¹ We studied the cursus in ‘A new statistical analysis of accentual prose rhythms in imperial Latin authors’, CP 79 (1984), 114 ff.Google Scholar, and the cursus mixtus in ‘Meter in accentual clausulae of late imperial Latin prose’, forthcoming in the June 1985 issue of CP.

² A detailed discussion of the forms of the three cadences of the cursus (for examples, see Table below) can be found in Lindholm, G., Studien zum mittellateinischen Prosarhythmus: Seine Entwicklung und sein Abklingen in der Briefliteratur Italiens (Stockholm, 1963), 39 ffGoogle Scholar. There are other forms of the cursus, notably the trispondaicus and the dispondaicus; these will be listed in the Table below, but we will not discuss them at present, because they are rare in occurrence and should be regarded as irregular clausulae: see Oberhelman, and Hall, , CP 79 (1984), n. 6Google Scholar.

See also Janson, T., Prose Rhythm in Medieval Latin from the 9th to the 13th Century (Stockholm, 1975), 128 ff.Google Scholar, for bibliography on the cursus.

³ The following is a brief bibliography on the cursus mixtus: Hagendahl, H., ‘La prose métrique d'Arnobe. Contributions à la connaissance de la prose littéraire de l'Empire’ Göteborgs Högskola Årsskrift 42 (1937), 1–215Google Scholar; Nicolau, M. G., L'origine du ‘cursus’ rythmique et les débuts de l'accent d'intensité en latin (Paris, 1930), 42 and 128 ff.Google Scholar; Harmon, A. M., The Clausulae in Ammianus Marcellinus (New Haven, Conn., 1910), 197 ff.Google Scholar; Dewing, H. B., ‘The origin of accentual prose rhythm in Greek’, AJP 31 (1910), 314 ff.Google Scholar; Novotný, M., État actuel des études sur le rythme de la prose latine, Eos, Supplementa, 5 (Lwów, 1929), 74 f.Google Scholar; Clark, A. C., The cursus in medieval and vulgar Latin (Oxford, 1910), 10 ff.Google Scholar; Ballou, S. H., ‘The Clausula and higher criticism’, TAPA 46 (1915), 151 ff.Google Scholar; Badalí, R., ‘Premessa ad uno studio sulla natura delle clausole simmachiane’, RCCM 8 (1966), 16 f.Google Scholar; Shewring, W., ‘Prose rhythm and the comparative method’, CQ 24 (1930), 172 fCrossRefGoogle Scholar. and 25 (1931), 16 f., and ‘Prose rhythm in the Passio S. Perpetuae’, JTS 30 (1928/1929), 56 f.Google Scholar; Shewring, W. and Dover, K., ‘Prose Rhythm', OCD ², 889Google Scholar; and Guillén, J., ‘Origen y constitución del census rítmico’, Helmantica 8 (1962), 309 ffGoogle Scholar.

It should be pointed out that even in a cursus mixtus text clausulae may be structured solely from an accentual perspective and, at other times, solely from a metrical one; for examples of this in the prose of Arnobius see Hagendahl, op. cit. 74 ff.

⁴ Actually the cretic-tribrach was not a preferred clausula in Cicero: we found only 21 instances in a sample of 908 clausulae from his speeches. It does become, however, a desired clausula in the prose of imperial Latin; in fact, it is the third most frequent metrical form in some of the Panegyrici latini. This increase in popularity can be attributed to the coincidence of the metrical scheme and the accentual pattern (cursus tardus). Moreover, the cretic-tribrach was viewed as a resolution of the cretic-spondee of Cicero. Cicero, of course, had a much more varied system of prose rhythm and did not limit himself strictly to cretic and trochaic rhythms. Later writers, however, concentrated almost exclusively on these patterns and in effect canonised them: see Wilkinson, L. P., Golden Latin Artistry (Cambridge, 1970), 157 ff. and 162 fGoogle Scholar. For Cicero's rhythms see Aili, H., The Prose Rhythm of Sallust and Livy (Stockholm, 1979), 51 ff.Google Scholar, and Primmer, A., Cicero Numerosus: Studien zum antiken Prosarhythmus (Wien, 1968)Google Scholar.

⁵ We use the term ictus here simply to denote the first syllable of each metrical unit in a prose clausula. The issue of an ictus in metrical prose is, to say the least, complicated and unresolved. See Nicolau, op. cit. 44 ff. and ‘Quelques considérations sur l'ictus et les rapports avec l'accent’, REL 7 (1929), 148 ff.Google Scholar; Wilkinson, op. cit. 142 ff.; and Allen, W. S., Accent and Rhythm: Prosodic Features of Latin and Greek: A Study in Theory and Reconstruction (Cambridge, 1973), 341 fGoogle Scholar. with notes.

⁶ We did detect the cursus mixtus in the De Platone and the De mundo, works ascribed to Apuleius; however, given the lack of consensus on their authenticity, it is safest to exclude them from consideration here.

