreting the words in their most obvious sense, refers to differences in pitch and quantity. But, on the other hand, they are frankly applying the terms learned from their Greek teachers to the nearest equivalents in Latin, just as the names of the Greek gods were fitted, more or less aptly, to the already existing Roman deities. That, in the matter of accent, the new terminology was faulty, is shown by the confusion in regard to the circumflex accent among Roman writers ; Vitruvius placing it on monosyllables like sol, 1/tXyJlos, vox ; Quintilian on the penult of tri- syllabic words whose vowel was long by nature, Cethegus, but Camillas ; Priscian, Martianus Ca- pella, and other late grammarians, placing it on the penult of Rojtia, for example, but not of Romac, where the ultima was long. It is worth while to note in passing, that this last theory, taken over

20 THE STRESS ACCENT IN LATIN POETRY

bodily from the Greek, is not found in Servius or Pompeius, who recognize the stress character of the Latin accent.1 The Latin writers do, indeed, speak of the more obvious difference between the Greek accent and their own. Quintilian, for ex- ample,2 complains of the monotony produced by the accent falling always on one of two syllables. " Itaque tanto est sermo Graecus Latino iucundior, ut nostri poetae, quotiens dulce carmen voluerunt, illorum id nominibus exornent." But the subtler distinction between pitch and stress — a distinction which has only recently been mastered by phoneti- cians, as Hendrickson points out in his, to my mind, conclusive reply3 to Bennett's "What was Ictus in Latin Prosody?" — they may well have missed. Especially does this seem to be the case when we reflect that the language of the cultivated classes, in the Classical Period at least, shows far less tendency to syncope than the popular speech. It is not claimed that the Latin accent was so heavily stressed as English or German, for instance, but, just as in French the phenomena of syncope and vowel reduction abundantly prove the stress character of the accent (although a musical accent seems to have been added later), so in Latin, the same phenomena prove that the essential differ- ence between the accented and unaccented syl- lables of a word was a stress and not a pitch

1 Vendryes, op. cit. p. 31. 2 Instit. Or at. xii. 10, 33.

3 A. J. P. vol. xx. p. 207.

WORD ACCENT ' 21

difference. This stress difference may have been almost unnoticeable under ordinary circumstances, when one was speaking remissione et moderatione vocis} — even in English, in quiet conversation the voice rises and falls as much at least as it strength- ens and weakens, — but when the voice was raised for any reason,2 it did become apparent, as it un- questionably does in French to-day. It would, frankly, be impossible to imagine a pitch accent entirely without differences of stress, or a stress accent unaccompanied by a rise and fall of the voice, because in the effort to produce a higher tone we unconsciously use more energy, and vice versa. Now if, as we have seen reason to believe, Greek in the time of Varro and Cicero had already begun to show traces of the change that has made modern Greek a stressed language, the difference between a pitch accent with a growing tendency toward stress, and a stress accent accompanied — as in the Romance languages — by a musical tone, may, not unreasonably, have escaped the notice of men eager only to find resemblances.

In conclusion, I quote three of the later gram- marians, because, while their contemporaries and successors were still repeating by rote the worn-out precepts borrowed from the Greek, their remarks show a quite modern spirit of experimentation.

1 Cic. Brut. xci. 314.

2 Cf. Servius's suggestion for determining the accented syllable of a word, quoted below.

22 THE STRESS ACCENT IN LATIN POETRY

Servius 1 (fourth century) : " Accentus in ea syllaba est, quae plus sonat. Quam rem deprehendimus, si fingimus nos (ad) aliquem longe positum cla- mare. Invenimus enim naturali ratione illam syllabam plus sonare, quae retinet accentum atque usque eodem nisum vocis ascendere." Pompeius 2 (fifth century) takes Servius's hint and enlarges upon it. " Et quo modo invenimus ipsum accen- tum ? et hoc traditum est. Sunt plerique qui naturaliter non habent acutas aures ad capiendos hos accentus et inducitur hac arte. Finge tibi quasi vocem clamantis ad longe aliquem positum. Ut puta finge tibi aliquem illo loco contra stare et clama ad ipsum ; cum coeperis clamare, naturalis ratio exigit, ut unam syllabam plus dicas a reliquis illius verbi ; et quam videris plus sonare a ceteris, ipsa habet accentum. Ut puta si dicas orator, quae plus sonat ? -ra ipsa habet accentum. optimus, quae plus sonat? ilia quae prior est. Numquid sic sonat -ti et -tnus, quemadmodum op f Ergo necesse est, ut ilia syllaba habeat accentum, quae plus sonat a reliquis, quando clamorem fingimus." In an- other place Pomponius writes : 3 " Et quid est ipse accentus ? ita definitus est : accentus est quasi anima vocis, id est accentus est anima verborum et anima vocis uniuscuiusque. Quemadmodum cor- pus nostrum non potest esse sine anima, sic nee

1 Comment, in Donat. p. 426, 16 K.

2 p. 127, 1 K. 8 p. 126, 27 K.

WORD ACCENT 23

ullum verbum nee ullus sermo sine accentu potest esse. Et quemadmodum anima nostra in toto corpore ipsa plus potest, sic etiam ilia syllaba plus sonat in toto verbo, quae accentum habet. Ergo ilia syllaba, quae accentum habet plus sonat, qua- si ipsa habet maiorum potestatem." Diomedes : (fourth century) writes: " Accentus est acutus vel gravis vel inflexa elatio orationis vocisve, intentio vel inclinatio acuto aut inflexo sono regens verba. Nam ut nulla vox sine vocali est, ita sine accentu nulla est; et est accentus, ut quidam recte puta- verunt, velut anima vocis." This remark, it seems to me, shows very careful observation. Looked at from one point of view the accent was elatio, from another, it was intentio. To see that it was really both, would have been too much to expect so long before the days of Experimental Psychology.

1 1. 430, 29 K.

II

NUMERI ITALICI ET SATURNII

The first utterances of every people are in verse, not verse in the sense of a definite arrangement of syllables that inevitably strikes the ear as different from the prose arrangement, but words forced into a rude kind of rhythm by being chanted again and again in worship of some god or over the daily tasks that are shared in common.1 Now whatever view may be held of the nature of ictus in quanti- tative poetry, there can be no two opinions of the nature of the beat in music. In the most primitive and the most sophisticated music alike, the down beat is the stressed beat — the placing of the foot on the first syllable of the measure. We are all perfectly familiar with the transformation of prose into rhythm by being chanted : —

Our Father which art in Heaven Hallowed be thy name,

or Du fond de l'abime je crie vers toi

O, mon Dieu.

1 Bowditch, Mission to Ashantee ; Westphal, Einleitung, Allge- meine Metrik ; du Meril, Introd. Poesies Populaires latines au doiizihne siecle, Paris, 1843.

24

NUMERI I TALI CI ET SATURNII 2$

Moreover, if, after centuries of quantitative poetry, the Christian hymns became purely accentual by being chanted in the service of the Church, — the singing was at first congregational and only grad- ually restricted to the priests, — much more would the first primitive chants base their rhythm on the accent of the words. As we have seen, this accent, in the Italic dialects as well as Latin, was one of stress, nor is it thinkable that the stress of the chant and the natural stress of the words should not coincide. This stress was helped out by allit- eration of the accented syllable, and by the endless repetition both of final syllables and of entire words. So the chant to Mars on the Iguvine Tablets is rhythmical : —

Serfe Martie Prestota Cerfier | Cerfier Martier Tursa Cerfier | (Jerfer Martier Totam Tarsinatem | trifom Tarsinatem Tuscer Naharcer | Jabuscer nomner )^ nerf cihitu | ancihitu

jovie hostata | anhostatu tursitu tremitu | sonitu savitu ninctu nepitu | hondu holtu preplohatu | previclatu.

Very similar is the Old Latin chant to Mars, quoted by Cato : 2 —

1 De Re Rustic a, 141.

26 THE STRESS ACCENT IN LATIN POETRY

/ r r /

Mars pater te precor

quaesoque uti sies | volens propitius

mihi domo | familiaeque nostrae.

quoius rei ergo

agrum terram | fundumque meum

suovetaurilibus [ circumagi iussi,

uti tu morbos | visos invisosque

viduertatem | vastitudinemque

calamitates | intemperiasque

proibessis, defendas averuncesque ;

/ f f r \ /

ut fruges frumenta | vineta virgultaque grandire dueneque | ev\nire siris, \L pastores pecuaque | salva servassis duisque dubnam salutem | valetudinemque mihi, domo | familiaeque nostrae : j harumce rerum ergo L j fundi serrae | agrique mei

lustrandi lustrique | faciendi ergo, sic uti dixi :

(Mars pater) macte hisce lactentibus suovetaurilibus | immolandis esto.

Alliteration, as Westphal has pointed out,1 is not the underlying principle of the verse, though it is of frequent occurrence. Repetition, indeed, both of sounds and of entire words, is the invariable characteristic of a poetry based on stress. Still frequent in the verses of Plautus and Terence, there is a visible falling off both of alliteration and

1 Op. cit. p. 220.

NUMERI ITALIC/ ET SATURNII 27

of the various forms of Reimart, during the Classi- cal Period, when stress was subordinated to quan- tity. Yet even here there is a difference. In the smooth hexameters of Ovid, which show sixty-five per cent of accords between quantity and word accent, repetition both of words and sounds is especially frequent, as it is in the more familiar Eclogues of Vergil, — the eighth, for example. In the popular and semi-popular poetry of the first three centuries of the Christian era, when quantity and word accent tend more and more to coincide, assonance, repetition of words and phrases, and even rhyme are increasingly frequent, until in the Christian hymns, stress and rhyme are the two almost equally important principles of the verse.

For the remaining fragments of Latin verse, prior to the Saturnians, I shall content myself with those the text of which is reasonably complete. It would be idle to quote the Carmen Sa/zare, for example, which was unintelligible to the Romans themselves in the time of Horace, and which has been emended by Baehrens 2 and by Zander,2 to give only two authors, in the most widely different manner.

Carmen Fratrum Arvalizim2, (inscribed on a marble tablet, discovered in 1778 and now in the Vatican) : —

1 Poetae Latini Minores, vol. vi. p. 29.

2 Versus Italici Antiqui, Lund, 1 890, p. 29.

3 Zander, op. cit. p. 25.

28 THE STRESS ACCENT IN LATIN POETRY

Neve luem ruem Marmar, .

1 three times

Enos Lases iuvate. — three times

sms mcurrere in pleores. J Satur fu fere Ma(vo)rs U limen sali sta berber

\ three times

\ three times

Semunis alternei

\ r *

advocapit cunctos

r r /

Enos Marmor iuvato — three times

' ' '

Triumpe, triumpe, triumpe

triumpe, triumpe (triumpe).

The prayer to Jupiter Dapalis, quoted by Cato:1 —

Jupiter Dapalis

f / r

quod tibi fieri

oportet in-domo

familia mea

calignam vini dapi

eius rei ergo

macte illace dape

\ / / pollucenda esto.

The Drinking-song from Varro.2 Zander unnec- essarily changes the order of the words.

r f r /

Novum vetus vinum bibo,

• / r /

Novo veteri morbo medeor.

A charm against foot-ache, quoted by Varro.3 The person using this charm was to sing it over

1 De R. R. c. 132. 2 De L. L. vi. 21. s De R. R. i. 12, 27.

NUMERI I TALI CI ET SATURNII 29

twenty-seven times, to touch the ground, and to spit.

Terra pestem teneto

Salus hie maneto.

A charm against sprains, quoted by Cato:1 —

/ / /

Huat, hanat, huat

/ r *

ista, pista, sista.

/ r *

dannabo danna ustra.

A charm against tumours and inflammations quoted by Pliny.2 The person was to say it over three times and spit on the ground three times.

Reseda, morbis, reseda scin, scin quis hic-pullus egerit radices nee caput nee pedes habeant.

An old saw quoted by Festus, p. 93 : —

Hiberno pulvere verno luto Grandia farra, camille metes.

The words of the goal-post, which marks the end of the race, to the defeated runner, quoted by Porphyrio on Horace:3 —

Quisquis ad me novissimus venerit, habeat, scabiem.

Lucien M tiller rewrites, Habeat scabiem quisquis

1 De R. R. c. 160. 2 Hist. Nat. xxvii. 131.

3 Ars Poetica, 1. 417.

30 THE STRESS ACCENT IN LATIN POETRY

ad me venerit novissimus, destroying both the rhythm and the spirit, for the three dactylic beats at the end represent the last desperate sprint of the runner and the sneer of the goal-post at his lack of success.

All the foregoing quotations, with the possible exception of the last two, are in the nature of chants, repeated over and over again, as children repeat in their play. There are three or four measures in the musical phrase, the down beat falling on the primary or secondary accent of the word. This accented syllable is usually (though not always) a long syllable, for quantity is an inherent principle in Latin derived from the Indo- European parent speech. Further, the movement of the voice is from the accented to the unaccented syllable — the most natural cadence in Latin — with an occasional anacrusis, common to both music and poetry, and made perfectly familiar to us by its use in the Christian hymns. It is inter- esting to note that the phrase consisting of three measures — by far the more usual — Enos, Lases iuvate I J | J J~l I J J II or lupiter Dapalis J J | J J | J J || for example, shows the type of the first and second half-verses in the Saturnians where the strong caesura in the middle of the line points to a composite nature.

