^{page 29 note 2} I cannot regard caesura as a mere ‘philological ghost’, with Sturtevant, E. H. (A.J.P. 1924, 329 ff.)Google Scholar, or agree with Eugene O'Neill, Jr, . (T.A.P.A. 1939, p. 293)Google Scholar that ‘the caesura may now be honourably retired’; but the matter is too important to be discussed adequately in a footnote, and will be treated in an article in Classical Philology.

^{page 29 note 3} Hickie, on Andoc. de Myst. 37Google Scholar.

^{page 29 note 4} Horn, . Od. iv. 597 f.Google Scholar; v. 131 f.; Soph, . Ant. 691Google Scholar; O.T. 1025; Eur, . H.F. 335Google Scholar; Ar, . Ach. 846 f.Google Scholar;Nub. 1410; Vesp. 238, 736 f.; Av. 56, 199 f., 1004; Lys. 365, 868 f.; Thesm. 212, 942; Lys. xii. 67; Xen. Anab. 1. viii. 11; Plat. Apol. 21c, 33c; Hyper. Adv. Dem. frg. ii (ed. Blass); Men, . Epitr. 125 (Jensen)Google Scholar.

^{page 29 note 5} For such inverted feet Enn, cf.. Ann. 222Google Scholar (ed. Baehrens), 2; 235, 1; 374; Heduphag. 529. 4, 10; Auson. IV. xxiv. 16 (ed. Peiper); and among the more literary inscriptions, Buecheler, and matsch, Lom, Carm. Lat. Epigr. 248. 5Google Scholar; 444. 2; 502. 3; 511. 4, 7; 518. 4; 650. 4; 1100. 1; 1105. 1; 1943. 2; 2068. 7.

^{page 29 note 6} Oral. 163: ‘inquinatus insuavissima littera.’ Quint, Cf.. Inst. Oral. XII. x. 29Google Scholar: ‘ilia quae sexta nostrarum paene non humana voce vel omnino non voce potius inter discrimina dentium efflanda est.’

^{page 30 note 1} Orat. 150: efficiet facile formulam. Of course any author might fall into this now and again. So the stylist Ovid, , who is not given to alliteration, has nerunt fatales fortia fila deae (Pont. i. viii. 64)Google Scholar and suffusis felle refugi (Trist. ii. 565); Livy, writes facilem futuram fugam (xxxv. xxx. 7)Google Scholar and Sallust nefaria facinora fecissent (Cat. li. 6)Careful search would doubtless provide other instances.

^{page 30 note 2} Pilsbury, , Latin Prose Composition, p. 30Google Scholar, n. 3.

^{page 30 note 3} Cf, . Rosc. Com. 4Google Scholar (essene); Verr. II. V.109 (hominene); N.D. i. 22 (varietatene), iii. 4 (responderene); Div. i. 53 (ipsene); Par. 12 (cogitassene).

^{page 30 note 4} Pilsbury, op. cit., p. 82.

^{page 30 note 5} III. XV. 3; Xliii.2; xlviii.8; Vll.ix.4; xxviii. 7; VIII. xvi. 12; IX. xxvi. 9; X. iii. 6; xxiii. 1; XXI. xxviii. 5; liv. 9; XXII xxxiii. 11; lix. 7; XXIII. xix. 3; XXIV. xviii. 10 (bis); XXVI. ii. 2; XXVII. v. 14; XXIX. xiv. 4; XXXIII. xv. 12; with a word (illique, ·plebi, cantu) intervening III. xxxv. 7; IV. xxxvi. 2; VI. xxxix. 5; IX. xxx. 8; XXVII. xxxiii. 6.

