the value of O. Lat. meo by three pronunciations of the English word immediate, which are all in use at the present time: (1) Careful speakers make it a word of four syllables, viz. immē' di-āt. (2) Careless or unlearned people reduce it to three, viz. immē'jut. (3) Between these two extremes there is a pronunciation immē'je-ut, in which the syllable je is greatly weakened in pronunciation, and the word is reduced almost to a trisyllable; O. Lat. meo may be best compared with this third or intermediate pronunciation. Professor Fay (Most. xiii) also aptly compares the dissyllabic pronunciation of bounteous, plenteous, radiant, happier, and the like, in our current hymns. In conclusion, I wish to quote the admirable account of synizesis, which is given by Spengel (Einl. zu Andria, xxxi): "Die Zusammenziehung zweier Vokale innerhalb eines Wortes beruht auf Unterordnung des kurzen Vokals unter den folgenden langen, wodurch beide éin Zeitmass bilden, wie in der Musik ein kurzer Vorschlag durch das Anlehnen an eine lange Note die selbständige Messung im Takt verliert. Um das einsilbige tuis, deos, u. a. richtig auszusprechen, dürfen wir nicht auf die erste Silbe den Ton legen, noch auch beide Silben gleichmässig betonen, sondern der möglichst kurz gesprochene erste Vokal lehnt sich als Vorschlag an den lang und voll gesprochenen Hauptvokal

an." 1

abiete." Similarly Wagner in his Aulularia (p. 62) and Ashmore in his Adelphoe (p. lxvii) state that the synizesis forms are to be pronounced myîs, dyô, dyes, dye, dyus, yōs, fwisse, etc. Even Lindsay writes somewhat loosely (L.L. 439), "When e is followed by a long syllable, it passed in unaccented usage intoy by synizesis, e.g. eo, eos." (Lindsay, however, expresses the view elsewhere (L.L. 22) that the double forms vinea and vinia, balteus and baltius, etc., are to be explained by the tendency to give a vowel in hiatus the close sound rather than by the tendency to change i and e in hiatus into consonantal i (y).) On the other hand, Professor Fay (Most. xiii) aptly describes synizesis as "quasi-elision between impinging vowels of the same word"; and little exception can be taken to the definition given in Allen and Greenough's New Latin Grammar (§ 642), "the running together of two vowels without full contraction." With Fay's description, compare Consentius (Keil, V, 403), who cites as an example of a vowel elided between a consonant and a vowel Vergilian d(u)odena (Geor. i, 232), which was, however, probably read by Vergil duodena (cf. p. 199). 1 Spengel (1.7., xxxii) also compares the Romance derivatives, e.g. according to his view: Ital. so from sc(i)o; Span. dos from d(u)o, d(u)os; Ital. fosse from

II. RELATION OF SYNIZESIS TO IAMBIC SHORTENING. LAW OF BREVIS COALESCENS.

Points of Similarity. The Roman Quantitative Problem.— Many Plautine scholars have refuted C. F. W. Müller's attempted substitution of iambic shortening for synizesis in the case-forms meo, tuo, deo, and the like, and have shown the extreme improbability of this hypothesis. In the present section I shall attempt to offer additional proof of this conclusion, but it seems desirable first to point out that those critics who have advocated iambic shortening in these forms have directed attention to an important and a curious series of facts which their opponents had entirely overlooked. It is to this somewhat vague treatment of synizesis on the part of its professed advocates, and especially to their neglect to inquire into the ultimate cause of the phenomena in question, that the vitality of Müller's theory seems largely due,1 and it therefore becomes important to explain clearly the problem of synizesis and to interpret the real meaning of the phenomena before we attempt a formal refutation of iambic shortening. Our first inquiry must be whether the great body of Plautine synizesis forms can be reduced to a law; for it is no real solution whatever to say, as is usually said, that synizesis occurs in "a limited circle of words and word-forms" (Brix, Einl. zu Trin. 20), or that it occurs in "certain words in very common use" (Fairclough, Intr. to Andria, lxi). The possibility of slurring the half-vowels i (e) and u in pronunciation being given (p. 170), we are called upon to explain

dissyllabic fuisse; Ital. di from monosyllabic die, dies, etc. On the pronunciation cf. also Bömer, l.l., 43: "Duas vocales pronuntiatas esse . . . ita, ut utriusque sonus audiretur," and Leppermann, .., 10: "Duae vocales una syllaba simul pronuntientur sed non in diphthongum coniunctae." Corssen also (II2, 752) aptly describes the half-vowel u in synizesis as 'ein zwischen v und u schwebender Mittellaut von unmessbar kurzer Dauer,' and in his general discussion (II2, 744 ff.) he makes very effective use at times of the rhythmical doctrine of the 'verschwindend kurzer Vocal.' If at other times he has clearly misused and perverted this doctrine, the underlying conception is none the less one of real value for the student of ancient metric.