Closely analogous to the primitive chant are the Sentential or maxims of everyday life.1 They, too,

1 Zander, op. cit. pp. 1-19.

NUMERI I TALI CI ET SAT URN II 3 1

have the unmistakable ear-marks of a popular origin ; namely, a strong stress rhythm, alliteration, and the frequent repetition of words; since, by reason of these three elements, such maxims are easy to remember, give greater pleasure in repeat- ing, and seem to carry more authority. Our own proverbs are precisely similar in nature : —

Many men, many minds. Nothing succeeds like success. Money makes the mare go. Be sure your sins will find you out.

Latin writers are fond of quoting these bits of popular wisdom. Cicero has a large number, gen- erally accompanied by some such expression as in proverbii consuetudinem venit, or tit est in vetere proverbio : —

Quot homines tot sententiae.1 Largitio fundum non habet.2 Fortes fortuna adiuvant.3 Summum ius summa iniuria. Minima de malis.5

The talk of Petronius' petits bourgeois is full of proverbs, especially is this the case in the Cena

TrimalcJiionis : —

1 De Fin. i. 15. 2 De Off. ii. 55.

3 Tusc. ii. 1 1 . Cf. the similar Di facientes ddiuvant, Varro, R. R. i. 1, 4.

4 De Off. i. 33. 5 De Off. iii. 102, 105.

32 THE STRESS ACCENT IN LATIN POETRY

/ * r f

Post asellum diaria non sumo 1

Sociorum olle male fervet.2

Qui asinum non potest, stratum caedit.3

Colubra restem non parit.4

Quod hodie non est eras erit.

In-alio peduclum in-te ricinum non vides.

Semper in hac-re, qui vincitur vincit.7

Assem habeas, assem valeas.8

Varro, Pliny, Gellius, the Grammarians, contain many more. Sometimes a proverb is quoted by different writers with a slight change of form, or with the verb omitted, as often in Cicero, or with only the characteristic words quoted. For ex- ample, Nonius has LSnge fugit qui suos fugit? and Petronius, LSnge fugit quisquis suos fiigit.10 This latter I agree with Zander in considering a corrupt form. Multis eget qui miilta hdbet11 in Gellius, while Seneca expresses the same idea, Qui miiltum hdbet phis ciipit.^2, Non semper Saturnalia erunt}z in Seneca; Semper Saturna- lia1^ (agunt) and Dii pedes landtos habent}^ in Petronius ; while Macrobius 16 writes, " atque inde prov erbium ducttim deos laneos pedes habere" and Porphyrio on Horace's words deseruit pede Poena

1 Petron. 24. 7 Petron. 59. 13 Sen. Apocol. 12.

2 Petron. 38. 8 Petron. 77. 14 Petron. 44.

3 Petron. 45. 9 Nonius 204. 22. 15 Petron. 44.

4 Petron. 45. 10 Petron. 43. 16 Mac. i. 8, 5.

5 Petron. 45. u Gell. ix. 8. I.

6 Petron. 57. 12 Sen. Ep. 119. 6.

NUMERI I TALI CI ET SATURNII 33

claudo} "hoc proximum est Mi quod dicitur deos iratos pedes lanatos habere."

Far more important than the Numeri Italici just considered, are the Numeri Saturnii, over which a long and bitter struggle has raged between those who advocate a quantitative and those who advo- cate an accentual basis of versification. In favour of the Quantitative Theory it may be urged that it is the view of all the Latin writers who treat of the measure from Caesius Bassus down, and against it, that it requires the arbitrary lengthening of a very large number of naturally short syllables. The Accentual Theory is in harmony with all we know of popular Latin verse, but, on the other hand, it requires a secondary accent on words of four syllables, like Cornelius, for example, and, unless we accept Thurneysen's (and Lindsay's) theory of but two accents in the second half -verse, a binary accent on words of three syllables accented on the antepenult, as maximas. Zander2 lessens the number of syllables arbitrarily lengthened, by suggesting an alternation of rhythm between the first and second half -verses ; the first, though regu- larly iambic, may become trochaic, and the second may become iambic. This view he supports by ancient verses, like —

/ f f u /

Hiberno pulvere luto verno

/ f r r

Grandia farra, camille, metes.

1 Porph. ad Od. iii. 2, 32. 2 Op. cit. p. ii.

D

34 THE STRESS ACCENT IN LATIN POETRY

But such an alternation of rhythm seems utterly out of place in an unaffected primitive verse. As du Meril well says, " La nature des langues exerce done une influence preponderante sur la premiere forme du vers ; on utilise les elements d'harmonie qu'elles possedent, sans songer a augmenter les difncultes de sa tache par des innovations sans raison et sans but. Dans presque toutes, la desi- nence des mots n'a qu'une valeur grammaticale ou meme purement euphonique, la syllable radicale, celle dont l'accentuation domine la prononciation des autres, est la premiere, et le mouvement natu- rel de la voix va du temps fort au temps faible."1 The difficulty of Zander's theory is increased by the fact that the alternation of rhythm may take place not only between the two halves of a line, as in, —

Grandia farra, camille metes, but between successive feet, —

Hiberno pulvere luto verno. To Zander's bibliography 2 need only be added the quantitative treatment of Klotz3 and of Rei- chardt4 (scarcely more than a restatement); and the accentual treatment of Westphal in his chap- ter " Die accentuirenden Verse der alten Italiker," 5

1 Op. cit. p. 50 et s.

2 Op. cit. p. xix et ss.

8 Grundzilge altromischer Metrik, Leipzig, 1890, p. 97 ss. * Jahrbucher fur Klassische Philologie (Suppl.), xix. p. 207. 5 Allgemeine Metrik, Berlin, 1892. p. 220 ss.

NUMERI I TALI CI ET SATURNII 35

of Lindsay,1 who follows Thurneysen 2 in making the verse one of five accents, and of Gleditsch,3 who makes it one of eight, like the Old German Lang ze He.

Caesius Bassus,4 the most ancient authority on the Saturnian metre, makes it a purely quantita- tive verse. His scheme for the first half is iam- bic, w _ w _ v_/ _ f and for the second trochaic, — w _ \j _ w, though he acknowledges that many of the verses are either too long or too short to fit the scheme. " Nostri autem antiqui, ut vere dicam quod apparet, usi sunt eo {i.e. Saturnio versu) non observata lege nee uno genere custodito, (ut) inter se consentiant versus, sed praeterquam quod duris- simos fecerunt, etiam alios breviores, alios longiores inseruerunt, ut vix invenerim apud Naevius, quos pro exemplo ponerem." As a matter of fact, none of the extant Saturnians fits Caesius Bassus's scheme perfectly, the quantities of the " model " verse, Dabunt malum Metelli \ Nacvio poetae are ^ — ^ —

^ |_w_^ , so that it would almost seem

to be the one thing which the Saturnians are not.

The arguments against the Quantitative Theory, stated briefly, are as follows : —

A. The clash between word accent and quantity

1 A. J. P. vol. xiv. pp. 139 ss. 303 ss.

2 Der Saturnier, Halle, 1885.

3 Rhetorick und Metrik der Griechen und Rower, in I. Miil- ler's Handbtich, 2 Bd. p. 820 ss. Miinchen, 1901.

4 Keil, vi. 1. p. 265 et s.

36 THE STRESS ACCENT IN LATIN POETRY

is exceedingly harsh in a majority of the lines; for instance, —

f r *

Subigit omne Loucanam

/ / /

Runcus atque Porpureus

Ne quaeratis honore,

so harsh and so frequent indeed, as to make it thoroughly unnatural in a popular verse. Espe- cially does this seem to be the case when the Saturnians are compared in this respect, with the earlier popular verses, with the popular poetry of the Classical Period itself, like the Mille-song of Aurelian's legions, in which there is little or no clash between word accent and quantity, with the semi-popular poetry of the early centuries of the Christian era, and with the Christian hymns.

B. Aside from final syllables, it must be con- fessed that the number of short syllables arbitra- rily lengthened is not great. Lucius ; however, with long i is contradicted, not only by every ex- ample of the word in early Latin poetry, but by the evidence of Oscan Luvkis (nom. sing, of stem Loucio)]1 and the argument for long i in early Latin from modern Italian Lucio has even less weight than for long e in the penult of mulierem, for the same period. There is, moreover, no authority for long u in puer, parisuma, nor long i in viro (Ind.-Eur. *wzro, but vir in Latin). The treatment of this word (viro, 1. 2 in the first

1 Lindsay, op. cit. p. 158.

NUMERI I TALI CI ET SATURNII 37

Scipio Inscription virum, 1. 32 in the epitaph of Atilius Calatinus) by the adherents of the Quan- titative Theory is interesting. Weil and Benloew 2

scan —

.» / / / /

Bonorum optimum fujisse virum

Populi primarium fu.isse virum

with the remark, " Virum a la place d'un trochee est tres-choquant, nous l'avouons ; mais les liquides se redoublent facilement, surtout apres une voyelle aigue : l'auteur aura fait violence a la langue en pronongant virrotn. C'est done la un effet d'accent, mais un effet tout exceptionnel. . . . N'oublions pas que nous avons affaire a une versification nais- sante, qui tantot force la prononciation au profit du vers, tantot sacrifice le mouvement du vers aux obstacles qu'y oppose une langue encore rude et peu fa^onnee au tour poetique." Bartsch 2 (with others) adds the genitive plural. He reads : —

Duonoro optimo fujise viro (viroro).

Havet,3 whose exhaustive treatise leaves no line, or fragment of a line, unconsidered, reads : —

Duo[noro | opti|mo ( ) fu ise vi|ro ( ) |

Popu li pri|mari|um ( ) fuijse vir]um ( ).

1 Theorie Generate de V Accentuation Latine, Paris and Berlin, 1855, p. 97-

2 Der Saturnische Vers und Die Altdeutsche Langzeile, Leipzig, 1867.

3 De Saturnio Latinorum Versu, Paris, 1880, p. 223.

38 THE STRESS ACCENT IN LATIN POETRY

These he calls Saturnia disticha and thinks they may have been of the nature of formulae. In the two lines of Naevius, however, where the word occurs in the same position (11. 58, 84), he ex- pressly states that the i is long, adding, " Sane mirum est vlri latine correptum esse. Sed simili modo perierare pro periurare habemus, quod adhuc explicatione caret; neque magis scimus cur dica- tur humanus et homo, publicus et populo" 1

Klotz,2 who allows but four feet to the measure, reads : —

' ' ' '

Bonorum optimum | fuise virum.

Zander3 has recourse to his theory of alternation of rhythm in all four lines, and reads : —

Duonoro optimo | fuise viro Populi primarium | fuise virum, etc.

Reichardt 4 follows Zander's marking, but sug- gests that the suppression of the last thesis was a liberty of which the writers of the Saturnians, on occasion, availed themselves, not only in epitaphs, fui\se vi\ro ( ) | where Havet finds it, but in Epic poetry as well.

C. The strongest argument, however, is the very large number of final syllables arbitrarily lengthened under the arsis (Klotz does not fail to see that this strengthens the argument for the

1 De Saturnio Latinorum Versu, p. 85. 2 Op. cit. p. 99.

3 Op. cit. pp. 60 and 58. 4 Op. cit. p. 224.

NUMERI ITALICI ET SATURNII 39

accentual nature of the verse). The syllables thus lengthened include, not only those, like the a of the nom. sing, in the 1st. decl. and of the nom. pi. neut, which though originally long, had been shortened before the time of the oldest Latin poetry,1 but also those that were never long at any period of the language, like -bits 2 in the dat. and abl. pi. and -que, atque, etc. I quote Havet's3 classes of lengthened final syllables : —

" i. Nominativus primae declinationis, ut terra, mea, sancta, tua, forma, fama, vita, divina, hasta, ea, cura, parisuma, ferocia, filia, Proserpina ;

ii. Nominativus secundae declinationis, ut Runcus, in- feros, inclitus, Putins, fortasse, faber ;

iii. Vocativus, ut summe, Laertie ;

iv. Neutrum plurali numero, ut exta, patria, occisa ;

v. Nominativus tertiae declinationis, ut mare, acer ;

vi. Genitivus, ut regis ;

vii. Dativus vel ablativus plurali numero, ut Te?npesta- tebus, piscibus, capitibus.

viii. Neutrum plurali numero, ut omnia, pec tor a, atrocia, sagmina ;

ix. Numerale, ut fortasse, quinque ;

x. Verbum, ut obliviscere, insece ; subigit,facit; quaira- tis ; cante ; pellere, fuisse ;

xi. Adverbium, ut facile ; comiter ; hice ; quamde, deinde ; semul ; -que, atque, itaque ; fortasse cume."

