¹ Fuhrmann, M., Das Systematische Lehrbuch (1960)Google Scholar, chap. 1.

² Cicero, , Topica 28 f.Google Scholar; cf. Nörr, D., Divisio und Partitio (1972)Google Scholar. Talamanca, M., ‘Lo schema ‘genusspecies’ nelle sistematiche dei giuristi romani’ (Problemi attuali di scienza e cultura: colloquio italofrancese: La filosofia greca e il diritto romano, Roma 14–17 aprile 1973, Tomo II. Acc. Naz. dei Lincei, 1977) 103, 145, 156Google Scholar argues that this is peculiar to Cicero.

³ As their Greek equivalents are by Plato, probably Speusippus, and often Aristotle (in the biological as opposed to the logical works; also Theophrastus): Balme, D. M., ‘Γένος and εἶδος in Aristotle's Biology’, CQ xii (1962) 81CrossRefGoogle Scholar. Diog. Laert. VII 39 reveals that some philosophers described as εἶδη of philosophical λόγος what others called γένη; Strabo I 2 6 seems distinctly confused about the εἶδη of discourse, λόγος, which is γενικός.

⁴ Boscherini, S., Lingua e Scienza greca nel De agricultura di Catone (1970)Google Scholar. Compare the pre-sophistic Works and Days of Hesiod, a combination of parts of very different character, where the third of the four main parts is taken up with a miscellaneous collection of pieces of advice concerned with everyday life; the work as a whole is only partly technical, however, morality and mythology also bulking large.

Helm, R., RE XXII 1, 150 (1953)Google Scholar recounts the history (to that date) of attempts to argue that Cato's text is badly defective or has been tampered with.

⁵ This is at its worst in the last chapters of all: (CL a model contract for sale of lambs, CLI Minius Percennius of Nola and his advice on cypress plantations, CLII elm-twig brooms as recommended by the Manlii for scraping wine-jars, etc.) but it could perhaps be argued that miscellaneous appendices are allowable. But the bulk of the work is only slightly better.

⁶ Cicero, 's De Oratore I 249Google Scholar, set in 91, speaks of Mago as the reigning authority, not the Sasernae.

⁷ Varro, , De RR I 222Google Scholar.

⁸ Ibid. I 18 2–3.

⁹ Skydsgaard, J. E., Varro the Scholar: Studies in the First Book of Varro's De Re Rustica (Analecta Romana Instituti Danici IV Suppl., 1968) 37Google Scholar.

¹⁰ Plato, , Philebus 16cGoogle Scholar.

¹¹ Civitas Dei VI 3Google Scholar.

¹² Ibid. XIX 1. Augustine describes Varro as proceeding diligenter et subtiliter; again, these words are typically used of the analytic method.

¹³ Cicero, , De Fin. V 16Google Scholar.

¹⁴ Gellius, A., NA III 10 16Google Scholar. Ritschl, F., Opuscula III 508Google Scholar suggests seven categories of great men as the basis (cf. the categories at the base of Nepos' Brief Lives).

¹⁵ Gellius, ibid.

¹⁶ Schmekel, A., Die Philosophie der mittleren Stoa (1892) 449Google Scholar: ‘die sonst unbegreifliche Art seine Werke zu disponieren’.

¹⁷ Op. cit. in n. 9, 35–6.

¹⁸ Gerschel, L., ‘Varron logicien’, Latomus xvii (1958) 65Google Scholar quotes this, and compares J. Collart, ‘Varron grammairien Latin’, ibid. (1954) who suggests various explanations—rhetorical or philosophical, esp. of course Pythagorean, influence. Gerschel thinks the division of things into a chain of being at De LL V 80–105Google Scholar is an ancient Indo-European one.

¹⁹ Dahlmann, H., RE Suppl. VI (1935) 1260Google Scholar.

²⁰ P. J. Enk, OCD ² sv. Varro.

²¹ De LL V 11Google Scholar.

²² Dahlmann, op. cit. in n. 19, 1249.

²³ Taylor, Daniel J., Declinatio: A Study of the Linguistic Theory of M. Terentius Varro (1974) 66CrossRefGoogle Scholar holds that etymology is the same thing as declinatio voluntaria and thus irrational (as De LL VIII 6Google Scholar says the latter is).

²⁴ As argued by Fehling, D., ‘Varro und die grammatische Lehre von der Analogie und der Flexion’, Glotta xxxv (1956) 214Google Scholar and xxxvi (1957) 48. But see Dihle, A., ‘Analogie und Atticismus’, Hermes lxxxv (1957) 170Google Scholar.

²⁵ De LL X 10Google Scholar; other numbers given for other Greek authors.

²⁶ Robins, R. H., Ancient and Mediaeval Grammatical Theory in Europe (1951) 54 ffGoogle Scholar.