^{page 30 note 6} I cannot present a complete account of such collocations as exemplum dislurbandorum iudiriorum (Corn. I, frg. 17, ed. Miller), but a consideration of Cicero's gerunds and gerundives dependent on causa (and the less frequent gratia) will illustrate his usage. Merguet cites from the speeches and philosophical writings 59 such phrases involving gerunds; of these none shows an object, save a neuter pronoun twice (Fin. v. 54; N.D. iii. 1). Of the 99 phrases involving gerundives 9 present the ‘jingle’ (Fin. i. 36; Off. i. 157; iii. 45; frg. philos. v. 95; Verr. II. ii. 137; Caecin. 6; Leg. Agr. ii. 31; Cat. i. 7; Mur. 27). The letters of Cicero have gerunds dependent on causa 8 times, only once with an object (again a neuter pronoun), and gerundives in 21 instances, none, however, in the genitive plural. It is quite obvious from these facts that Cicero's practice was to use the gerund of intransitive verbs or of transitive verbs taking a neuter pronoun as object, otherwise the gerundive without reference to the sound.

^{page 30 note 7} Inst. Orat. VIII. iii. 50 f.: ‘…ταυτολογα, id est eiusdem verbi aut sermonis iteratio. Haec enim, quamquam non magnopere a summis auctoribus vitata, interim vitium videri potest, in quod saepe incidit etiam Cicero securus tarn parvae observationis.’ Ib. ix. iv. 41: ‘Videndum etiam ne syllabae verbi prioris ultimae sint primae sequentis. Id ne quis praecipi miretur, Ciceroni in epistolis excidit “res mihi invisae visae sunt, Brute ” et in carmine “0 fortunatam natam me consule Romam”.’ Cf. in general ib. ix. i. 11; iii. 66–86; iv. 42; x. i. 7; Auct, . ad Herennium iv. 18Google Scholar: ‘si vitabimus eiusdem litterae nimiam adsiduitatem’; ib. 32: ‘Haec tria proxima genera exornationum, quorum unum in similiter cadentibus, alterum in similiter desinentibus verbis, tertium in adnominationibus [puns etc.] positum est, perraro sumenda sunt.’

^{page 30 note 8} Off. i. 61.

^{page 30 note 9} Leg. Manil. 37.

^{page 30 note 10} Sest. 142.

^{page 30 note 11} Clu. 185.

^{page 30 note 12} Verr. 11. iv.88.

^{page 30 note 13} Sest. 93.

^{page 30 note 14} Fin. v. 44.

^{page 30 note 15} Ib.

^{page 30 note 16} Tusc. i. 116.

^{page 30 note 17} De Orat. i. 88.

^{page 30 note 18} Quinct. 11.

^{page 30 note 19} Tusci. i. 41.

^{page 30 note 20} Phil. vi. 1.

^{page 30 note 21} As he does Verr. 11. i. 71; v. 161; Clu. 75; Cat. iii. 3; Flacc. 53; Har. resp. 38; Mil. 46; Off. iii. 80; Ac. i. 8; Fin. v. 15; Rep. ii. 6.

^{page 30 note 22} Cic, Cf.. Orat. 163Google Scholar: ‘verba…legenda sunt potissimum bene sonantia, sed ea non ut poetae exquisita ad sonum sed sumpta de medio’; ib. 149: ‘quam suavissimis vocibus’.

^{page 31 note 1} Such lines as μιμντω αὐτθι τος πειγμενς περ ' Aρηος (II. xix. 189) celsa petebas (Sil. Ital. Pun. xvii. 326 f.) appear only sporadically.

^{page 31 note 2} If the Helenae Encomium is not genuine it is at least a reflection of Gorgias' style.

^{page 31 note 3} This is quite clear from Cicero's discussion Oral. 164–7.

^{page 31 note 4} v 993.

^{page 31 note 5} 309, 319 f.

^{page 31 note 6} De Sen. 55. Cf. Mil. 30: ‘vi victa vis vel potius oppressa virtute audacia est’; 78: ‘evomere videar quam verius’.

^{page 31 note 7} Met. xv. 878 f. (repeated almost verbatim Tr. IV. x. 129); cf, . also Her. xi. 11Google Scholar; xviii. 144 for alliteration in v connected with other sentiments.

^{page 31 note 8} Cic, . Tusc. i. 34Google Scholar.

^{page 31 note 9} Serv, Cf.. on Aen. iii. 183Google Scholar (casus Cassandra canebat): ‘Haec compositio iam vitiosa est, quae maioribus placuit, ut “Anchisen agnovit amicum” [Aen. iii. 82] et “sale saxa sonabant”’ [v. 866]. Cf, . also Verg. Georg. ii. 294Google Scholar f.(of the sturdy aesculus): ‘immota manet multosque nepotes, multa virum volvens durando saecula vincit,’ etc.