1 Lindsay, Lat. Lang. p. 371.

2 Lindsay, Lat. Lang. p. 403, from an original -bhos.

3 Havet, op. cit. p. 57 ss.

40 THE STRESS ACCENT IN LATIN POETRY

The ablative singular of the third declension he considers a doubtful case of lengthening for two reasons : (i) because of the two endings -e and -I, corresponding to the two endings of the accusative -em and -im and the two endings of the genitive plural -urn and turn; and (2) because instead of -e, -ed might have been written at this period, for ex- ample, patred, loved, ordined, rumored, pulvered. But there are only a few instances of I in the ablative singular of consonant stems,1 and the ex- tension of the ablatival d, especially to such a word as love, is very doubtful. It is persistently written in the S. C. de Bacchanalibus of 186 B.C., but as persistently omitted in a nearly contempo- rary inscription,2 nor is there any trace of d in the ablative of nouns in Plautus and the earliest Dramatic literature.3 It may be remarked in pass- ing, that inasmuch as the ablative suffix in d ap- pears to be confined in Ind.-Eur to O-stems, the same argument that caused Havet to lengthen Latin vlr (from Ind.-Eur. *wlro) should have pre- vented his extension of the ablatival d beyond O-stems in Latin.

Thurneysen,4 and Lindsay in his two suggestive articles,5 allow but two accents to the second half- verse. Against this, the following considerations may be urged : —

1 Neue, i.2 p. 212 et ss. 2 C. I. L. ii. 5041, Spain, of 189 B.C. 3 Lindsay, op. cit. p. 391. 4 Op. cit. p. 13 ss. 5 Op. cit. p. 303 ss.

NUMERI I TALI CI ET SATURNII 4 1

A. It is, to say the least, very difficult to find a primitive verse with just five accents. Bartsch postulates an original common Epic verse for all Indo-Germanic poetry, consisting of eight feet, with a caesura after the fourth foot. From this root form he derives the Greek hexameter and the Saturnians, as well as the Indian sloka and the Old German Langzeile} A verse with eight accents (the trochaic septenarian) is the favourite metre of the soldiers' songs in the time of the Caesars, and recurs in the Spanish Epic ; a verse of eight feet is much used in the Byzantine poetry (though that of six is also common) and in the poems of the Troubadours, and, divided into two verses of four feet each, with end rhyme, such verse is familiar to us from the Christian hymns. So verses of four, six, or eight feet, seem to be the primitive, spon- taneous form, while those of five — English blank verse, for example — are artificial and modern.

B. If there were but two accents in the second half-verse, we ought to find Saturnians in which the second half is made up of four (or even three) syllables, for they, according to the rest of the scheme, could bear two accents, and this, as a mat- ter of fact, we do not find.

C. It certainly is "strange," to use Lindsay's own word, that consentiunt, Calypsonem, Aleriaque, etc., in the second half-verse should receive but one accent, while words of four syllables in the first

1 Cf. Westphal, op. cit. p. 56 et s.

42 THE STRESS ACCENT IN LATIN POETRY

half-verse always receive two accents, and even aetate (1. 21), with three syllables, receives two — Aetate quam parva -1

D. A half-verse like gloria atque ingenium in the third Scipio Inscription, since it occurs in the last half of the line, is allowed to have but two accents, although it consists of seven syllables, and if it stood before the caesura would un- doubtedly receive three. Why should the fact that it stands after the caesura deprive the syl- lables of their full value ? The same may be asked with regard to

Hone vino ploirume consentiont Ripmai)

(the inscription of Atilius Calatinus twice quoted by Cicero and ending consentiunt gentesy makes the two-syllabled Romai a more probable conjec- ture than Romanai) and also

Hie cepit Corsica Aleriaque urbe.

In both lines,2 the number of syllables before and after the caesura is the same, but the six syllables before the caesura receive three accents, while the six following the caesura are put off with two.

E. The second half of the line is the more im- portant, because upon it the attention rests during the moment of adjustment before the next line is begun ; but this accentuation makes the caesura a precipice over which the verse rushes, to end with

1 Lindsay, op. cit. p. 314. 2 First Scipio Inscription.

NUMERI ITALICI ET SATURNII 43

an ignominious splash on the rocks below. In the second half-verse fortis vir sapiensque} Thurney- sen2 contents himself with marking the syllable -ens, in sapiens que ; Lindsay goes further and marks the two accents fortis vir sapiensque, thus put- ting five syllables under one metrical stress {fortis vir sapi-). This seems like a theory for the theory's sake, inasmuch as the poetry has wholly disappeared.

F. Caesius Bassus, Marius Victorinus, Terenti- anus Maurus, and others3 — " Unde apud omnes grammaticos super hoc adhuc non parva lis est " — agree in making the Saturnian a verse of six feet, especially are they sure about the three trochees in the last half. Now the later quantita- tive treatment of Latin verse would undoubtedly influence their view of the character and disposi- tion of the syllables in the feet; but the strik- ing, the fundamental, part of a verse, the part which no metrician could miss, is the number of feet. Besides, the tendency of the later Satur- nians is to become longer,4 which makes Thurney- sen's suggestion 5 at least a possible one, that when the Saturnian disappeared from literature, Sic horridus ille Defluxit numerus Saturnius,6

1 Second Scipio Inscription. 2 Op. cit. p. 13.

3 They are quoted in full by Havet, op. cit. pp. 310-327.

4 Cf. the quantitative Saturnians of Terentianus Maurus.

5 Op. cit. p. 56. 6 Hor. Ep. ii. 1, 158, 9.

44 THE STRESS ACCENT IN LATIN POETRY

it continued to exist among the common people and gradually went over into the trochaic septe- narian, the poor man's poetry, a few examples of which are preserved by Suetonius and others. But, a thing which he does not appear to see, the number of feet in each half-verse is still equal (four instead of three), and the second half-verse, while it has four accents, is catalectic, pointing to the smaller number of syllables, not feet, charac- teristic of the second half-verse of the Saturnians. Keller,1 whose discussion is, on the whole, the most satisfactory that I have seen, divides the ex- tant verses into " strong " or " classical " Satur- nians, which, incidentally, fall in with his scheme, and "older" or "cruder," which do not. But there is not a shred of evidence for such a divi- sion. Why should we suppose that the floruit of the Saturnian metre was reached in the time of Naevius ? Is it not at least as probable that the two oldest Scipio Inscriptions represent the purer native tradition, and that the increased number of unaccented syllables in the third Scipio Inscription shows a more pronounced borrowing from the Greek ? Keller enumerates sixteen rules for the " strong " Saturnians, certain of which2 apply also to the others. The points he really holds to are

1 Otto Keller, Der Saturnische Vers als rhythmisch erwiesen, Prague, 1883 ; Keller, Der Saturnische Vers, 2 Abhandl. Prague, 1886.

2 Namely, Rules 1, 3, 4, 6b, 9, 10, na, 12, 14, 15, 16.

NUMERI ITALICI ET SATURNII 45

three;1 namely, the strong pause in the middle of the line, the impossibility of two accented syllables following each other, and the equal impossibility of either half-verse ending in an accented syllable.

Now the strong pause in the middle of the line, naturally at the end of a word — for Ritschl has not been followed by the more modern editors in such readings as

Hone vino ploirume con . . . sentiont R(oma?iai) —

is the least disputed characteristic of the Satur- nians. It is, moreover, of the highest importance, bringing them into harmony both with the earlier Numeri Italici by pointing to a composite nature, and, through the Law of the Last Half, with the hexameter. There is also practical unanimity among the adherents of the Accentual Theory, in regard to the accent falling on the penultimate syllable of each half-verse. For toward the end, Thurneysen seems half inclined to yield his ac- centuation of apud-vos, remarking in a foot-note,2 " Auch diese Falle schwinden, wenn man mit Keller apud-zws, inter-se betont. Dann ist der letzte Vers accent ebenso fest wie der erste." But Kel- ler's rule that two accented syllables may not fol- low each other not only reduces him to the necessity of declassing the oldest and best-authenticated Saturnians, but it is contradicted by the general

1 p. 39. 2 Op. cit. p. 49.

46 THE STRESS ACCENT IN LATIN POETRY

usage of primitive poetry. To mention only a few- instances, the prayer to Mars, above cited, —

Mars pater, te precor ;

the old German of Otfrid's Evangelienharmonie} —

habt er in war min,2 \ / / \ ist sedat sinaz,3 \ / / \ iir kind ellu ; 4

the old English poem of Beowulf,5 —

falcom gefraege lange hvvile

r r \ r

ne leof ne lad"; the Cuckoo-song of the French peasantry, —

Jeunes gens qui etes a marier

Oh ! ne vous mariez pas dans le moi de Mai ;

J'ai vu le coucou ! ! ! Me, Me,

J'a vu le coucou ! ! ! Me, Me ;

and the familiar child's rhyme, —

Rain, Rain, go away ! Come again another day !

This is the well-known theory of a "supressed thesis." Otfried Mueller was the first to suggest

1 Quoted by Westphal, op. cit. p. 67 et s.

2 O. iii. 2?. 3 O. i. 547. 4 O. iv. 2633.

5 Quoted by Kaluza, Der Altenglische Vers., Berlin, 1894 ; part ii. p. 7 ss.

NUMERI ITALICI ET SATURNII 47

it, though in applying it to the Saturnians he con- fined it to the second and fifth feet. But it is no invention of the theorists, it is rather das ewig kindliche of poetry.

[L] CORNELIO L F. SCIPIO1

[A] IDILES. COSOL CESOR

1. Hone oino ploirume consentiont R(omai)

2. Duonoro optimo fuise viro

3. Luciom Scipione filios Barbati

4. (Co)nsol censor aidiles hie fuet (apud-vos)

5. Hec cepit Corsica Aleriaque urbe

6. Dedet Tempestatebus Aidem mereto (votam) [LCORNELIO.jC N. F. SCIPIO2

7. Cornelius Lucius Scipio Barbatus

8. Gnaivod patre prognatus fortis vir sapiensque

9. Quoius forma virtutei f parisuma fuit

10. Consol censor aidilis quei fuit apud-vos

11. Taurasia Cisauna Samnio cepit

(^12. Subigit omne Loucanam opsidesque abdouxsit

Wolfrlin3 thinks the prose heading of (1-6) is much older than the Saturnians which follow, on account of the ruder form of the letters, on account

1 C. I. L. i. 32. Consul 259 B.C.

2 C. I. L. i. 30. Consul 298 B.C.

3 Revue de Philologie> vol. xiv. p. 113 ss.

48 THE STRESS ACCENT IN LATIN POETRY

of the more ancient spellings, Cornelio (n. case) over against filios (1. 3), cosol, cesor, over against consoly censor (1. 4) and because the simple order of offices is changed in line 4, metri gratia. He agrees with Ritschl, Desau, and others, in placing it before (7-12), though he considers 240 B.C. (the date generally given) too early, and suggests 200 B.C. I cannot agree with him, however, in finding in the expression Duonoro Optimo, a trace of Greek influence, for it is an idiom common to the popu- lar speech of many languages (Cf. the Hebrew, Holy of Holies). The fact that the second half- line is not so regularly shorter than the first, seems to me an argument for giving the priority in time to (1-6). In line 1, the first half -line consists of six syllables, and the second of six, if we emend Romai, of seven, if Romanai; line 5 has six syl- lables in the first half and six (or seven) in the second; and in line 6, the sense seems to require some such participle as votam, although the stone is broken so close to the preceding word that we cannot be sure (quite different from line 2, where the space proves viro to be the last word). Further, the only monosyllables occupying a whole foot (Jionc, /lie, hec) refer to Scipio himself, making it possible that the additional emphasis of the slow tempo (one full beat) is not accidental, character- istic, as it is, of all primitive poetry, as Mars pater, te pre cor. It may be, however, that we have here only an instance of the Schva Indogermanicum,

NUMERI ITALICI ET SATURNII 49

hice, being pronounced, although it was not written.

Line 4 is emended with certainty from the similar line (10) in the second inscription, where quei is read instead of hie, as also in (9) the relative pro- noun has taken the place of the demonstrative. Thurneysen 1 is certainly wrong in accenting apud- vos, on the analogy of tecum, mecum, " wo deutlich der Ton auf den Pronomen runt," for the latter follow the usual accentuation of a dissyllabic word, cum, as Priscian says,2 being merely an enclitic, while in apiid-vos (like the English among-you) the unemphatic pronoun is treated as an enclitic and the accent falls, as before, on the penult. This accentuation is supported by the versification of Plautus and Terence, for example, Trin. 421, abs. te accepi, 619 erga-te, 733 penes me; Merc. 585 apud-me.3

In accenting, fortis-vir sapiensque, Lindsay re- marks, "fortis-vir, a word like our gentleman ; " and so it is, but if we were writing the words in English, we would say, "gentleman and scholar," with the primary accent on the first syllable, but a secondary accent on " man " for the sake of the rhythm. So Kipling writes, " On the road to Mandalay," " 'Er petticoat was yaller," "Elephints a-pilin' teak," and so on ; of the thirteen trisyllabic

1 Op. cit. p. 24. 2 xiv. 6, p. 27, H.

3 Lindsay, "Latin Accentuation" (second paper) Classical Re- view, vol. v. p. 403.

50 THE STRESS ACCENT IN LATIN POETRY

words in the poem, six are accented on the first syllable or on the last; and so have a secondary- accent on the other, as also the proper name of four syllables, " Supiyawlat." Now English poetry is based solely upon accent; there are no distinctions of long and short syllables ; and yet all of these syllables, with a secondary accent, are what may be termed heavy syllables. In Latin poetry on the other hand, the distinction could never have been unknown. Their alphabet was borrowed from Greek colonists in Italy, so that their intercourse with Greeks, though perhaps slight, was long continued. It is not, therefore, surprising, if, even before the great waves of Greek influence in the time of Ennius and his successors, the writers of the Saturnians modified their native accentual metre by the recognition of quantity. The influence of quantity was unquestionably first felt in the second half-verse of the Saturnians.1 Half -verses like mdximas legiones (1. 33), lacrimis cum multis (1. 88), "read themselves," with a primary accent on the first syllable of maximas, filiam, and a binary accent on the last, as in the popular verse chanted by the soldiers on the oc- casion of Caesar's triumph over Gaul : —

Ecce Caesar nunc triumphat qui subegit Gallias, etc.