²⁷ Gellius, A., NA XVI 8Google Scholar.

²⁸ Though Swoboda in his edition of Nigidius Figulus' fragments may be right in claiming that his Commentarii Grammatici were not a mere collection of notes with no organising principle at all, it seems clear that the work was not arranged as an ars, though different subjects were grouped in different books. Noun cases and spelling were not treated till Book xxiv, and Gellius twice complains of at any rate local obscurity in language similar to that he had used of Stilo (NA XVII 7Google Scholar and XIX 14 3). But Nigidius' De diis did show an interest in classification: frag. 67, quidam deos et eorum genera temporibus et aetatibus distinguunt, inter quos et Orpheus; frag. 68, he said, disciplinas Etruscas sequens, that genera esse Penatium quattuor.

²⁹ Douglas, Pace A. E., ‘Clausulae in the Rhetorica ad Herennium as evidence of its date’, CQ x (1960) 65, cfCrossRefGoogle Scholar. CR xvii (1967) 105Google Scholar. But it must be significant that none of its historical examples go beyond the eighties and that its relation to the De Inventione of Cicero is so close as to suggest that the two authors learnt from the same master. Cicero, 's De Or. I 208Google Scholar and the fragments of Antonius' ‘very thin’ libellus on rhetoric (probably written in the nineties) make it pretty clear that this work was not yet organized as an ars.

³⁰ Plato, , Phaedrus 266bGoogle Scholar.

³¹ The enthymeme, of which the major premise is only probable. Rhet. ad Her. II 28Google Scholar: an argumentatio consists of propositio, ratio, rationis confirmatio, exornatio and complexio; Cicero, , De Inv. I 57Google Scholar calls it ratiocinatio, not argumentatio, and agrees that there are five parts—or three according to some. The question is dealt with at some length.

³² Müller, K. K., RE II 2 1637 (1896)Google Scholar.

³³ But since Cato's now very out-of-date handbook, the Romans appear to have preferred to learn about warfare from reading about the res gestae of their ancestors, with some help from Greek technical works (Cicero, , Acad. II 12Google Scholar; Sallust, , BJ 85 12Google Scholar). As for political writing, the Romans produced little, but Cicero's De Re Publica still has much to say of the three good and three bad forms of constitution, and the perfect mixed form. See I 38, prefatory definition vital, but the rest of the process unnecessary for Cicero, who is not aiming at completeness, and is addressing intelligent men on a well-known subject.

³⁴ I 12; cf. Barwick, K., ‘Probleme in den Rhet. 11. Ciceros und der Rhetorik des sogenannten Auctor ad Herennium’, Philol. cix (1965) 57Google Scholar.

³⁵ di Benedetto, V., ‘Dionisio Trace e la Techne a lui attributa’, Ann. della Sc. Norm. di Pisa xxvii (1958) 169, xxviii (1959) 87Google Scholar. On the other hand, Asclepiades of Myrlea, for one, insisted that grammar was a τέχνη.

³⁶ De Or. II 107Google Scholar.

³⁷ See n. 71. It has been remarked that at all stages the rhetorical subdivisions isolated by the ancients were inspired by practical convenience rather than logic or a desire for completeness.

³⁸ De Or. II 43 ffGoogle Scholar.

³⁹ Ibid. 235 ff.

⁴⁰ Cf. Topica 8–20 on intrinsic arguments: ‘sometimes definition, sometimes partium enumeratio, sometimes etymology is required’.

⁴¹ Cicero in the Orator describes the plain style as suited for argument, and in 102 instances his Pro Caecina as an example: the whole point of the speech was to define and distinguish ambiguous words. It is interesting that he puts it in these terms, as we might be tempted to look at his argument as being that words do not matter at all, only the intention of the maiores in framing the interdict. In his theoretical discussion in the Orator Cicero significantly stresses definition.

For his discussion of dialectic in the Orator, and of its use to the speaker, see 14 ff. and note: nec vero sine philosophorum disciplina genus et species cuiusque rei cernere nec eam definiendo explicare nec tribuere in partes possumus (here we do have divisio and partitio distinguished; note how these three procedures head the list) nec iudicare quae vera quae falsa sint neque cernere consequentia, repugnantia videre, ambigua distinguere. Cicero returns to the subject in 113 ff.; in 116 he stresses that in all rational discussion it is essential first to define the matter at issue, and then explicato generis cuiusque rei videndum est quae sint eius generis sive formae sive partes; definitions should be longer and more accessible than those of the philosophers.