1 Cf. Stolz's somewhat hasty acceptance of this theory of Müller and Skutsch, Müller's Handb. II3, 2, 34; cf. also Gleditsch, Metr.3 258.

why this slurring occurs in the dialogue metres in meō, deūm, eūm, eã, die, diū, but does not occur in aúred, fílið, grátis, linteùm, pósteà, ánteà, pridie, interdius;1 why it occurs in scio, fui, eat, eămus, eodem, but never in nésciò,2 ádfuì, éxeat, tránseàt, pródeat, èxěámus, èxeúntem, èxeúndum, intr(o)eúndi, etc. ;3 why it occurs in fuerunt, pužlla, puēr,

1 Ahlberg seems to think of this problem, when he writes of deus (Procel. I, 95): "Quoniam hoc subst. non nisi in certis collocationibus verborum encliticam vim habet, . . . synizesis non adest in versibus his." Cf. below, pp. 177, 193 ff.

21, n. I.

2 Except in the peculiar locutions nescio quis, nescio quid, etc. Of the two scansions which are metrically possible in such cases, viz. nesc(i)ó-quis and nescio-quis, the former only should probably be admitted for O. Lat., but whether we assume the process of shortening or of synizesis, the task of explaining its applicability to the trisyllabic word nescio is equally difficult in either case. In discussing these locutions, it is necessary to assume that the analogy of the simple verb scio has in some way made itself felt, probably in the first case through the interrogative locutions sc(i)ō-quis and scin-quis; for the latter, which is usually overlooked, see Schrader's note and citations, De -ne prosodia, Luchs's study (Hermes, VI, 264 f.) of the indefinite locution nescioquis and its peculiar metrical treatment is well known, but Schrader's note shows that he is mistaken in entirely rejecting, and that the editors are mistaken in emending, the occasional occurrence of the peculiar scansion nesc(i)ô-quis in the case of the interrogative locution also, viz. Ba. 795; Tri. 880; Pacuv. tr. fr. 294; And. 734 (accepted by Spengel, Einl. xxxii). Finally, Luchs denies that the indefinite locution ever has the full scansion nescio-quis, but even this conclusion may perhaps be doubted, v. Seyffert, Bursian's Jahresb. 1894, 262, who accepts the reading of B, néscio-quem, in Men. 407 (marked corrupt in ed. min.). In O. Lat. a rigorous distinction cannot be drawn between the interrogative and the indefinite locutions. - Since writing the above, I have become convinced that the assumption of synizesis in nesc(i)õquis is improbable, and that the usual value of the indefinite is much rather nescioquis, with shortening due to the addition of the enclitic quis (cf. Stolz, Müller's Handb. II, 2, 55; Skutsch, Tépas, 136). The indefinite has, however, occasionally the original value nesciɔ̃quis (cf. siquidem and sīquidem), while the interrogative has sometimes within the verse the same value as the indefinite, viz. nescio-quis.

3 A single apparent exception occurs in the case of the compound perduellis, which is always trisyllabic in Pl., but in this case it is evident that the synizesis of the simple form duellum has become so regular as finally to create a new word (duello only Am. 189, cf. Müller, Pl. Pr. 236, 264); and Müller (Z., 237) is undoubtedly correct in his view of this particular word, viz. that it does not involve the coalescence of two vowels, but was pronounced in O. Lat. dvellum (later bellum). The locution nescio-quis, which may also seem an excep. tion, has already been discussed. The isolated quattuor of Mo. 630 (quattuor, ed. min.; quattor, Schöll in ed. mai., cf. quatt(u)or Enn. A. 90, 609 M.) does not concern us closely; for it occurs in the first foot, which may have the free