1 Scholl, De Accentu Latino, Leipzig, 1876, p. 32, in a note, " Verbo moneo etiam in Saturnius posteriorem versus partem maiorem fere concentum praebere, quam prior em"

NUMERI I TALI CI ET SATURNII 5 1

It seems to me that one may go a step further and say that the influence of quantity was more strongly felt in the fourth and fifth feet of the Saturnians than in the sixth, a view which is con- firmed by the usage of the later hexameter, where the clash between quantity and word accent is considerably more frequent for the sixth foot than for the fifth (a little less than 2 to i).

Third Scipio Inscription1

13. Quei apice insigne dial(is) (fl)aminis gesisti

14. Mors perfe(cit)tua ut essent f omnia brevia

15. Honos fama virtusque gloria atque ingenium

16. Quibus sei in longa licu[i]set tibe utier vita

17. Facile facteifs] superases gloriam maiorum

18. Qua-re lubens te in gremiu Scipio recip(i)t

19. Terra Publi prognatum Publio Corneli.

The tone of this inscription is at once more personal and more modern. As. Boissier 2 remarks, " II semble qu'ici le vieux saturnien s'attendrisse et qu'il veuille s'accommoder a des temps nou- veaux." The most noteworthy point in the versifi- cation is the greatly increased length of the first half-line in comparison with the second. In this it resembles the latest of the well-authenticated Saturnians.3

1 C. I. L. I. 33, Consul 180 B.C.

2 Journal des Savants (1881), p. 167.

3 The epitaph of M. Caecilius, C. I. L. I. 1006, 130-100 B.C.

52 THE STRESS ACCENT IN LATIN POETRY

Hoc est factum monumentum Maarco Caicilio

Hospes gratum est quom apud-meas restitistei seedes Bene rem geras et valeas dormias sine qura.

As has been already remarked, this is more than halfway to the popular trochaic septenarian, for example : —

• / r r / / /■

Postquam Crassus carbo factust Carbo crassus factus- est.

Fourth Scipio Inscription1

20. Magna sapientia multasque virtutes

21. Aetate quom-parva posidet hoc-saxsum

22. Quoiei vita defecit non honos honore

23. Is hie situs quei numquam victus est virtutei ; — -

24. Annos gnatos viginti is(div)eis (man)datus

25. Ne quairatis honore quei minus sit mand(at)u(s).

Multasque (1. 20) is like aetate (1. 21); in each half-line the thesis of the first foot is suppressed. In the second half of 24 there remains only an upright stroke on the stone for the first letter of the second word. Havet prefers the emendation (loc) which suits his metre as (dtv) suits mine. The read- ing of the half-line is very doubtful.

Sorana Inscription2

26. Quod re-sua d[if]eidens asper(e)afleicta

27. Parens timens heic vovit voto hoc solut(o)

1 C. I. L. I. 34. 130 b.c.

2 C. I. L. I. 1 1 75. 150-135 B.C. according to Ritschl.

NUMERI ITALICI ET SATURNII 53

28. [De] cuma facta poloucta leibereis lube(n)tes

29. Dono danunt Hercolei maxume mereto

30. Semol te orant se(u)oti crebro condemnes.

Havet reads Herclei, with the remark,1 " Pro Hercolei quod metro repugnat aut Herclei, pro- nuntiandum est syllabis duabus aut quattuor fortasse Herecolei ; scilicet primum ex 'HpatcXfjs fieri debuit *Heracoles, deinde *Herecolesf postremo Hercoles (sic *balancum, balineum y balneutri)." This I cite as illustrative of his method. When balineum is written with four syllables — in Plautus, for example — it represents 6 v w — , not 6 w — . Why, then, should Hercolei be supposed to have a dif- ferent number of syllables than are written, except, of course, quod metro repugnat ?

These oldest inscriptions, and, above all, the first and second Scipio Inscriptions, are of the utmost importance in determining the norm for the Satur- nian metre, because we may be reasonably sure that we have them in the form in which the Ro- mans had them, while in the case of verses resting on Ms. authority, both accidental and intentional changes may have been made by generations of copyists.

An examination of these thirty lines; then, give the following rules for the Saturnian metre : —

I. Every Saturnian is divided by a caesura

1 Op. cit. p. 233.

54

THE STRESS ACCENT IN LATIN POETRY

into two parts, equal in time but not in the number of syllables.

II. Each half -verse is made up of three trochaic beats, with an occasional anacrusis.

III. The third and sixth beats, which are the strongest, must coincide with the primary accent of the word ; the first, second, fourth, and fifth beats may fall on a less strongly accented syllable.

IV. The thesis may be suppressed in the first, second, fourth, fifth feet, though never in two suc- cessive feet, nor in the third or sixth foot.

The scheme, therefore, for the first half-verse would be : —

' A

II.

' A

S \J \J KJ

III. S \J

i.e. any combination of these feet, making not less than six nor more than eight syllables (average seven).

And for the second half -verse : —

IV.

S \J

' A / \j \j

V.

' A

VI.

i.e. any combination of these feet, making not less than five nor more than seven syllables (average six).

There are usually three words in the first half-

NUMERI I TALI CI ET SAT URN II 55

verse and two in the second. Elision does not take place between the half -verses ; in other places it may or may not take place, according to the necessity of the versification. A long vowel is sometimes shortened before an initial vowel or h (not elided), as it is occasionally in Accius, Ennius, and even in later poets.

Two second half-verses, parisuma fuit (1. 9) and omnia brevia (1. 14), seem to have but two accents, unless with Havet and others we read omnia brevia, for which, as has been said, there is no warrant in the early poetry. Keller explains them as belonging to the alios breviores mentioned by Caesius Bassus. They both contain, however, the average number of syllables (six) and seem rather formulaic in character, so perhaps the poet fitted them into the scheme as best he could and let it go at that. The Saturnians are not more irregular than other primitive poetry. In the first thousand lines of Beowulf, for example, Kaluza1 finds ninety variants on Sievers's " five types " for the old Ger- man kurzzeile or halbzeile, which corresponds in certain respects to the half-verse of the Saturnians, though perhaps the comparison has been pushed too far.2 It would, indeed, be just as absurd to expect regularity and perfection in primitive verse as in primitive sculpture. But just as the latter, in spite of its conventional misrepresentations, and

1 Kaluza, op. cit. part i, p. 32 ss.

2 Cf. Bartsch, op. cit.

56 THE STRESS ACCENT IN LATIN POETRY

all the hardness and stiffness of unsubdued mate- rial, shows some of the beauty of the human form, so through the limping measures of the former we can trace the beginnings of inspiration. It is poetry, that is the essential thing about it, and any theory which destroys the poetry, no matter how well it reads, is worthless.

There are five verses quoted by different writers from inscriptions : 1 —

a. Uno cum plurimae consentiunt gentes.

31. • \ >> \ f f/

b. Unicum plurimae consentiunt gentes.

32. Populi primarium fuisse virum.

These lines are from the epitaph of Atilius Cala- tinus (which Wolfflin 2 thinks served as the model for the first Scipio-Inscription) quoted by Cicero.3 Havet emends, unum complurimae; Reichardt, Hunc unimi plurimae ; Lindsay, uno complurimae, with the remark, " I give a double accentuation to (allitera- tive) co7nplurimae and primarium, but not to con- sentiunt. The reading complurimae is favoured both by the alliteration and by the ' echo ' of the other line of the distich." The important point, it seems to me, is that the two readings uno cum and unicum must have sounded the same ; the second

1 The information in regard to the sources of the following verses is taken from Havet, Zander, Baehrens, and Lindsay (all cited above).

2 Op. cit. p. 116 et s.

3 (a) De Fin. ii. 35, 116; (d) De Sen. 17, 61.

NUMERI I TALI CI ET SATURXII $?

syllable cannot, therefore, have been accented, while the first syllable and the <:^;;z-syllable were accented. Complurimae is certainly right (though with Zander x I would restore the ancient spelling,

Oino comploiriMnae consentiont gentes),

but then the question arises, why should " allitera- tive complurimae " receive a double accentuation and consentiunt, with the same number of syllables and beginning with the same letter (presumably, therefore, alliterative), not, except, to quote Havet's illuminating phrase, quod metro repttgnat ?

33. Fundi t, fugat, prosternit maximas legiones.

From the epitaph of Acilius Glabrio, 181 b.c. (circ.) quoted by Caesius Bassus de Metris.2

34. Magnum numerum triumphat hostibus devictis

Quoted, apparently from an inscription, by Censorinus.3

35. Duello magno diremendo regibus siibigendis.

From the inscription of M. Aemilius Lepidus, in honour of his father, L. Aemilius Regillus, 179 b.c, quoted by Caesius Bassus.4

It would be a mere jeu d 'esprit to put back into

1 Op. cit. p. 58.

2 vi. 265 K. Utrum exemplem suspicor esse a Caesio aut aliquo grammatico fictwn, Zander, op. cit. p. 57. 3 vi. 615 K.

4 vi. 265 K. Utrum exemplem suspicor esse a Caesio aut aliquo grammatico Jiclum, Zander, op. cit., p. 57.

58 THE STRESS ACCENT IN LATIN POETRY

Saturnians the lines given in prose by Livy1 or in hexameters by Priscian.2 Equally unimportant would be the attempt to fill out, or to place in one part or another of the verse, the stray words and phrases quoted, in some instances, by the gram- marians. For the Odyssia of Livius Andronicus and the Bellum Punicum of Naevius, therefore, I give only complete lines, requiring no emendation, or the very slightest.

From the Odyssia of Livius Andronicus (Ob. 204 B.C.).

36. Verum mihi Camena insece versutum. Quoted by Gellius, xviii. 9, 5, for insece. It is the opening line of the Odyssia.

37. Neque^enim te oblitus-sum Laertie noster.

Quoted by Priscian 3 for voc. sing, in -ie. Mss. neque enirn, neque tamen, Laertiae, Lertie, O Laertiae, and Laertie. Tarn is Korsch's suggestion. Zander, with Reichardt following (as usual), reads

Neque tarn ted oblitus sum Laertie noster. .38. Argenteo polybro aureo eclutro.

ap. Non. 544 M., s.v.polybrum; eclutro is Baehrens's suggestion. Cf. eicXovrpov. Mss. et glutro.

1 For example, i. 35, 5-14 ; de Anco Marcio, v. 16, 8 ss. ; vi. 29, 5 ss., etc. . 2 For example,

Inferus au superus tibi fert deus funera, Ulixes (i. p. 96), Cum socios nostros mandisset impius Cyclops (i. p. 419), At celer hasta volens perrumpit pectora ferro (i. p. 335). 3 i. p. 301 H.

NUMERI ITALICI ET SAT URN II 59

\ tt

39. Tuque mihi narrato omnia disertim. ap. Non. 509 M., s.v. disertim. Three Mss. have tu quae and one tuque. I prefer tuque because it gives the usual number of words in the first half- line.

40. Quando dies adveniet quern profata Morta~est. Quoted by Gellius hi. 16, 11 for Morta, as the

name of one of the three Fates.

* f tt ? t ft

41. (Aut) in PylunTadvenie(n)s aut ibrommentans.

Quoted by Festus1 for ommentans. Mss. ad-

venies. Corr. Scaliger. Aut is Baehrens's almost

certain emendation.

* t tt f \ tt

42. Ibi]demque vir summus adprimus Patroclus.

Ap. Gell. vi. 7, 11. After a discussion to prove that adverbs compounded with ad should be ac- cented on the first syllable, this verse is quoted with the remark, adprimum autem longe primum Livius in Odyssia dicit.

t f t' t \ tt -*y

43. Partim errant nequinont Graeciam redire.

Festus,2 nevuinont pro nequeuntt ut solinunt ferinunt

pro solent feriunt diccbant antiqui.

t / tt t \ *,

44. Apud nympham Atlantis filiam Calypsonem.

Quoted by Caesellius Vindex3 for Calypsonem, ace. sing.

t r ^^ tt t \ tt X

45. Utrum genua amploctens virginem oraret. )

1 Thewrewk de Ponor, p. 218, 14. 2 Thewrewk de Ponor, p. 162, 24. 8 Ap. Pris. i. p. 210 H.

V

60 THE STRESS ACCENT IN LATIN POETRY

Diomedes,1 vulgo dicimus " amp lector" veteres immutaverunt " amploctor" crebo dictitantes. One Ms. has orraret?