⁴² Douglas, A. E., Cicero (Greece and Rome: New Surveys of the Classics, no. 2, 1968)Google Scholar is typical of many; I quote him as he is usually so sensible and sensitive a critic of Cicero, (cf. now in ANRW I 3 (1976) 95)Google Scholar. Brink, C. O., Horace on Poetry I (1963) 1Google Scholar is fairer to Cicero; but it is hard to see the technical framework for which he argues in Horace's Ars Poetica, and indeed no didactic poetry sticks closely to such a framework, partly for poetic reasons and partly perhaps because of the origin and models of the genre in archaic Greece: Pöhlmann, F., ‘Charakteristika des römischen Lehrgedichts’, ANRW I 3813Google Scholar.

⁴³ It does not use the term ars, but it may be partly in order to conform to schematic purity that it omits to list figures of speech, which had not yet been classified, see n. 71 below.

⁴⁴ Acad. I 2 5Google Scholar (put in the mouth of the Academic Varro).

⁴⁵ De Fin. I 22Google Scholar: tollit definitiones, nihil de dividendo ac partiendo docet; non quomodo efficiatur concludaturque ratio tradit; non qua via captiosa solvantur, ambigua distinguantur, ostendit. Note what, as usual, takes first place in a list of the benefits of logic. Cf. II 11 ff., a long criticism of Epicurus' procedure (which was to stress the quid and quale of things) and of his three genera cupiditatum: II 26–7, these are really only two; vitiosum est in dividendo partem in genere numerare; II 30, hic si definire, si dividere didicisset si loquendi vim si denique consuetudinem verborum teneret.

Cf. for the Latin Epicureans Tusc. Disp. II 7–8Google Scholar, complaints beginning that they neque distincte neque distribute … scribere; certainly definition and division again. Talamanca, op. cit. in n. 2, 55 n. 202 observes that ‘i procedimenti diaeretici’ in Philodemus ‘non appaiono rilevare’.

⁴⁶ Swoboda, frag. 113 and note.

⁴⁷ Pliny, NH xv 127Google Scholar.

⁴⁸ The philologists contrast this title with the common ones of Ἄτακτα and suchlike, but this throws little light. Cloatius' date is given by Verrius' use of him. I cannot follow Funaioli (GRF 470) and others in holding that the frag. from Book II proves that Cloatius was also listing sacred objects; it deals with the bloodless altar of Apollo at Delos, adored by Pythagoras, and could well come from the preface to another book of vegetarian scope. We know from a frag, of Cloatius' work on etymology, the liber a Graecis tractorum, that he knew Theophrastus' botanical works, so perhaps the Ordinata Graeca was more than the mere ỏνομαστικόν it seems at first sight.

⁴⁹ See Balme, op. cit. in n. 3.

⁵⁰ Sloane, P., ‘John Locke, John Ray and the Problem of the Natural System’, Journ. of the Hist. of Biol. v (1972) 1CrossRefGoogle Scholar gives an idea of the debates of the 16–18th centuries, not only in England.

⁵¹ He may himself have formalised ‘Tuscanic’ as an order to set beside Doric, Ionic and Corinthian. I doubt if it is adequate to say ‘auch Vitruvs haüfiges Schematisieren ist auf Varro zurückzuführen’ (D. Fehling, review of Fuhrmann, op. cit. in n. 1, Gnomon xxxiv (1962) 113)Google Scholar. Vitruvius knew Varro's book on architecture in the Disciplinae, and admired the De LL, but he has none of Varro's numerological eccentricity.

⁵² There is little evidence for the organisation of the XII Tables. Nineteenth century scholars have probably introduced too much Greek logic into the arrangement we have in e.g. FIRA 1 21 ff.

⁵³ Pomponius, D. I 2.2.39, perhaps from Cicero's lost De Iure Civili in Artem Redigendo.

⁵⁴ Miquel, J., ‘Stoische Logik und römische Jurisprudenz’, ZSS lxxxvii (1970) 85Google Scholar. I can refer to only a fraction of the ‘fast unübersehbare Literatur’ on Greek influence on Roman Law, or its absence; see esp. Talamanca, op. cit. in n. 2, which begins with an account of the scholarly disputes of recent years.

⁵⁵ Wieacker, F., ‘Griechische Würzeln des Institutionsystems’, ZSS lxxx (1953) 93Google Scholar.

⁵⁶ E.g. Wieacker, op. cit.: only ‘fragmentarische Beeinflussung’; in English, Watson, A., Lawmaking in the Later Roman Republic (1974) 179 ffGoogle Scholar.

⁵⁷ Bretone, M., ‘La tecnica del responso Serviano’, Labeo xvi (1970) 7Google Scholar = Techniche e Ideologic dei Giuristi Romani (1971) 73Google Scholar.

⁵⁸ De Inv. II 65Google Scholar; Topica 28.

