¹ The article keeps virtually silent on Cicero's Greek or Latin predecessors or successors. (It may be mentioned in passing that the considerable literature on Seneca's rhythm contains much of interest to the student of Cicero.) The notes occasionally refer to Cicero's contemporaries. It should be noted that | will be used to mark what I consider rhythmic closes, sometimes in anticipation of the argument; but not every such close will invariably be indicated. In § 5 below, is used to mark the possibility of overlap between rhythmic units: is used where the first close may occur, | where the second may.

² It should not be thought that two of our examples are special cases because the consonant is c. Cicero avoids ac before c, but his use of atque before it is essentially confined, as with other consonants, to rhythmic closes (otherwise he can use et). He may be variously differentiated from Nepos, who even writes ‘simul ac conspexit’ (Pel. 5.3), from Varro, who avoids atque before other consonants but writes ‘atque commutatis’ (L. 6.38, cf. Plancus and Cato, Fam. 10.24.5, 15.5.2), and from such later rhythmic writers as Quintilian, who is freer in using atque unrhythmically before c (Winterbottom, M., CR 42 [1992], 449Google Scholar).

³ In the last two quotations atque even appears (despite Orat. 233!) where ac would have given rhythm 4. For statistics on rhythms, over different ranges of text, see esp. Th. Zielinski, , Das Clauselgesetz in Ciceros Reden (Philologus Supp. 9.4 [1904], 589–844)Google Scholar, and also Müller, E., De numero Ciceroniano (Diss. Kiel, 1887), 11–30Google Scholar, Wolff, J., De clausulis Ciceronianis (Jahrb.f. cl. Phil. Supp. 26 [1901]), 593–9Google Scholar, Zielinski, , Der constructive Rhythmus in Ciceros Reden (Philologus Supp. 13.1 [1914]), esp. 3–10 and tables 1–2Google Scholar, Broadhead, H. D., Latin Prose Rhythm (Cambridge, 1922), Ch. 6Google Scholar, Primmer, A., Cicero numerosus. Studien zum antiken Prosarhythmus (SAWW no. 257, 1968), esp. 161–72Google Scholar, Aili, H., The Prose Rhythm of Sallust and Livy (Stockholm, 1979), Ch. 3Google Scholar.

⁴ Zielinski, , D. constr. Rhythmus (n. 3), 4Google Scholar.

⁵ ‘Necessarily’, because the evidence strongly encourages the view that no consonant combinations lengthen a preceding final syllable. (Zielinski's, discussion, D. Clauselgesetz (n. 3), 762f.Google Scholar, is a surprising aberration.)

⁶ Leseproben aus Reden Ciceros and Catos (Rome, 1968), 169–71Google Scholar. None of Caelius' letters is rhythmical. For Cicero on the relation between oratorical rhythm and poetic metre cf., besides the Orator, De Oral. 1.70, 151, 3.175f., Brut. 32, Part. 72.

⁷ ‘Cola and clausulae in Cicero's speeches’, in Craik, E. M. (ed.), Owls to Athens [Festschr. Dover] (Oxford, 1990), 349–59Google Scholar.

⁸ Cicero's, remarks on hiatus, Orat. 150–52Google Scholar, must be borne in mind when we postulate it; but the considerations in § 5 below may complicate the matter.

⁹ Wolff (n. 3) discusses a selection, 637–40.

¹⁰ Att. 1.20.1 (?) and 2 (supported by e.g. Fam. 15.14.4, Agr. 2.15) occur in thoroughly rhythmic periods; Att. 5.13.1 ‘praefulci atque praemuni’ (contrast 5.17.5) struck me so forcibly as grandiose in phrasing that it first made me see the significance of atque for this subject. Cicero's correspondents also avoid atque + consonant, save before c or in simul atque (nn. 2, 11). It occurs a fair number of times in Caesar and his continuators (Gal. 1.2.5, 34.3, 2.6.2, 3.8.2, 4.1.8, 2.2, 3.3, 25.3, etc.)

¹¹ So even in the fifth-century Turin palimpsest at Scaur. 50, where, the consonant apart, atque is used in an improbable position, and is obviously caused by the atque three words before. The unrhythmical simul atque at Fin. 4.34 (for example) is supported by the similar phrases at 1.30, 2.31, 3.16. For such uses of atque + consonant in authors who otherwise avoid it, cf. e.g. Var. R. 2.3.5 ‘aliter … atque fit’, Ant. Div. fr. 107 Cardauns (cit. Gellius) ‘simul atque parti sunt’, and note ‘simul atque Dolabella’ in Caelius, Fam. 8.6.1.