46. Ibi manens sedeto donicum videbis. |

47. Me carpento vehentem domum venisse.

Chairsius,3 donicum pro donee. The (single) Ms. has vehementem. Havet 4 reads vehente int Thur- neysen,5 vehentem, Bucheler6 and Zander,7 endo do- mum. If I were emending the second half-line I should write, domum venisse patris, from Homer's line,8 —

aoruSe iXOoifxev ml iKco//,e#a Sahara 7rar/)Os.

48. Simul-ac dacrimas de~ore noegeo detersit.

Festus, IVoegeum, amiculi genus? Noegeum can- didumP Dacrimas should probably be written for Livius.11 Ms. lacrimas.

49. Merjcurius cumque eo filius Latonas.

Quoted by Priscian ^ for Latonas, gen. sing. Havet13 and Baehrens,14 following Bartsch, supply venit, at the beginning of the verse, but without probability, for the resemblance to Homer's line 15 is not striking.

1 384, 7 K. 2 Paris. 7493.

3 197, 15 K. 4 Op. cit. p. 352. 5 Op. cit. p. 14.

6 Neue Jahrbucher fur Phil, lxxxvii. p. 332. 7 Op. cit. p. 88.

8 f 296. 9 186, 32 Th. de P. 10 187 Pauli Excerpta.

11 Th. de P. p. 48. I2 I. p. 198. 13 Op. cit. p. 372.

14 Op. cit. p. 40. 15 6 322.

NUMERI ITALIC! ET SATURNII 6 1

50. Nam divina Monetas filia im docuit. Quoted by Priscian1 for Monetas, gen. sing. The Irish Mss. divina, the others (the larger number) diva. All Mss. filiam. Filia must be nominative, and since an accusative may very well have been added,2 I have followed Fleckeisen (and Zander) in reading im, not me (Lindsay) for the passage in Homer is third person.

51. Topper facit homines ut prius fuerunt. Quoted by Festus3 for topper. Mss. utrius fuerint. tit prius is Duntzer's suggestion,4 fuerunt, Biiche- ler's.5 Zander6 (and Reichardt) homoncs.

■ 52. Topper citi ad-aedis venimus Circai.

53. Simul duona eorum portant ad-naves.

V 54. Millia alia deinde isdem inserinuntur.

These three lines are quoted together by Festus,7 immediately after 1. 51, from Livius, in Odyssia vetere. From their subject they can scarcely be- long to the story of Circe, and Lindsay follows Thurneysen in attributing them to Naevius. The Mss. read Circae and the third line, millia alia in isdem inserinuntur, in all three, therefore, the sec- ond half-line has five syllables. Lindsay suggests Circai as a " perfectly justifiable alteration," though he reads Circae. Baehrens and Zander frankly

1 I. p. 198. 2 0 480 et ss. 3 532 Th. de P.

4 De Versu que??i vocant Satarnio, p. 45.

5 1. 1. 6 Op. cit. p. 91. 7 1. 1.

62 THE STRESS ACCENT IN LATIN POETRY

rewrite lines 53 and 54. The latter is very suspi- cious for the reason that it is the only one of the extant Saturnians in which a single word occupies the whole of the second half-verse. I have, there- fore, written deinde, which might easily have dropped out, and followed Baehrens in placing isdem in the second half.

55. Sancta puer Saturni filia regina.

Quoted by Priscian 1 as an instance of puer for puella. Baehrens suggests maxima regina, Zander, omnium regina, "vel aliquid, infinitaconiectura." 2 He marks it dcsperatus.

From the Bellum Punicum of Naevius (ob. 198 B.C.): —

/ 56. Eorum sectam sequuntur multi mortales.

57- Ubi foras cum-auro illic exibant.

V 58. Multi aliPe Troia strenui viri.

Servius Danielis ad Verg. Aen. ii. 797. There is no need of any change.

59. Iamqu(e) eius mentem fortuna fecerat quietem.

Priscian,3 etiam simplex (quies) in usu invenitur trium generum.

60. Inerant sign(a) expressa quo-modo Titani.

Jj 61. Bicorpores Gigantes magniqu(e) Atlantes.

62. Runcus atque Purpureus filii Terras.

1 1. 232, 5 H. 2 Op. cit. p. 87. 3 I. 242, 20 H.

NUMERI I TALI CI ET SAT UR Nil 63

Quoted by Priscian,1 for Terras, gen. sing, and again2 (1. 60 and 61 only) for Titani, nom. pi.

63. Silvicolae homines belliqu(e) inertes.

Macrobius,3 silvicolae Faiini. Zander, followed by Reichardt, reads homones.

64. Bland ( e) et docet percontat Ae|n(ea) quo pacto.

65. Troiam urbem liquerit.

Nonius,4 Liquerit significat et ' reliquerit? In another place 5 he quotes the line again, this time with reliquisset, but I agree with Zander 6 in giving greater weight to the former reading, because there, Nonius makes the word the subject of a note. According to Havet7 the Mss. give (for the first place) enos, e?ias, ennius, and percontcnas, and (for the second) aen, aeneam, acnius, emiius, aencidos. Quintilian, however, says,8 " Ne miremur quod ab antiquorum plerisque Aenea ut Anchisa sit dictus." Lindsay thinks that, " Quo-pacto is a word-group like quomodo" his own remark, however, shows the difference. In quo-pacto, the two parts did not coalesce so completely as to be felt as a single word, both (I think) on account of the long penult, and the greater individuality of the word pactum. But even if one accepts his theory of "sentence accentuation," as I do, in the main, it

1 1. 198, 15 H.	2 I.217.	3 Sat. vi. 5, 9.
4 335. 1 M.	5 474, 5 M.	6 Op. cit. p. 101
7 Op. cit. p. 343.	8i. 5, 61.

64 THE STRESS ACCENT IN LATIN POETRY

does not follow that in a primitive verse like the Saturnian, relatively unimportant words should never be accented.

66. Prim(a) incedit Cereris Proserpina puer.

Priscian,1 hie et haec puer vetustissimi.

* / 67. Deinde pollens sagittis inclitus arquitenens. \ 68. Sanctus love prognatus Pythius Apollo.

Quoted by Priscian2 following 66, and by Macro- bius3 (6y and 68 alone) for arquitenens. Mss. Sanctusqne Delphis prognatus. The -que is cer- tainly out of place in 68. Zander puts it in 67, reading inclutusque, but this makes the connection too close between the first and second half-lines. It seems to me more likely that sanctusqne was written by some scribe for sanctus love, who then added the meaningless Delphis. Zander rewrites the line, —

Sanctus love Deli Putius prognatus Apollo ....

but this is unlikely, (1) because prognatus occupies the third place in lines 8 and 19, next to its ablative, and (2) there is no undoubted instance of a " run-over" line among the extant Satur- nians. It seems to me that the least violent remedy is to lengthen the -us, in arsi, on the ground that it is a conventional, formulaic ex- pression, a sort of " tag," which the writer forces

II. 231, 13 H. 2 I. 231,13 H. 3 Sat. vi. 5, 8.

NUMERI I TALI CI ET SATURXII 65

into the service of his verse because of its familiarity.

/ 69. Postquam avem aspexit in | templo Anchisa

70. Sacr(a) in-mensa Penatium ordine ponuntur ; . 71. Immolabat auream victimam pulchram.

Probus1 ad Verg. Ec. vi. 31. Biicheler suggests Penatum, and Havet2 reads In auream molabat, quoting Lucretius, Vergil, and Horace for the tmesis, but no change is necessary, Penatium finding an echo in the auream of the following line.

72. Urit vastat populatur rem | hostium concinnat.

Nonius,3 Concinnare, conficere vel colligere. I have followed Thurneysen's suggestion 4 in transpos- ing populatur vastat, cf. Fundit, fugaty prostemit

(i. 33).

73. Virum praetor adveniet auspicat auspicium.

Nonius,5 Auspicavi pro auspicatus sum. Havet reads adveniens as in 1. 41, Baehrens adveneit. The double accentuation of auspicium is made probable by the repetition of the proposition, so in 75 foil.

74. Censet eo venturum obviam Poenum. Nonius,6 C ens ere significat existimare, arbiti'ari. Mss. censent and censet.

75. Su|perbiter contemtim content legiones.

1 p. 14 K. 2 Op. cit. p. 388. 3 90. 23 M.

4 Op. cit. p. 33. 5 p. 468. 20 M. 6 p. 267. 17 M.

F

66 THE STRESS ACCENT IN LATIN POETRY

Nonius,1 Contcmtim, contemncnter. Lindsay re- marks, " contcmptim conterere recurs in Plaut. Poen. 537," and then accents Siipc'rbiterconte'mtim,giv'mg the short syllable su- a full beat, and disregarding the alliteration.

76. Septimum decimum annum ilico sedent. j Nonius,2 Ilico, hi eo loco. Havet emends sederent to get rid of the uncomfortable short vowel in the penultimate syllable of the line, L. Muller sedentes.

77. Sicilienses paciscit obsides ut reddant. Nonius,3 Paciscunt. One Ms.4 gives only the verse quoted above, others, "Id qiwque paciscuntnrh moenia sint que Lutantinm 1'econciliant captivos plurimos idem Sicilienses paciscit obsides tit reddant." Bucheler assigns idem, I think rightly, to Nonius, but it is possible that it may begin line yy.

78. Ei venit in-mentem hominum fortunas. Quoted by Priscian6 for fortunas, gen. singular. Two Mss. have mentem, the majority mentey which is probably an abbreviation.

79. Hone|rariae honustae stabant in flustris. Isidorus,7 Flustrnm motus maris sine tempe state fluctnantis .

80. Res divas edicit Praedicit castus. Quoted by Nonius 8 under castitas.

1 p. 516, 1 M, also 515. 8 sq. s. v. superbiter. 2 p. 325, 6 M. 3 474, 16 M. 4 Paris 7665. 5 Paris 7667 paciscunt.

6 i. 198, 15 H. 7 de Nat. Rer. 44. 8 197, 14.

V

NUMERI ITALIC! ET SATURNI1 6 J

81. Summe deum regnatur quianam me genuisti? Festus,1 Quianam pro quare et cur positum est apud antiquos. Quianam genus isti is twice written in the Ms. The reading me genuisti is Havet's.2

/ 82. Sesequ(e) ii perire mavolunt ibidem.

83. Quam cum-stupro redire | ad suos popularis.

Festus,3 Stuprum pro turpitudine. Lindsay sug- gests poplaris as a possible reading, quoting Fleckeisen, Plaut. Rud. 740, and TLoirXapis (Arch. Ep. Mitth. i. p. 7).

84. Sin illos deserant fortissimos viros.

85. Magnum stuprum populo fieri per gentis.

Following 82, 83 in Festus. In these two pairs of lines the similar ending in the first half is worthy of notice, in 82 and 83 the rhyme perire, redire, in 84 and 85 the dactyl (accentual) instead of the usual trochee.

86. amborum uxores.

87. Noctu Troiad (e) exibant capitibus opertis.

88. Flentes ambae"abeuntes lacrimis cum-multis.

Servius ad Aen. iii., 10 Nacvius inducit uxores Acncae ct AncJiisae cum lacrimis Ilium relinqucntes. I have placed the primary accent of c&pitibus on the first syllable, as infacilius, bdlineum in Plautus, and lengthened the last syllable in arsi following Vergil's P ectoribus Lillians spirantia consulit extat

1 p. 340 Th. de P. 2 Op. cit. p. 301.

3 p. 460, 27 et ss. Th. de P. 4 ^Sw. iv. 64.

68 THE STRESS ACCENT IN LATIN POETRY

89. Ferunt pulchras creterras aureas lepistas.

This verse is quoted three times, by Caesius Bas- sus,1 by Marius Victorinus,2 who does not think it belongs to Naevius, and by Marius Plotius.3 The Mss. have pateras creterras, cratcras, creterras, but creterra seems to be the old form. (See Georges, Lex. Wortf. s.v.)

90. Magni metus tumultus pectora possidet.

Nonius,4 Metus masculino Naevius. One Ms. has possidit, the other possidet, which may, however, be scanned possidet (3d conj.).

91. Novem Iovis Concordes filiae sorores.

Caesius Bassus5 and Mar. Victorinus.6 This verse is slightly confirmatory of the emendation Sanctus love prognatus (1. 68).

92. Patrem suum supremum optimum appellat. Varro,7 Naevius . . . supremum a superrimo dictum.

93. Scopas atque verbenas sagmina sumpserunt.

Paulus ex Fest.,8 Sagmina dicebant herbas ver- benas. On the opposite page after Naevius is the line Ius sacratum Iovis iurandum sagmine. Mss. scab os, scapas, s capos.

94. Simul alis aliunde rumitant inter sese. Paulus ex Fest.,9 Rumitant rumigerantur. I have

1 266 K. 2 139 K. 3 531 K. * 214. 7. 5 266 K. 6 139 K.