^{page 31 note 10} Alliteration in v, e.g. in Lucretius (ed. Bailey): i. 72, 202, 271; ii. 964; iii. 494, 1046; iv. 675, 1071; v. 217, 266, 931 f., 957, 1226; vi. 114 f., 137, 541,1140.

^{page 32 note 1} p. 157.

^{page 32 note 2} p. 195.

^{page 32 note 3} p. 182.

^{page 32 note 4} p. 213.

^{page 32 note 5} p. 219.

^{page 32 note 6} p. 270.

^{page 32 note 7} p. 182.

^{page 32 note 8} O.T. 371.

^{page 32 note 9} Blaydes: ‘The “tauismus”, if I may use the expression, of this line is observable, being perhaps designedly used by the poet, with reference to the initial letter of Tiresias’ name, to heighten the sarcasm by the parechesis, in the same way as Agamemnon, according to some, taunts Calchas, Mντι κακ7nu;, οὐ πώπογ μοι τκρἠγυον εἷπας. Muretus long since (V.L. i. 15) compared the verse of Ennius, “0 Tite, tute Tati tibi tanta, tyranne, tulisti”’, a verse which the Auctor ad Herennium (iv. 18) regards merely as bad writing.

^{page 32 note 10} Campbell.

^{page 32 note 11} The editors fail to be struck by the hammer ing of kappa in Soph. O.T. 1262:

ἔκλινε κοῖλα κλῇθρα κμππτει στγῃ.

^{page 32 note 12} viii. 69; xxii. 209.

^{page 32 note 13} Cho. 511.

^{page 32 note 14} 687 ff.

^{page 33 note 1} Frg. 360, 14 (Nauck²).

^{page 33 note 2} Theaet. 151A.

^{page 33 note 3} Pol. 293E.

^{page 33 note 4} Phil. 26E.

^{page 33 note 5} Leg. 805A.

^{page 33 note 6} Leg. 845D.

^{page 33 note 7} 68E.

^{page 33 note 8} 1444^b 29 f.

^{page 33 note 9} I445^b 22 f.

^{page 33 note 10} Pro Eux. xxv. 12 f. (Blass).

^{page 33 note 11} ix. 44.

^{page 33 note 12} ix. 43.

^{page 33 note 13} Av. 1164 f.

^{page 33 note 14} Thesm. 845.

^{page 33 note 15} Frg. 301, 6 (Kock).

^{page 33 note 16} Adesp. 602 (Kock).

^{page 34 note 1} i. 179 f.

^{page 34 note 2} Pharr.

^{page 34 note 3} Il. vi. 269 f.

^{page 34 note 4} Il. vi. 484.

^{page 34 note 5} 62.

^{page 34 note 6} Dem. xviii. 265; xix. 337; xxi. 226.

^{page 34 note 7} In all the figures used in this paper are included not onlyσ but ζ, ξ, and ψ; and σσ is counted as two sigmas.

^{page 34 note 8} 2, 15, 58, 61, 62, 65, 68, 139, 141, 146, 161, 164, 172, 199, 200, 251, 282, 290, 293, 298, 314, 328, 331, instead of σσ and relieving the hearers' ears of Euripides' sigmas in this way. Actually this change in pronunciation has little to do with the matter; in the samples of tragedy used in this paper (see notes 5, 6, and 7 on this page) the use of ττ would relieve Aeschylus of only 26 of his 1,699 or 1,700 sigmatic sounds, Sophocles of 32 out of his 1,649, and Euripides 36 of his 2,115, i.e., of c. 4·3, 5·3, and 5 respectively per 100 lines; conversely, the use of σσ for ττ in the passages taken from Plato Comicus, Eubulus, and Aristo phanes would raise their figures respectively by 14·3, 4·36, and 6·5 per 100 lines. In neither case would the gap between tragic and comic usage be bridged over, though it would nearly be so in the case of Plato Comicus, where the sample is small and the poet happens in four fragments to be harping on κ τταβος and γλττα.