¹² ‘Quam anxie Cicero clausulas quaesiverit’, Wolff, (n. 3) begins, 637Google Scholar.

¹³ Of the speeches only Q. Rose, Font., and Phil. 1 have overall a lower percentage than 7.5, and they are as preserved much shorter works. The percentages given here and subsequently use Muller's (and Friedrich's) pages as those of a complete and accessible edition; the precise figures should not be trusted too implicitly. I include only the instances I think probably sound; but one will in any case be at the mercy sometimes of antiquated information. In the very first speech, the evidence for atque at Quinct. 48 looks far weaker now that we have Reeve's, M. D. edition (M. Tullius Cicero, Oratio pro P. Quinctio [Stuttgart and Leipzig, 1992], 22, 51)Google Scholar.

¹⁴ See Zielinski, , D. Clauselgesetz (n. 3), 773f.Google Scholar; Fraenkel (n. 6), 136, is led astray by his loose criteria. I have noted fifty-seven places where fieri, etc., gives a better rhythm; there are certainly more. I have found only fourteen where fieri, etc., gives a better rhythm (including Corn. 1 fr. 34 Crawford: ‘fieri oportere’, as in half the examples). The scansion is archaic even for Plautus and Terence, who use it mostly at line-end, e.g. Bacch. 299, Poen. 788, Ph. 593; see Skutsch, on Enn, . Ann. 11 SkGoogle Scholar.

¹⁵ Cf. Davies, J. C., ‘Phrasal abundantia in Cicero's speeches’, CQ 18 (1968), 142–9CrossRefGoogle Scholar. Such pairs, some recurring, are common even in Caesar, and his continuators. In the latter cf. e.g. ‘disciplinam seueritatemque’ (B. Alex. 48.3, cf. 65.1), ‘Ienitatem et clementiam’ (B. Afr. 86.2, cf. 92.4).

¹⁶ Div. Caec. 6 (here plainly an almost self-contained rhythmic unit), Ver. 7, Ver. 2.2, 3.151, 4.83; related phrases without alliteration 1.82, 3.111, 155, 4.30, 5.1.

¹⁷ The phrase itself is naturally a well-worn one: note Cicero's remark at Att. 1.14.3, and cf. e.g. Sail, . Hist. 1 fr. 77.10 (Philippus)Google Scholar, B. Alex. 60.1.

¹⁸ Comparison with the less emotive Att. 2.24.3 brings out how Cicero has given the sequence a rhetorical shape. ‘Day and night’ as a device to enhance is as old as Homer, cf. Il. 24.745, al., and e.g. Epicur, . Men. 135Google Scholar, Enn, . Ann. 336 SkGoogle Scholar, Lucr. 2.12, 3.62, B. Afr. 26.4; on its forms in Cicero cf. Madvig, on Fin. 1.51Google Scholar, Landgraf, on S. Rosc. 6Google Scholar.

¹⁹ See Rizzo's, S. apparatus (M. Tulli Ciceronis pro Cluentio Habito oratio [Milan, 1991])Google Scholar.

²⁰ Nisbet, R. G. M., M. Tulli Ciceronis in L. Calpwnium Pisonem Oratio (Oxford, 1961), 154Google Scholar.

²¹ A break after mali is desirable. The relative clause is not a denning one: ‘uiso atque persuaso’ elaborates ‘opinio’. ‘opinio mali’ presents part of the standard Stoic definition: cf. 25, 74f.; cf. further Kidd on Posidon. fr. 165.6–8 EK, and note Varro's use of ‘magni boni opinionem’ at L. 6.50.

²² Fraenkel (n. 6), 31, clearly recognizes that the final verb can constitute a ‘colon’.

²³ Cf. Nep, . Timoth. 2.3 ‘recens fili ueterem patris renouauit memoriam’Google Scholar.

²⁴ Note efflagitauit in Corn. 1 fr. 53 Crawford.

²⁵ For more examples of such phrases see Adamietz, J., M. Tullius Cicero, Pro Murena. Mit einem Kommentar herausgegeben (Darmstadt, 1989), 233Google Scholar. atque ceruicibus nostris shows an extended rhythm of the kind mentioned in § 1 above.