7 Z. Z. vii. 51. 8 p. 469 Th. de P. (JPauli Excerptd).

9 p. 369 Th. de P. {PatUi Excerptd).

NUMERI ITALIC! ET SATURNII 69

adopted Boethius's suggestion alls, to avoid the double resolution in the first half-verse. Mss. alius. I can find no Ms. authority for inter se, though it is a very slight change and is read by Havet, Baehrens, Zander, Reichardt, and Lindsay. Gf. apud-vos (11. 4, ex em., and 10).

95. Apud empori(um) in-campo hostium pro moene.

Festus,1 Moene singulariter dixit Ennius. O. M til- ler was the first to notice that this line was a Saturnian, and substituted Naevius for Ennius. Havet2 may, however, be right in suggesting that the line of Ennius and the name Naevius have been omitted by a copyist.

96. Summas opes qui regum regias refregit.

Quoted by Diomedes3 and by Atilius.4 It may not be by Naevius.

97. Dabunt malum Metelli Naevio poetae. — ■ Quoted by Caesius Bassus,5 optimus (Saturnius) est quern Metelli proposuerunt de Naevio aliquo- tiens ab eo vcrsu laeessiti, also by Mar. Vict.6 Mar. Plotius,7 Atil. Fortun.,8 Ter. Maur.,9 Pseud.- Ascon.10 Malum dabunt is given by the first three.

98. Immortales mortales si | forent fas flere.

99. Flerent divae Camenae Naevium poetam.

!p. 124 Th. de P. 2 Op. cit. p. 296. 3 i. p. 512 K.

4 vi. p. 293 K. 5 vi. p. 266 K. 6 139 K.

7531K. 8294K. 925I7.

10 In Cic. Verr. i. 10, 29.

70 THE STRESS ACCENT IN LATIN POETRY

ioo. Itaque postamquarrPest Orcho traditus thesauro. 101. Obliti-sunt Romae loquier lingua Latina.

The famous epitaph written by Naevius to be inscribed on his own tomb. Quoted by Gellius1 along with similar epitaphs of Plautus and Pacu- vius. Thurneysen 2 thinks that the last verse must either be transposed or regarded as a later imitation, because the caesura is neglected and there are six accents. Similarly he considers the verse, —

r r r r r r

Terra pestem teneto salus hie maneto

radically different from the epic Saturnians.3 Of the other Numeri Italici he quotes only Hlberno ptdvere and Novzim vetus, none of those consist- ing of three beats, although we found that the number was considerable.

It is rather the fashion with the later editors to throw doubt upon the antiquity of the Naevius In- scription, but if it was composed by Gellius on a purely quantitative basis, then it is certainly an inferior piece of work. Comparing it with four lines taken at random from Naevius, the regard (or perhaps disregard) of quantity seems about the same : —

KJ ^ W

98. Immortales mortales si forent fas flere.

99. Flerent divae Camenae Naevium poetam. i. 24, 2. 2 Op. cit. p. 52. 3 Op. cit. p. 54.

NUMERI ITALICI ET SATURNII J\

WWW W ±s

ioo. Itaque postquam est Orcho traditus thesauro. 101. Obliti-sunt Romae loquier lingua Latina.

W ^/ W WW

75. Superbiter contemtim content legiones.

W W ^ W w

91. Novem Iovis Concordes fi]iae sorores.

w w w w w

87. Noctu Troiade exibant capitibus opertis.

WW ^ w

88. Flerent ambae abeuntes lacrimis cum multis.

It would be hard to see how a half-verse com- posed entirely of long syllables (like the first half of 98, 101, 87) could be read as poetry without a stress beat, and to suppose that this beat clashed with the natural accent of the word in all but the last foot would be to make the ancient sing-song measure of the native prophets more Greek than the iambic and trochaic metres of Plautus and Terence, where word and verse accent tend broadly to coincide.

Alliteration is not of prime importance in the Saturnian verse, and it is a subject that has been very fully treated.1 Such evidence as it affords is in favour of the Accentual Theory, scansions like Gnaivod patre progndtns not only introducing a violent clash between word and verse accent, but disregarding the alliteration as well.

1 Cf. Keller, op. cit. p. 33 et ss., Loch, De Usu Alliterationis apud Poetas Latinos, Halle, 1865; Dingeldein, Der Reim bei den Griechen und R'dmern, Leipzig, 1892; and others.

72 THE STRESS ACCENT IN LATIN POETRY

A reading of the verses based primarily upon the natural accent of the words (i) is in harmony with the Latin or rather Italic principle of word accent, and (2) brings the Saturnians into line with what we know of the earlier and later popular poetry; while the greater influence of quantity in the last half explains the early and complete naturalization of the Greek hexameter. It is per- haps not without interest to note, that in the hexameters of Lucilius which approach most nearly to the popular standard, there are 293 lines (49.4%) in which word accent and quantity coincide in the last three feet, and 52 lines (8.7%) in which there is no clash whatever throughout the entire line. But, on any theory, it must be confessed that the Saturnians limp. In some lines three unaccented syllables are slurred over, in others a single long syllable is held a full beat, though both irregularities find abundant illustration in the primitive poetry of every people.

In each Saturnian there is a strong pause be- tween the two halves of the line, often a complete break in the sense. This points, as has been said, to a composite nature. A common type of the Numeri Italici is x v; I 7 w u Ku| (Buos Lases invate, for example ), which recurs in the first half- line in 50 out of the 100 Saturnians quoted above. This is varied to/ Al/uu| ' kj |,or/^^l/A| ' ^, or (rarely) /v|/wuu|/v;lor/^l ' ^1 < ^ ^ I . The second half -verse is less regular, the

NUMERI I TALI CI ET SATURNII 73

type /^l/ v| £ \j\ {Jupiter Dapdlis) being varied to 7 A I / w I / w I , or 7 w I / A I / u? or S yj\ 7 ^ w I 7 w I ( or 7 wu| x aI 7^ I, very rarely 7 vuin the last foot. In the midst of these ir- regularities, however, there is one rule that is never violated ; the third and sixth beats fall, de rigueurt on the primary accent of the word. After the strong caesura and the falling {i.e. trochaic) metre, this seems to me the most important characteristic of the Saturnians. In some of the verses all six beats coincide with the primary accent of the word, as, —

Gnaivod patre prognatus fortis vir sapiensque

Honos fama virtu sque glori(a) atqu(e) ingenium

Ne quairatis honore quei-minus sit mandatus

Quando dies adveniet quern profata Morta^est

but very often it is necessary to hold a long syllable for one full beat. This is more often the first syllable of a word, as, —

Hone oino ploirume con sentiont Romai

Aetate cum parva posidet /loc-saxsum

rarely (there are seven instances in all) the last, in a proparoxytone word of three syllables as, —

Taurasis Cisauna Samnitf cepit.

The last beat in each half-verse is, therefore, the strong beat ; the voice, slipping over the less im- portant first and second, beats to rest with satisfac-

74 THE STRESS ACCENT IN LATIN POETRY

tion upon the third. This explains, too, the fact that the thesis is most frequently suppressed in the fifth foot ; coming next to the last beat of the whole line, it is the weakest of the six and falls quite indifferently on a single long syllable, a long followed by a short, or on the first of three short syllables — though never on a short followed by a single short syllable. The same preponderance of the last beat in each hemistich is characteristic of the longer ariyoi ttoXltlkol — namely, those of fif- teen syllables, developed out of the iambic tetram- eter catalectic of Classical times 1 — and as Gaston Paris2 has shown, of French poetry. It would be insufficient in English or German verse on account of the heavy stress accent of both languages, but it harmonizes perfectly with the nature of the accent in French, in late Greek, and in Latin, and it seems to restore a fugitive beauty even to the verses contemptuously relegated by Ennius to the fauns and satyrs.

1 Christ, Metrik der Griechen und Romer, Leipzig, 1879, p. 375.

2 J&tude sur le Role de V Accent latin dans la Langue fracaise, Paris and Leipzig, 1862, p. 106 et ss.

Ill

THE QUANTITATIVE METRES

In the Numeri Italici and, as we have seen rea- son to believe, in the Numeri Saturnii, metrical accent and word accent coincide. The same is true, to a great extent, of the verses of Plautus and Terence, where a syllable may be shortened either by the word stress or the verse stress.1 That this view of the versification of Plautus and Terence was taken by the Romans themselves is strongly suggested by the following passage :2 " Annianus poeta praeter ingenii amoenitates literarum quoque veterum et rationum in Uteris oppido quam peri- tus fuit et sermocinabatur mira quadam et scita suavitate. Is dffatim ut ddmodum prima acuta, non media, pronuntiabat atque ita veteres locutos censebat. Itaque se audiente Probum grammati- cum hos versus in Plauti Cistcllaria legisse dicit : Potine tu homo facimis facere strenuum ? aliorum dffatim est. Qui faciant sane ego vie nolo fortem perhiberi virtim, causamque esse huic accentui dicebat, quod dffatim non essent duae partes ora- tionis, sed utraque pars in unam vocem coaluisset, sicuti in eo quoque, quod exddversum dicimus, se- cundam syllabam debere acui existimabat, quoniam

1 Cf. Klotz, op. cit. p. 88.

2 Annianus apud Gellium, N. A. vi (vii) 7.

75

76 THE STRESS ACCENT IN LATIN POETRY

una non duae essent partes orationis ; atque ita oportere apud Terentium legi dicebat in his versi- bus : In quo Jiaec discebat ludo exddversum loco Tostrina erat quae dam."

Did word accent play any part in the metres borrowed from Greece, or were they, as Bennett holds,1 an orderly succession of long and short syllables, and nothing more f

We shall look in vain for help from the writers of the Classical Period themselves, for the two passages from Cicero, quoted by Schoell,2 are not to the point. Schoell quotes 3 " non enim sunt alia sermonis, alia contentionis verba, neque in alio genere ad usum quotidianum, alio ad scaenam pompamque sumuntur." And so far the words might seem to apply to accent ; but Cicero is speaking of the different moods of oratory, and he goes on, " sed ea nos cum iacentia sustulimus e medio, sicut mollissimam ceram ad nostrum arbi- trium f ormamus et fingimus. Itaque ut turn graves sumus, turn subtiles, turn medium quiddam tene- mus, sic institutam nostram sententiam sequitur orationis genus idque ad omnem aurium volupta- tem et animorum motum mutatur et vortitur." The second quotation is even more disingenuous. Schoell says — and Bennett quotes him, evidently without looking up the passage — that Cicero 4 praises Ennius " quod non discederet a communi more ver-

1 Op. cit. p. 362. Vendryes quotes him with approval.

2 Op. cit. p. 23. 3 De Orat. iii. 177. 4 Or at. xi. 36.

THE QUANTITATIVE METRES J J

borum." What Cicero does say is as follows:1 "Sed in omni re difficillimum est formam, quod j^apaKTrjp Graece dicitur. exponere optimi, quod aliud aliis videtur optimum. Ennio delector, ait quispiam, quod non discedit a communi more ver- borum ; Pacuvio, inquit alius : omnes apud hunc ornati elaboratique sunt versus, multa apud alterum neglegentius ; fac alium Accio ; varia enim sunt iudicia ut in Graecis non facilis explicatio, quae forma maxime excellat. In picturis alius horrida, iuculta, abdita et opaca, contra alius nitida, laeta, conlustrata delectatur. Quid est quo praescriptum aliquod aut formulam exprimas, cum in suo quod- que genere praestet et genera plura sint? Hac ego religione non sum ab hoc conatu repulsus exis- timavique in omnibus rebus esse aliquid optimum, etiam si lateret, idque ab eo posse qui eius rei gnarus esset iudicari." There is nothing in all this which, taking the words in their context, can be wrested into evidence. But Ouintilian has a remark that seems to throw light on the question. He says : 2 " Ceterum scio iam quosdam eruditos, quosdam etiam grammaticos, sic docere et loqui, ut propter quae- dam vocum discrimina verbum interim acuto sono finiant, ut in illis : —

. . . Quae circum litora circum Piscosos scopulos . . .s

1 1 quote the entire paragraph. 3 Am. iv. 254.

2 Inst. Orat. i. 5, 25.

*]% THE STRESS A CCENT IN LA TIN FOE TR Y

ne, si gravem posuerint secundam, circus did vide- atur non circuities. Itemque cum quale interrogan- tes gravi, comparantes acuto tenore concludunt : quod tamen in adverbiis solis ac pronominibus vindicant, in ceteris veterem legem sequuntur. Mihi videtur condicionem mutare, quod his locis verba coniungimus. Nam cum dico circum litora, tanquam unum enuntio dissimulata distinctione : itaque tanquam in una voce una est acuta, quod idem accidit in illo : —

. . . Troiae qui primus ab oris.1 Evenit, ut metri quoque condicio mutet accentum ;

. . . Pecudes pictaeque volucres : 2

nam volucres media acuta legam, quia, etsi natura brevis, tamen positione longa est, ne f aciat iambum quern non recipit versus herous. Separata vero haec a praecepto non recedent, aut ei consuetudo vicerit, vetus lex sermonis abolebitur."