^{page 35 note 1} Dionysius frg. 27 (Kock):

κα τοῖς μοῖσιν γγελσι πἠμασιν

τ σῖγα συλλξαντες ὡς αὐτο σοφο.

^{page 35 note 2} Eur. Med. 476: ἔσωσ σ' ὡς ἴσασιν 'ελλἠνων ὅσοι.

^{page 35 not 3} πλεονζει στχος τῷ σῖγμα.

^{page 35 note 4} I have merely assured that they shall be fairly scattered and exhibit different types of emotion.

^{page 35 note 5} Pers. 290–489; P.V. 936–1035; Sept. 369–749 (omitting the lyrics); Ag. 581–680; Eum, 566–665. The sigmatic sounds total respectively 309, 280 (or 281), 294, 263, 269, 284,=1,699 or 1,700.

^{page 35 note 6} Antig. 681–780, 991–1090; O.T. 1–100, 345–444, 513–612; O.C. 728–827. The totals here are respectively 268, 262, 281, 295, 249, 294,=1,649.

^{page 35 note 7} Alc. 606–740, 1008–36; Med. 1–95, 446–545, 1008–80 (omitting 1077);HF. 1–100, 248–347; Hel. 711–810. Here the totals are respectively 387, 89, 265, 312, 190, 287, 291, 294, = 2,115 for 731 verses.

^{page 35 note 8} Kock notes the Antiope, the Auge, the Bellerophon, and the Oedipus; and one is tempted to add the Ion, the Medea, and the Oenomaos. ‘εορτα, frg. 30.). One might note also that Lasos of Hermione went to the trouble of composing an ᾠδἠ ἄσιγμος (cf. Athen. x. 455c).

^{page 36 note 1} Frg. 134 (Kock): ταῖς ξυστσιν ταῖς χρυσοπστοις στρνυται, a line as innocent of any emotion as well could be.

^{page 36 note 2} Ach. 860–1066 (om. lyrics); Nub. 1214–1334 (om. lyrics); Av. 1–100; Plut. 322–421.

^{page 36 note 3} Grand average for tragic samples is 283 per 100 lines; for Ar., 233; for Pl. and Eub. combined, 248; for comic samples all told, 241.

^{page 36 note 4} Sept. 369–479 (om. lyrics).

^{page 36 note 5} Pers. 290–389.

^{page 36 note 6} O.T. 513–612; Antig. 991–1090.

^{page 36 note 7} Med. 446–545.

^{page 36 note 8} 309 vs. 317 per 100 lines.

^{page 36 note 9} 307 per 100 lines.

^{page 36 note 10} Alc. 1008–36.

^{page 36 note 11} 606–740.

^{page 36 note 12} 287 per 100 lines.

For note 13 see opposite.)

^{page 36 note 14} 936–1035.

^{page 36 note 15} 937–43:

σβου, προσεχου, θπτpε τν κρατο^ντ' ε.

ἐμοἴ δ' ἔλασσον ζηνς ἢ μηδν μλει.

δρτω, κρατετω τνδε τν βραχὺν χρνον

ὅπως θλει. δαρν γρ οὐκ ἄρξει θεοῖς.

λλ' εςορ γρ τνσε τν δις τρχιν,

τν το^ τυρpννου το^ νου δικονον

πντως τι καινν γγελν λἠλυθεν.

^{page 37 note 1} 966 f.: τς ςς λατρεας τν μν δυσπραξαν, σαφς πστασ', οὐκ ἂν λλξαιμ' γώ.

^{page 37 note 2} 972 f.: χλιδ; χλιδντας ὧδε τοὺς μοὺς γὼ χθροὺς ἴδοιμι κα σ δ' ν τοτοις λγω.

^{page 37 note 3} 975 f.: πλῷ τοὺσ πντας χθαρω θεοὺςὅσοι παθντες εὖ κακο^σ μ' κδκως.

^{page 37 note 4} 979: εἴης φορητς οὐκ ἄν ε πρσσοις καλς.

^{page 37 note 5} 987 f.: οὐ γρ σὺ παῖς τε κἄτι το^δ' νοστερος ε προςδοκᾷς μο^ τι πεσεσθαι πρα;

^{page 38 note 1} 965: ς τ

^{page 38 note 2} 1000: πρ τς παρούσας πημονς ρθς φρονεῖν.