²⁶ Cf. Sest. 16; and note the scholiast's comment on Aer. Al. Mil. fr. 22 Crawford (cf. fr. 21), Schol. Bob. p. 174.8–10 Stangl.

²⁷ tune Reid: turn codd.; cf. 125, Hon. fr. 26 Grilli.

²⁸ Note the pungent effect of the disyllabic verbs for degrading activities at Parad. 37 ‘qui tergent, qui ungunt, qui uerrunt, qui spargunt’; verbs with four longs would not have the same force at all. Compare in English the monosyllables Spenser uses of the demeaning tasks inflicted by Radigund: ‘To spin, to card, to sew, to wash, to wring’ (F.Q. 5.4.31.6).

²⁹ Cf. Zielinski, (n. 3), D. Clauselgesetz, 729–33Google Scholar, D. constr. Rhythmus, 9 f.

³⁰ Zielinski's, own remarks on hiatus and brevis in longo, D. Clauselgesetz, 732f.Google Scholar, provide further ammunition against him. On word-end, cf. Aili (n. 3), 56; one should not count places like Caec. 28 above.

³¹ The bare ablative emphasizes the means of defence, cf. e.g. Red. Sen. 31, Cael. 11.

³² This last line of approach is espoused by Zielinski, , D. constr. Rhythmus (n. 3), 25fGoogle Scholar.

³³ Cf. Fraenkel (n. 6), 99f.

³⁴ Fraenkel himself well illustrates short questions and answers combining into a single rhythmic unit (41, 213). Split resolution does not matter in Cicero. One cannot use Cicero's discussion of rhythm in the Orator to show that in cases of apparent overlap the final unit would have seemed a desirable rhythmic close in Cicero's own terms. Even if we regarded those terms as authoritative for us (which would be difficult), most of the examples below would not suit Those from Phil. 13.16 and N.D. 3.94 would be exceptions (as would e.g. Ver. 4.23 ‘Phaselis fuit |’, Dom. 76 ‘honestauit |’).

³⁵ Cf. on performance e.g. Nisbet (n. 7), 357.

³⁶ Tac. Hist. 3.39.1 (cf. Suet, . Vit. 14.2Google Scholar) makes apparent the strength of the phrase in this context.

³⁷ The Greek internal accusative gives a weight to θεαν which Cicero achieves through the separation of diuinum (a certain conjecture), sempiternae takes on the sense of both παστου and πρς τν σμπαντα χρνον (R. Giomini in his edition [Leipzig, 1975] wrongly, I think, denies that the latter phrase is rendered by Cicero). The word is much grander than παστου, and also coheres effectively with diuinum.

³⁸ The last part adapts a formula: cf. Serv. A. 7.614, 8.1, Mommsen, , Römisches Staatsrecht ³ (Leipzig, 1887), i.696 n. 1Google Scholar. But note the unrhythmic adaptation at Veil. 2.3.1, and the less dense one at Vir. Ill. 64.7. Note too the dense paraphrase at Rab. Perd. 20.

³⁹ Bake's sed for th e MSS' ut seems better than Ziegler's at: cf. Brut. 58, Fin. 1.1. The oratorical versions of the theme are naturally more intense, Rosc. 66, Pis. 46 (cf. also Parad. 18); those passages tend to confirm the close after ‘ardentibus taedis’ (note the change in order), ‘cruciatu’ may be pointed; cf. Calp, . Decl. 49Google Scholar ‘habet … poenam suam: cruciatur conscientia, pudore torquetur …’.

⁴⁰ In the preceding sentence ‘sunt finitima omnino, sed tamen differt aliquid’, rhythm adds an argument in favour of differunt; I venture to think it preferable on other grounds, despite Madvig, on Fin. 3.15Google Scholar, and the editions of Müller, Pohlenz, Douglas, and others.

⁴¹ Cf. Calvus and Brutus, Ep. fr. ix a 2 SB (see Gudeman, A., P. Cornelii Taciti Dialogus de Oratoribus ² [Leipzig, 1914], 316–19Google Scholar), Orat. 229, Quint. Inst. 12.10.12.1 am grateful to the referee of CQ for suggestions, and especially to Professor M. Winterbottom for much bracing discussion.