⁷ See e.g. Harmon, op. cit.

⁸ Of the seven works of Jerome we have examined so far, four (Vitae Paul. et Hil., Adversus Ruf., Adversus Pelag., and Comment. in lsaiam) display the cursus; of twelve works of Ambrose examined, the cursus was found in eight (De par., De myst., In Psal. 118 exp., Hexaem., De bono mort., Epist., De sacr., and De lapsu virg.).

⁹ Our demonstration of purely accentual rhythms in the pseudo-Apuleian Περ⋯ ⋯ρμηνείας and the Asclepius may support a later date for their composition.

¹⁰ See Oberhelman, and Hall, , CP 79 (1984), 130Google Scholar.

¹¹ We have only the brief De structuris of Sacerdos (GL 6:492–5): see Nicolau, op. cit. 101 ff., for the importance of this treatise for the history of accentual forms. Cf. OCD ², 889.

¹² The texts we used were Mommsen, Th., Theodosiani libri XVI cum constitutionibus sirmondianis (Berlin, 1905, reprinted 1954)Google Scholar and Meyer, P. and Mommsen, Th., Leges novellae ad theodosianum pertinentes (Berlin, 1905, reprinted 1954)Google Scholar.

¹³ Even though the Theodosian codifiers were charged with clarifying the language and making other changes in earlier constitutions (CTh 1.1.5 ad Senatum 429), we presume the changes would not have been so great as to alter significantly earlier clausulae; moreover, the large size of our sample (see below) allows us a representative examination.

¹⁴ Inductive statistics tell us that the margin of error of a sample of almost 1,500 items drawn at random is plus or minus 3·0% at the 99% confidence level. Thus, for example, if we should find that 75·0% of the clausulae in our sample are accentual, the probability is 99 out of 100 that the true value lies somewhere in the range of 72·0%–78·0%. This margin of error is quite low enough for any statistical survey.

¹⁵ In order to avoid subjectivity in determining the nature of the metrical clausulae (a problem that has plagued much of the modern scholarship on Cicero's rhythms), we used the handy tables in Hagendahl, op. cit. 257–60. Hagendahl has devised 128 combinations of any long and/or short syllables on the basis of A. W. De Groot's work in comparative statistics. De Groot is quite right in insisting that the clausula begins where the preceding syllable is indifferent to quantity. For a summary of De Groot's theory see Wilkinson, op. cit. 139 ff. Cf. also Douglas, A. C., ‘Clausulae in the Rhetorica ad Herennium as evidence for its date’, CQ 54 (1960), 65 ffCrossRefGoogle Scholar. and Shewring, W., ‘Prose rhythm and the comparative method’, CQ 24 (1930), 161 ffCrossRefGoogle Scholar.

¹⁶ See Oberhelman, and Hall, , CP 79 (1984), 117 ffGoogle Scholar.

¹⁷ Inductive statistics tell us that, if the legal texts are in fact non-accentual, then the probability of observing a frequency of 90·0% cursus forms is less than 1 in 1 million.

¹⁸ See the table in Oberhelman, and Hall, , CP 79 (1984), 122 ffGoogle Scholar.

¹⁹ A strict adherence to a few forms was pointed out long ago by Collinet, P., ‘Une programme d'étude sur l'emploi du cursus rythmique par la chancellerie impériale romaine’, REL 5 (1927), 256Google Scholar. Collinet's brief survey is the only study known to us which deals with rhythm in the legal texts; he, however, gives no figures and cites from the legal texts only a few examples upon which his conclusions rest.

²⁰ We place an accent on non here, although it is usually a proclitic. Cf. Hagendahl, ibid 17.

²¹ For the practices of the early papal chancelleries see Poole, R. L., Lectures on the History of the Papal Chancery Down to the Time of Innocent III (Cambridge, 1915), esp. pp. 3, 93, and 99Google Scholar; Brazzel, K., The Clausulae in the Works of St Gregory the Great (Washington, D.C., 1939), 28 ff and 57 ff.Google Scholar; and Di Capua, F., Il ritmo prosaico nelle lettere dei papi e nei documenti della cancelleria Romana dal IV al XIV secolo (Roma, 1934), xxviiGoogle Scholar. See also the tables of accentual clausulae in the letters of the popes of the medieval period in Janson, op. cit. 109 ff. and Lindholm, op. cit. 165 ff.

²² Again, inductive statistics tell us that the probability is less than 1 in 1 million that we should observe such frequencies of metrical clausulae in a purely accentual text unless there was a deliberate attempt to accommodate them.

²³ For the coincidence of ictus and accent in Arnobius, see Hagendahl, op. cit. 74 ff. Cf. OCD ², 889.