^{page 38 note 3} 1014: σκέψαι δ’, ν μ τοῖς πεις πεισθῇς λόγοις.

^{page 38 note 4} 1021: ἂψορρον ἢξεις εἰς είς φάος. Δις. δέ σοι.

^{page 38 note 5} 681–780.

^{page 38 note 6} 739: καλς ρήμης γ’ ᾱν σὐ γς ᾰρφοις μόνος.

^{page 38 note 7} 742: ᾥ παγκάκιοτε, δι δικης ίὼν πατρί.

^{page 38 note 8} 746: ὧ μιαρòν ᾖθος καί γυναικòς ὒστερον.

^{page 38 note 9} 756: γυναικòς ὦν δούλευμα, μή κώτιλλέ με.

^{page 38 note 10} 767: νος δ’ έστì τηλικοτος λγήσας βαρύς.

^{page 38 note 11} 780: πόνος περισσός οτι τν “Αιδου σέβειν.

^{page 38 note 12} 991–1090.

^{page 38 note 13} 997: τί δ’; ὡγώ τ òσν φρισσω στόμα.

^{page 38 note 14} 1052: ταύτης σù μέντοι τς νόσου πλήρης ἒφυς.

^{page 38 note 15} 1033–43, 1055–7, 1059, 1061.

^{page 38 note 16} O.T.425.

^{page 38 note 17} 316–462(of which 345–444 are included in our examples).

^{page 38 note 18} 350–3, 362, 364, 371, 374, 378, 387–9, 395–8, 410, 413, 425, 430, 440.

^{page 38 note 19} 425: ἂ σ’ ξισώσει σοί τε καì τοῖς σοῖς τέκνοις.

^{page 38 note 20} 387–9.

^{page 38 note 21} 395–8

^{page 38 note 22} 371.

^{page 38 note 23} 413.

^{page 39 note 1} , Soph.O.T. 513–612Google Scholar, where the strongly hissing sigmatic lines are 526, 532, 533, 535, 610, and the ones particularly expressive of hatred and contempt are 532–42, 548, 551 f., involving lines of 6, 7, 3, 6, 1, 1, 2, 2, 2, 1, 2, 3, 4, and 2 sigmatic sounds; O.C. 728–827, where the heavily sigmatic lines are 744 and 789, while the lines expressing most contempt and hatred are 761 f. and 770 f., embracing respectively 1, o, 5, 3 σ , Eur.Med. 446–545Google Scholar, where the hissing lines are 452,467, 476, 481, 5I5, 531, 538–40, the contemptuous and baleful ones are 465–72,488–90,492–4, 510 f., 514 f., including (in addition to two lines of 6 and 7 sigmas) lines exhibiting 2, 4, 1, 3, 3, 3, 3, 1, 3, 2, 2, 4, 3, 3, 2, and 2 sigmatic sounds; Alc. 606–740, where the markedly sigmatic lines are 620, 625, 697, 717, 731, the ones expressing hatred and contempt are 632–46, 665, 696–702, 710, 713 f., 716 f., 720, 728–38, including 3 lines of 6 or 7 sigmas, and others exhibiting respectively 3, 2, 4, 1, 4, 2, 1, 4, 5, 2, 2, 2, 4, 3, 0; 2; 2, 1, 4, 3, 4, 4; 3; 4, 1; 3; 5; 2, 1, 5, 4, 3, 5, 4, 4, 1, and 2.

^{page 39 note 3} It was only in the period of decadence that verbiage dominated subject matter, so that Melanthius(, Plut.Mor. 41D)Google Scholar could complain of a production by the tragic poet Diogenes that he could not see his play for the words:… οὐκ ἒθηκατιδεῖν αὐτήν ὑπò τν ὑνομάτων èπιπροθουμένην.

^{page 39 note 4} In, Aesch.Cho. 563 fGoogle Scholar. Orestes promises that he and Pylades will speak in the Phocian dialect; but when the time comes (653 ff.), their language is no different from that of the other characters in the play.