1

1

eāmus, eā, eūm, eō, cf. ais (ain), but never in fuerint, puĕri (cf. Ritschl, Proleg. clxiv), ăběamus, rěděamus, intěrěa, sěd-ĕum, ǎd-ěum, ǎb-ěo, quid-ăis.1 These marked differences in treatment appear at first sight somewhat perplexing and difficult of solution, but I am indebted to the kindness of my colleague, Professor M. Anstice Harris, for the valuable suggestion that the explanation must be sought in the preference of the Roman language or of all the Roman γένος διπλάσιον metres for a particular rhythm.2 On the whole, much might be said in favor of the thesis that the fierce aggressiveness, the sharp energy of the Roman national character, which created a thoroughgoing system of regressive accentuation, also inclined the Romans to prefer rhythms like the trochaic, which begin with the arsis; cf. above, p. 168, and Christ, Metr.2 208 f.3 Yet the principle of the preferred rhythm has in Latin a simpler and a more obvious application in the well

1

dom of the yévos loov, cf. Ba. 1204 filii (anap. sept.). Much less do the reduced forms of hūius, illīus, etc., belong here; for these were probably huis, illīs (Luchs, Studem. Stud. I, 319 ff). Ritschl (Proleg. clxiv, n.) admitted synizesis in antea and postea, but the few and doubtful examples have long ago been corrected; hence it is unnecessary to explain anteă and posted as separable composita in these cases, i.e. ante eã, post eã. The separable prepositional composita of the value — ~ —, e.g. a-suo, ex-tuis, ad-suam, ex-ea, etc., when they stand detached from a substantive, are more frequently full forms, i.e. dsuò, éxtuìs, and similar to the verbal composita éxeò, nésciò. Since, however, prepositional composita of this kind retain many of the characteristics of their component parts, they also freely exhibit the contract forms, but by no means so frequently as the simple proclitics suō, suām, tuis, etc.

1 Havet, De Saturnio, 32, quotes quid ais from the verse-close, Eu. 654, but the verse is one which all the editors agree in regarding as corrupt and emend in some form (Fleck. and Dzi. : qui3d äis).

2 My indebtedness for helpful suggestion and aid is much greater than appears in this brief acknowledgment, and if I have been successful in avoiding at this important point the somewhat narrow traditional view of the phenomena under discussion, it is largely due to the fact that I have been generously assisted by Professor Harris's critical judgment and clear insight into the fundamental principles of rhythm. For the influence of the preferred rhythm upon English word-forms in the loss or retention of weak syllables, see Hempl's study "Learned and Learn'd," Publ. Mod. Lang. Assoc. XII, 318 ff.

3 According to Roppenecker, De emend. cantic. Plaut., Freising, 1894, p. 19, the Latin language in the time of Plautus was less suited to ascending rhythms, i.e. the iambic and especially the anapaestic.

known difficulty which the Romans experienced both in their prose and in their poetry in pronouncing an iambic word or iambic word-beginning, when the latter occurs in certain combinations that frequently arise in the continuous sequence of the sentence. Fortunately, it is not necessary to investigate afresh those combinations of syllables which the early Romans pronounced with difficulty; they are precisely those which lead to the well-known phenomena of iambic shortening, and they have therefore often been carefully investigated and conveniently tabulated. I need only remark that the difficulty of pronouncing these syllable-groups was connected in ordinary or unrhythmical prose with a certain position of the word-accent, and numerous modifications in word-forms were thus assisted by word-accentuation; on the other hand, in rhythmical prose and in poetry, where the word-accent was replaced by a metrical accent (largely similar in kind, cf. Trans. Am. Phil. Assoc. XXXV, 52), changes in word-forms were produced also by the latter. Hence poetry has at its disposal both series of modified word-forms, but such is the constitution of iambic and trochaic verse that the effects of the rhythmical accent are usually only a continuation of the effects which the word-accent tends to bring about in common speech, and only rarely run counter to the latter, e.g. the pretonic syncope seen in m(e)um gnátum may be produced equally well by the rhythmical accentuation m(e)úm gnatúm. In the present discussion, however, the legitimate operation of both accents will be recognized.

The difficult quantitative combinations which the Latin language seeks to escape by the use of iambic shortening are conveniently enumerated by Dziatzko-Hauler (Einl. zu Phormio, 51) as follows: (1) ≤; (2) u_úu; (3) It is universally recognized that in all these sequences it is the presence of the short syllable which constitutes the disturbing element, and that the pronunciation of this syllable, eg. in the sequence, makes it difficult to give the immediately following long its proper value, when the voice is hastening to the second and accented long; hence this short syllable is properly called a Brevis Brevians,