²⁴ Of the planus forms in the Codex, 88·2% contain the cretic-spondee or ditrochee; 79·1% of the velox, the ditrochee or cretic-spondee; and 70·7% of the tardus, the dicretic or cretic-tribrach.

²⁵ See Crusius, F., Römische Metrik: eine Einführung (München, 1967), 27Google Scholar, and Aili, op. cit. 49. For examples in Catullus see Fordyce, C. J., Catullus (Oxford, 1961), 121Google Scholar. M. D. Reeve has pointed out to us that hexameter poets may have to shorten or lengthen the final -o (particularly -tio nouns and -io verbs) because of metrical considerations. In the legal texts, however, the issue is certain: final -o is short, unless it is preceded by -i-.

²⁶ See Courtney, E., A Commentary on the Satires of Juvenal (London, 1980), 185Google Scholar.

²⁷ The Novellae will be cited by the emperor, followed by the edict and section number in Meyer and Mommsen's text.

²⁸ The Codex will be cited in accordance with the numerical system in Mommsen's text.

²⁹ Our work with contemporary literary texts, e.g. Symmachus and the panegyrists, has detected a duality in the final -o in verbal forms: the -o in future imperatives is short, as in esto; the final -o is also short in the first person singular forms, as is not uncommon in poetry (see Fordyce, loc. cit.; Crusius, loc. cit.; and Aili, loc. cit.); third conjugation -io verbs prove the exception, however, by retaining a long -o. In our survey of the clausulae of the legal texts, we found no example of first person singular forms with the ending in -o.

³⁰ For Arnobius see Hagendahl, op. cit. 103 ff. See Augustine's discussion of the form feceritis in his De doctr. christ. 4.20.41 and the commentary on this passage in Sullivan, T., S. Aureli Augustini De doctrina christiana liber quartus (Washington, D.C., 1930), 138 fGoogle Scholar. See also Fordyce, op. cit. 107 and Aili, loc. cit.

³¹ Novermŭs āccīdĕrê (Valent. 14.1); fábricam iūssĕrīmûs (CTh 15.1.14); arbitrio sūmpsĕrīmûs (16.1.2); and intúito iūssĕrīmûs (CTh 9.40.13).

³² Sometimes the word-order has been unduly strained, as in profutura civitati augébitur mūltĭtūdô (Valent. 5.1), urget inmitis execútio mīlitārîs (Valent. 1.3.2), hereditario poterit iūrĕ trānsmīttĕrê. (CTh 14.24.1) and qua nullum carere dḗbĕt ōrācǔlûrm (Valent. 19).

³³ See Wilkinson, op. cit. 156 f., for the Rhetorica ad Herennium and Cicero's Orator. See also Augustine, , De doctr. christ. 4.20.40–1Google Scholar.

³⁴ The word-order optūrĕ dēbḗmûs is retained in Theod. 4.2 in order to secure a cursus Mixtus clausula.

³⁵ Sometimes the change is done solely to produce an exact agreement between accent and ictus and not merely a cursus mixtus clausula. E.g. in ita ab invitis iubḗmǔs ārcḗrî, (Maiorian. 11) and relaxūmǔs ōbnṓxĭīs (Maiorian. 2.1) there is exact agreement; if more usual word-order had been used, i.e. arcérī iǔbḗmûs and obnóxiīs rĕlāxūmûs, cursus mixtus forms would have been realised, but without exact correspondence of accent and ictus.

³⁶ Cf. Janson, op. cit. 32.

³⁷ Justinian's compilers were not interested in prose rhythms of any sort. We could detect no trace of purposeful metrical or accentual rhythms in a sample of 700 clausulae.

³⁸ Laurand, L., Etudes sur le style des discours de Cicéron, aveç une esquisse de l'histoire du ‘cursus’ ² (Paris, 1926), ii. 217Google Scholar, and Shewring, W., ‘Prose rhythm and the comparative method’, CQ 25 (1931), 20 fGoogle Scholar.

³⁹ MS O in several other passages preserves readings of a rhythmical nature that other codices do not: e.g. Marcian. 1.4 has antiquitus statuta sunt (no accentual form with the infrequent trochee-cretic), but O preserves antiquitus sūnt stătt (cursus velox with ditrochee); at Valent. 12.3, O has infortunia sūnt ēxprḗss (cursus velox with dispondee), while Γ and Δ read. infortunia ēxprḗssă sûnt (no accentual scheme with choriamb-cretic, assuming the presence of hiatus); also, at Maiorian. 1.3, the text reads unusquīsquĕ vḗstrûm (ditrochee with no accentual scheme), while O has unusquīsquĕ vēstrṓrûm(cursus planus with cretic-spondee).

⁴⁰ O here has servaturus, which is surely picked up from the next line, which also contains servaturus.

⁴¹ Valebit is the customary form in the legal texts; e.g. it occurs in the first few chapters of the Codex at 1.2.5, 1.2.6, 1.2.8, 1.2.9, and 1.5.3.