In this passage, Quintilian seems to say that the accent of separate words may be slipped to another syllable when the words are joined in a sentence, i.e. word accent is subordinate to sen- tence accent — Troiae qui primus. Here, it seems to me, is the opening wedge by which stress forced its way back to supreme importance.

We know that the early Latin verses were based on accent, that in the popular poetry of the Clas- sical Period, of which we have a few fragments,

1 Aen. i. I. 2 Georg. iii. 243.

THE QUANTITATIVE METRES 79

like the Mille song of Caesar's legions, quantity and word accent coincide, as they do to a great extent in the New Poetry of the second century after Christ — the P ervigilium Veneris, for ex- ample ; and, not to mention the curious melange of accent and quantity in the Carmen Apologeti- cutn and Instrnctiones of Commodianus (250 a.d.), that, little by little, accent took the place of quan- tity in the Christian hymns. It seems impossible that, for a short period, under foreign influence, the principle with which Latin poetry began and ended, should have been, not subordinated to quantity, — for that is granted, — but turned abso- lutely out of doors.

An examination of the hexameter from Ennius to Claudius shows the following percentage of coincidences between word accent and " quanti- tative prominence" — to borrow a phrase from Professor Bennett: Ennius 60.5%; Lucilius, 66.2%; Lucretius (500 lines), 64.3%; Vergil, Eclogues (209 lines), 63 % ; Georgics (200 lines), 62.9 % ; Aeneid (400 lines), 60. 1 % ; Horace, Epis- tles (180 lines), 63 % ; Satires (164 lines), 60.4% ; Ovid, Metamorphoses (300 lines), 65 % ; Persius, Sat. i (135 lines), 64.5 % ; Lucan (200 lines), 57.9%; Petronius, Bellum Civile (150 lines), 62.4 % ; Juve- nal, Sat. i (171 lines), 56.7%; Ausonius, Mosella (200 lines), 60.6% ; Auctor Contra Paganos (120 lines), 53.6%; Prudentius (200 lines), 62.3%; Claudian (200 lines), 56.8%. It is worth noting

80 THE STRESS ACCENT IN LATIN POETRY

that (a) we find the highest percentage (66.2 %) of coincidences in Lucilius, who wrote with scarcely more care than he would speak, dashing off a couple of hundred verses, as Horace says, stans pede in uno ; and that the only line which shows a clash in each foot between word accent and "quantitative prominence" is in the passage1 where he is showing how not to write. It runs : — Quo me habeam pacto, tarn etsi non quaeris, docebo Quando in eo numero mansi quo in maxima non est Pars hominum, (ut valeam ; cum tu tam mente labores) Ut periisse velis, quem visere nolueris, cum Debueris. Hoc ' nolueris ' et ' debueris ' te Si minus delectat, quod Te^vtW. Eisocratium est XrjpuSes que simul totum ac cru/A/xapa/acoSes, non operam perdo.

{b) In Lucretius, who rigorously subordinated form to matter, the percentage is the third highest

(64.3%).

(c) In Vergil it decreased from the more fa- miliar Eclogues to the carefully elaborated Geor- gics and A eneid (though it is about the same for the first, sixth, and twelfth books of the A eneid) \ and even more interesting is the fact that the last forty-six lines of the eighth Eclogue — the answer of the shepherdess to her lover — not only contain the inter-rhymed line, —

Limus ut hie durescit et haec ut cera liqueseit, but also 70 °/o of accords.

1 Baehrens, Poetae Latifii Minor es} vol. v. 145.

THE QUANTITATIVE METRES 8 1

(d) Finally, in the smooth hexameters of Ovid, when the last difficulty of technique had been overcome, the percentage reaches the second high- est (65 %).

It seems fair to say from the percentages above quoted, that not until the hexameter degenerated into a mere declamation of the schools, and lost all claim to being called poetry, did its authors come to disregard accent altogether in favour of quantity. Further, as has been remarked,1 the percentage of accords is far higher for the second half of the hexameter than for the first. Taking the lines previously examined, in Ennius 38 % of all the lines show the reading ^wUwUy | for the last three feet ; in Lucilius, 49.4 % ; Lucre- tius, 51.8%; Vergil, Eclogues, 48.4; Georgics, 36 % ; Aeneid, 35.7 % ; Horace, Epistles, 53.3 % ; Satires, 50 % ; Ovid, 46 % ; Persius, 54 % ; Lucan, 41.5 %; Petronius, 54.6 % ; Juvenal, 38 % ; Auso- nius, 51%; Aicctor Conti'a Pciganos, 15%; Pru- dentius, 48.5 % ; Claudian, 22 %.

These figures would be inexplicable if quantity were the only principle at work in the versification of the Latin hexameter. The Law of the Last Half is a part of the heritage of the hexameter from the native accentual Saturnian metre, — for we do not find it in Greek, — and helped out by allitera- tion and assonance, from the same source, accounts

1 Ritschl, Opusc. i, ii, praef. p. xii.

82 THE STRESS ACCENT IN LATIN POETRY

for the immediate popularity of the Annales of Ennius. It was the echo of the older stressed verse in the second half of the quantitative hexameter that made it at once intelligible to the people.

Now to return to the question of how Latin poetry was read in the Classical Period. First, I think, we may consider that Hendrickson has proved Bennett's dictum, " Latin poetry is to be read exactly like Latin prose," paying no regard whatever to accent {i.e. stress) to be untenable, for then it would no longer be poetry ; and that Bennett means when he states his belief that it was so read by the "ancients," not " Servius and the other ancient metricians," but the Romans of the time of Vergil and Horace, the present writer has heard him say again and again. But how are we to account for the clashes between word accent and verse accent, as determined solely by quan- tity? As has been shown, the proportion of ac- cords to clashes is about 60 % to 40 %. In each foot, therefore, if the word accent coincides with the verse accent on the first syllable — the dactylic hexameter alone is here considered — that syllable is pronounced with a slightly increased stress ; if it does not, the two stresses nullify each other in the mind, and the foot is read with absolutely " level stress." But the number of feet in which they coincide is sufficient to carry the verse, especially since in the first foot they coincide three times out of four, and in the fifth and sixth feet they coin-

THE QUANTITATIVE METRES 83

cide in an overwhelmingly large proportion of lines.1

To illustrate, take the opening lines of the Aeneid, because they represent the least favourable showing, the number of accords being smaller in the first fifty lines than in any other group of fifty lines examined.2 The verses are marked as fol- lows : in the top line of markings, which represent stress, the feet in which word accent and quantity coincide, have the stress-mark on the first syllable; the feet in which they do not coincide are read without change of stress, and this is indicated by a line over the entire foot, in place of the stress mark. The second line of markings represents quantity.

Stress

quantity Z. \j kj\ A. \j \j\ | 1^-^ w|jL_

1 . Arma virumque cano, Troiae qui primus ab oris Stress

quantity _w| |_ w \j | | _^w u |Z^

2. Italiam, fato profugus, Lavinaque venit.

1 Humphreys, Infltience of Accent in Latin Dactylic Hexam- eters, T. A. P. A., vol. ix. (1878), p. 39 ss. The percentages found by Professor Humphreys are for the fifth and sixth feet only.

2 The figures are, —

Bk. i. 1-50, 168 accords to 132 clashes

50-100, 182 accords to 118 clashes

100-150, 178 accords to 122 clashes

150-200, 177 accords to 123 clashes

Bk. vi. 236-285, 184 accords to 116 clashes

286-335, I9I accords to 109 clashes

Bk. xii. 670-720, 187 accords to 113 clashes

720-770, 176 accords to 124 clashes

84 THE STRESS ACCENT IN LATIN POETRY

Stress

quantity Zuv|Z _ | | \Zkjkj\Z.—

3. Litora, multum ille et terris iactatus et alto

Stress

quantity Zuu|_ _|_ w kj | | _£ w w | Z. _

4. Vi superum saevae memorera Iunonis ob iram, Stress —

quantity S_ w w | ^_ _ | | \ Z. uu|Z_

5 . Multa quoqueet bella passus, dum conderet urbem . Stress

quantity Z_ _ | Z. w kj \ _kj kj \ _ w w I — w w ! ^L_

6. Inferretque deos Latio, genus unde Latinum

Stress \

quantity /__ \Z_kj ^ I I Z^ _ UuvU_\

7. Albanique patres atque altae moenia Romae, Stress

quantity S_ kj kj\ I _ w w I I ZLkjkj\ ZL_

8. Musa, mihi causas memora, quo numine laeso, Stress

quantity Z. \j \j | |Zww| |£vv]Z_

9. Quidve dolens, regina deum tot volvere casus Stress

quantity |_v^u|Zvv| _ w w|_£wvy|Z._

10. Insignem pietate virum, tot adire labores Stress

quantity _ Uu|_ _ ! Z w w | |Zwu|Z_

11. Impulerit. Tantaene animis caelestibus irae? Stress

quantity /_ _| Zu w|_wv|_ww|Zuw| Z_

12. Urbs antiqua fuit, Tyrii tenuere coloni,

Stress

quantity | _ w w I l-wuUuwlZ_

13. Karthago, Italiam contra Tiberinaque longo

THE QUANTITATIVE METRES 85

Stress ^

quantity Z^w |ZWwl — w ul-- !- vw|Z_

14. Ostia, dives opum studiisque asperrima belli;

Stress

quantity Z _| 1 1— w wl Zw w I Z _

15. Quam Iuno fertur terris magis omnibus unam

Stress

quantity \j \j |_ \j\j\jL\j \j \ |Zww| Z

16. Posthabita coluisse Samo ; hie illius arma, Stress

quantity £_ _l_uu|Z_ |_ v; w I Z w w I Z vy

17. Hie currus fuit; hoc regnum dea gentibus esse*

Stress

quantity Z _| Zvu|_ _ | Z _ | Z \j \j\ Z w

18. Si qua fata sinant, iam turn tenditque fovetque.

Stress

quantity _ww|_w^|_ _|Z_| ZuuU-

19. Progenium sed enim Troiano a sanguine duci

Stress \

quantity _w^|_ww| 1 UuwU-

20. Audierat, Tyrias olim quae verteret arces ;

Stress

quantity Z wu| 1 | | Z ^ w I Z _

21. Hinc populum late regem belloque superbum

Stress 1 \

quantity 1 —^-^j w| |_ _|Z^w|Z_

22. Venturum exscidio Libyae : sic volvere Parcas. Stress

quantity Z uwl-wuUuul- _|Z^W|Z_

23. Id metuens veterisque memor Saturnia belli, Stress

quantity Z ^ \j |_ _l_ _|Z_|Z^w|Z_

24. Prima quod ad Troiam pro caris gesserat Argis

86 THE STRESS ACCENT IN LATIN POETRY

Stress quantity

25.

Stress quantity

26.

Stress

quantity

27.

Stress

quantity

28.

Stress

quantity

29.

Stress quantity

Stress quantity

31-

Stress

quantity

32.

Stress

quantity

33-

Z. w^_wl l_._|-_ -1-^ v/l^_

Nec dum etiam causae irarum saevique dolores

/

— ww I — wwl — t UU I- 1 — \J \j\ —

Exciderantanimo : manet alta mente repostum

_ww|_ ww| I Z_4/.._^| Z._

Iudicium Paridis spretaeque iniuria formae,

\jL _

IZ-

Et genus invistHH, et rapti Ganymedis honores ;

jL _| Z w w! I Z_| / w w i^_

His accensa super iactatos aequore toto

Z_|_ww|_w^l-/w_ I / wwl^_ Troas, relliquias Danatun atque inmitis Achilli,

I i_ww| IZ.W w |Z._

Arcebat longe Latio, multosque per annos

|_ _| |_ww|/wu|/_ x

Errabant, acti fatis, maria omnia circum.

Zw w| \Z.

W W I

Tantae rriolis erat Romanam condere gentem.

An examination of these lines proves beyond question that coincidence between word accent and quantity is not wholly dependent on the caesura, as Plessis thinks.1 For in lines 1, 7, 12, 15, 18,

1 Metrique grecque et latine, Paris, 1889, p. 32 ss.

THE QUANTITATIVE METRES 87

19, 26, 27, 28, 29, and 33, where the caesura falls after the long syllable of the third foot, the accords agree neither in number nor position. Lines 4, 5, 8, 15, and 21 show accords in the first, fifth, and sixth feet, but the caesura falls in very different parts of the line. So lines 2, 22, 26, and 32 re- semble each other in having accords only in the fifth and sixth feet, but this resemblance does not extend to the caesura ; and, while in the lines with five accords, 7, 14, 18, 29, and 33, it is always the third foot that is read with level stress, line 14 differs from the others in having the main caesura after the first foot. There are, in fact, no two successive lines that resemble each other both in the number and position of the accords, and in the position of the caesura, thus bringing to light another element in that greatest marvel of Vergil's metrical technique — its infinite variety. More- over, if for two or three successive lines more than the average number of feet are read with level stress, in the following lines less are read; but each time the climax is marked by a line containing five accords. Nor does it seem likely that the poet was unconscious of this method of avoiding mo- notony, when he uses it with such extraordinarily marked effect. In form, therefore, as well as in matter, Vergil is a thoroughly Latin poet. And just as he made the Trojan Aeneas an essential part pf Roman tradition, the founder of the State, Quires of the Quirites, so he naturalized the Greek

88 THE STRESS ACCENT IN LATIN POETRY

hexameter, subordinating, but not obliterating, stress, the fundamental principle of the native poetry.

In the passage quoted above, Quintilian speaks rather tentatively of sentence accentuation, as if it were a new doctrine and not the conventional teaching, iam quidam eruditi, quidam etiam gram- matici docent, he writes, and rnihi videtur, and in the very next sentence he returns to the vetus lex sermonis. " Namque in omni voce acuta intra numerum trium syllaborum continetur, sive eae sunt in verbo solae sive ultimae, et in eis aut proxima extremae, aut ab ea tertia. Trium porro de quibus loquor, media longa aut acuta aut flexa erit; eodem loco brevis utique gravem habebit solum, ideoque positam ante se, id est ab ultima tertiam acuet. Est autem in omni voce utique acuta, sed nunquam plus una, nee unquam ultima, ideoque in disyllabis prior. Praeterea nunquam in eadem flexa et acuta quoniam in flexa est acuta. : itaque neutra cludet vocem latinam." The same teaching is found in Bk. xii., 10. 33: "Sed accentus quoque, cum rigore quodam, turn simili- tudine ipsa minus suaves habemus ; quia ultima syllaba nee acuta umquam excitatur, nee flexa circumducitur, sed in gravem, vel duos graves cadit semper. Itaque tanto est sermo Graecus Latino iucundior, ut nostri poetae, quotiens dulce carmen esse voluerint illorum id nominibus exornent." It does not seem likely that they would wish so to

THE QUANTITATIVE METRES 89

"adorn " their verses, if the law of sentence accent had a very wide application. After Quintilian the subject is not infrequently mentioned by the gram- marians. Lindsay has collected the instances in Ch. hi. of his Latin Language ; p. 165 ss. Two words are thought of as forming a word group, with but a single accent. As, for instance, Pom- peius writes : " Quotienscumque duae partes ora- tionis in unam colliguntur, iam quoniam pro una sunt, unum accentum habebunt, prout fuerit syllaba ilia. Si dicas ' interea loci,' interea una pars ora- tionis est, loci una pars orationis est. Quando iam sic utramque dicis, ut pro una sint, ambae partes unum habebunt accentum. Ergo duae partes ora- tionis quando unam faciunt, necesse est ut unum accentum habeant." 1 On the next page, speaking again of separate words, he writes, "Ultima enim numquam habet [accentum] aut in versu aut in prosa." Schoell 2 quotes this last sentence from Pompeius and on the same page Consentius,3 his treatment of whom shows the same lack of direct- ness of which we have before had occasion to speak. Consentius is writing De Scandendis Ver- szbns, and Schoell begins his quotation : —

" Sine accusatione consistit versus huius modi : Conditus in nubem medioque refulserit orbe,4 et Tu quoque litoribus nostris Aeneia nutrix.5 Hi et tales non auctoritate aliqua praerogativa artis

1 v. p. 130, 22 K. 2 Op. cit. p. 27. 3 v. 398, 24 K.

4 Georg, i. 442. 5 ALn. vii. 1.

90 THE STRESS ACCENT IN LATIN POETRY

aut consuetudinis defenduntur ; nihil additum, nihil detractum, nihil mutatum habent, sed iuxta commu- nis linguae enuntiationem integri nati sunt, neque ulla ea parte aliquid dubitationis admittunt." On this he remarks, " Hoc dicere non potuit Consen- sus, nisi cum Bentleio 1 conditus in nubem et tit qtcoque litoribiis, pronuntiaret, non nubem ac litori- bus." Consentius goes on to say that the verses quoted are " without apology" because they con- tain no short syllable lengthened in arsis as (and he quotes)

"Emicat Eurya/kr et munere victor amici;" 2

no short vowel followed by a mute and liquid, as (he quotes) "et vol?/mim linguas," and " pecudes pictaeque voices' ' ;3 no lengthening or shortening like " X&nton me crimine dignum duxisti" ; or relli- quias Danaum," or Ennius's "<?batu Athenis," or " /talium " 4 (with long/) or "aquosus Orz<?n," or Horace's "feraeque s?/£'tae" and Lucan's dixisse Plwbus ; and finally no elisions. The section ends with a discussion of elision, nor is there a word in it, from beginning to end, about accent. With such misrepresentations as this, both at first and at second hand, it is small wonder that the

1 Sched. d. m. Ter. p. xix., ed. Lips.

2 jEn. v. 337.

3 Quintilian, 1. I., and Sergius, ad A en. i. 384, have a word to say on this point.

4 Mentioned also by Quintilian, op. cit. i. 5, 18.

THE QUANTITATIVE METRES 91

teachings of the grammarians should seem "cha- otic," as they are pronounced by Professor Bennett.

Now, the step from Troide qui to cano is not a long one, and it seems to me likely that it was taken by such uninspired declaimers as the Auctor Contra Paga?ias and by the ecclesiastical versifiers of the Middle Ages. For then Latin had become, if it had not always been, a stressed language, as is universally acknowledged, and this offers a per- fectly reasonable explanation for the utter disre- gard of word accent in the late hexameters.

Naturally the common people never saw the sense in this subordination of word stress to verse stress, and so in the soldiers' songs and other frag- ments of popular poetry that have come down to us, word and verse stress tend to coincide, as, for instance : —

1. " Gallias Caesar subegit, Nicomedes Caesarem

Ecce Caesar nunc triumphat, qui subegit Gallias Nicomedes non triumphat, qui subegit Caesarem."1

2. Urbani servate uxores, moechum calvum adducimus Aurum in Gallia effutuisti, hie sumpsisti mutuom.2

3. Gallos Caesar in triumphum ducit, idem in curiam Gallos bracas deposuerunt, latum clavum sumpserunt.3

4. Mille, mille, mille, mille, mille decollavimus

Unus homo, mille, mille, mille decollavimus Mille, mille, mille, mille, bibat qui mille occidit Tantum vini nemo habet quantum fudit sanguinis.4

1 Suetonius, life of Julius Caesar, c. 49. 3 Op. cit. ch. 80.

2 Op. cit. ch. 5L 4 Volpiscus, Life of Aurelian, ch. 6.

92 THE STRESS ACCENT IN LATIN POETRY

5. Mille Francos, mille semel Sarmatas occidimus Mille, mille, mille, mille, mille, Persas quaerimus.1

The same is true of the semi-popular poetry, which began to appear early in the second century after Christ, for example : — •

6. Floro2 poetae scribenti ad se

Ego nolo Caesar esse Ambulare per Britannos Latitare per Germanos Scythicos pati priunas rescripsit Hadrianus

Ego nolo Florus esse Ambulare per tabernas Latitare per popinas Culices pati rotundos.

7. The iambics quoted by Baehrens 3 from the Liber Ludicrorum of Apuleius : —

Calpurniane salve properis versibus ! Misi, ut petisti, mundicinas dentium, Nitelas oris ex Arabicis frugibus, Tenuem candificum nobilem pulvisculum, j Complanatorem tumidulae gingivulae, Converritorem pridianae reliquiae, Ne qua visatur tetra tabes sordium, Restrictis forte si labellis riseris.

8. The Pervigilium Veneris — in which not only are words and phrases repeated with a peculiarly

1 Op. cit. ch. 7. 3 Op. cit. p. 376.

2 Quoted by Spartianus, Vit. Hadr. 1 6.

THE QUANTITATIVE METRES 93

charming naivete, while assurance, "alliteration and even rhyme are found, but word accent and quan- tity tend all through the poem to coincide, and do entirely coincide in one-third of the lines. t Finally, in the Christian hymns, the versification is based, now on word accent, now on quantity — sometimes even on a simple count of the syllables, as it is in the GTiyoi ttoXltlkoi of the Byzantine writers — until in the end quantity was displaced, and accent alone determined the structure of the verse. This new kind of versification, which was, at the same time, oldest of all, is first clearly distinguished from the Classical versification based on quantity, by Marius Victorinus (4th century ). Although the words of Laberius1 a writer of Mimes who flourished 50 B.C., Versorum 11011 numerorwAi uu- mero studuimus, and of Quintilian,2 Pocma nemo dubitaverit imperito qnodcini initio fusum et auriutn mensura et similiter de curve ntiam spatiornm obser- vation esse generatum, seem to point to the same distinction. Marius Victorinus writes : 3 " Metro : quid videtur esse consimile ? Rhythmus. Rhy- thmus quid est ? Verborum modulata compositio non metrica ratione, sed numerosa scansione ad iudicium aurium examinata, ut puta veluti sunt cantica poetarum vulgarium. Rhythmus ergo in metro non est ? Potest esse. Quid ergo distat a me- tro ? Quod rhythmus per se sine metro esse potest,

1 Laberius, v. 55, ed. Ribb. (2). 3 vi. p. 206 K.

2 Inst. Qr. ix. 4, 114.

94 THE STRESS ACCENT IN LATIN POETRY

metrum sine rhythmo esse non potest1 Quod liquidius ita definitur, metrum est ratio cum modu- lation, rhythmus sine ratione metrica, modulatio. Plerumque tamen casu quodam etiam invenies rationem metricam in rhythmo, non artificii obser- vatione servata, sed sono et ipsa modulatione ducente." Diomedes' definition, rhythmus est versus imago modulata? is to the same effect, as is also what Servius says (of the Saturnian verse 3), carminibus Saturnio metro compositis, quod ad ryhythmum solum vulgares componere consuerunt. At the end of the fourth century, therefore, the dis- tinction was already thoroughly established.4 Later. Beda Venerabilis (died 672), enlarging on Victo- rinus's words, writes : 5 " Videtur autem rhythmus metris esse consimilis, quae est verborum modulata compositio, non metrica ratione, sed numero sylla- barum ad iudicium aurium examinata, ut sunt car- mina vulgarium poetarum ; quern (sc. rhythmum) vulgares poetae necesse est rustice, docti faciant docte. Quo-modo et ad instar iambici metri pul- cherrime factus est hymnus ille praeclarus : —

O rex aeterne domine

Rerum creator omnium.

Qui eras ante saecula

Semper cum patre filius. 6

1 This bears hard on Professor Bennett's theory of " quantitative prominence."

2 p. 470 K. 3 ad Georg. ii. 385.

4 Hiimer, Untersuchungen uber die dltesteif lateinish-echristlichen Rhythmen, p. 6 et ss. 6 vii. p. 258 K, P 5th century.

THE QUANTITATIVE METRES 95

et alii Ambrosiani non pauci. Item ad formam metri trochaic canunt hymnum de die iudicii alphabetum : —

Apparebit repentina

Dies magna domini

Fur obscura velut nocte

Improvisos occupans." In the Rex aeterne domine, the first beat in the second and fourth lines would seem to be deter- mined by quantity rather than by word accent. But, as Greenough1 has pointed out, in Christian poetry there were so many dissyllables, like Christe, Dens, Pater, Lueis, demanding naturally the first place, that the license came in of giving a trochaic accent to the first foot instead of an iambic, as in the modern hymn, —

From all that dwell below the skies Let the Creator's praise arise. It is also worthy of note that while the accentual Greek poetry was content with the coincidence of word and verse accent in the last foot of the line, or, in the longer lines, of each hemistich — which is true also of modern French poetry — in Latin, where the influence of stress is greater and more persistent, it is demanded in every foot.

Professor Greenough shows2 that Horace in his

1 "Accentual Rhythm in Latin," in Harvard Studies in Classical Philology, vol. iv. p. 113 ss.

2 /. /. ; cf. also Metres Lyriques d^ Horace, par O. Riemann, Paris, 1883, pp. 65 and 70.

g6 THE STRESS ACCENT IN LATIN POETRY

two favourite metres, the Sapphic and Alcaic, by observing the strong caesura after the fifth syllable — which is not done in Greek — makes word- and verse-accent coincide in a very large proportion of feet. Incidentally he remarks, "In a large num- ber of iambic verses taken consecutively from the remains of Ennius and Naevius, as they are given in Merry's collection, out of 1500 ictuses, only about 22 per cent fail to conform to the word accent, and this counting all cases of verbs com- pounded with prepositions, though it may well be that the preposition was at that time accented, and all cases of a dissyllable at the end of a verse, though the last verse ictus must have been very weak." Also that in a thousand verses of Seneca, the tragedian, there is "not one that cannot be read in the Christian fashion." As has been said, word- and verse-ictus tend to coincide in the dia- logue of Plautus and Terence, though the theory must not be pushed too far. "There is just so much disregard of accent as to produce what Ritschl happily calls the harmonische Disharmonie of Plau- tine verse." 1 So it would seem that there is no real break from the early accentual verses, like Mars pater te precor, to the Dies irae, dies ilia, of Chris- tian times, but that, from first to last, Stress is responsible, not only for the formation of words, but, in a greater or less degree, for the structure of verse as well.

1 Lindsay, Appendix to his edition of Plautus' Captivi, p. 372.

I

i