⁵⁹ I find it hard to accept the argument of Marshall, A. J., ‘The Structure of Cicero's Edict’, AJP lxxxv (1964) 185Google Scholar, that Cicero's classification ‘may represent only a mental process preliminary to actual composition’ rather than the way the finished edict appeared: the method is one primarily concerned with presentation and communication (though the distinction Marshall makes is not unfamiliar to the Stoics).

⁶⁰ For definition, Coing, H., ‘Zur Methodik der Republikanischen Jurisprudenz: zur Entstehung der grammatisch-logischen Auslegung’, Studi in on. di V. Arangio-Ruiz I (1953) 365Google Scholar.

⁶¹ Watson, op. cit. in n. 56, 192.

⁶² Schulz, F., History of Roman Legal Science (1946) Part II, chaps. III and IVGoogle Scholar; cf. Geschichte der Römischen Rechtswissenschaft (1961), the more recent German version.

⁶³ Watson, op. cit. in n. 56, 157, with refs.

⁶⁴ Behrends, O., Die Wissenschaftslehre im Zivilrecht des Q. Mucius Scaevola Pontifex (Nachr. d. Akad. der Wissensch. in Göttingen, I: Philol.-Hist. Klasse, 1976 no. 7)Google Scholar thinks Scaevola used species and genus in a specifically Stoic sense of experienced particulars and rationally known concepts and that it was Servius who introduced definition and classification. This is hard to accept, and Cicero would not have despised a Stoic-inspired work as pre-philosophical (see my review, JRS lxvii (1977) 188)Google Scholar. More reasonable is Schmidlin, B., ‘Horoi, Pithana und Regulae: zum Einfluss der Rhetorik und Dialektik auf die juristische Regelbildung’, ANRW II 15 (1976) 101Google Scholar, who takes a middle line in the great controversy as to the extent of Greek influence on late Republican jurisprudence, but calls Mucius' Ὅροι ‘zweifellos der spektakulärste Versuch, die ars dialectica der Jurisprudenz dienstbar zu machen’; this cannot be true if we are to give any credence at all to Cicero's statements about Servius (whom Schmidlin does not mention). He perhaps also credits Scaevola with a more thorough understanding of Stoic logic than is likely.

⁶⁵ Bruck, E. F., ‘Cicero versus the Scaevolas: Law of Inheritance and Decay of Roman Religion’, Seminar III (1945) 1Google Scholar.

⁶⁶ Wieacker, op. cit. in n. 55: ‘die innere Darstellung des grossen Standardwerks bleibt ganz unerkennbar’, but Schulz was wrong to think Gaius' distinction de personis and de rebus was already there.

⁶⁷ Watson, op. cit. in n. 56, 163, 182 (reconstruction from Paul's epitome).

^67a Talamanca, op. cit. in n. 2, 16 even thinks that with Labeo we come to the most significant stage in the use of diaeretic techniques by the Roman jurists; he does not however think that any of these were doing the sort of large-scale job that Cicero was interested in, but rather dealing with individual concepts, not integrated into a single synthesis till Gaius (as far as we can tell).

⁶⁸ With Palladius in the fourth century, the system is collapsing—his four elements of agriculture are earth, air, water—and industry.

⁶⁹ Dialogus 19 3–5.

⁷⁰ Apollodorus insisted that the orator must always divide his speech into the four basic parts, since this, like the other rules, emerges from the nature of the materials (Brzoska, , RE I 22892)Google Scholar.

⁷¹ Instit. I 323 ffGoogle Scholar. (note loose use of genus still). See Münscher, , RE VII 2 1604Google Scholar; Barabino, G., P. Rutili Lupi Schemata Dianoeas et Lexeos (1967) 13Google Scholar: ‘ci troviamo di fronte ad uno dei primi ed ancora incerti tentativi di classificazione’; cf. Stegemann, W., RE XX i326Google Scholar: the rhetorician Phoibammon uses the classification of figures of speech found in Quintilian (those involving ἐνδεῑα, πλεονασμός, μεταθέσις, ἐναλλαγή) also for figures of thought: ‘wir werden an einen stark grammatisch beeinflussten Rl etor des 1 oder 2 Jahrhunderts nach Ch. denken können’. Quint. Instit. III 41–4Google Scholar shows there had been a recent attempt, which he opposed, by an authoritative source (Greek or Roman?) to establish far more than three kinds of oratory; it was based on hints in some earlier Greek works and in the De Oratore (above, p. 20). Surviving late rhetoricians do classify more than ever—see e.g. the third-century Menander.

At a theoretical level the subject was alive in the later 1st cent. B.C. Boethius' own De Divisione (Migne lxiv 875d) says that Andronicus of Rhodes, doubtless in connection with the revival of Aristotelian studies to which he contributed so much, wrote a περὶ διαιρέσεως; he may have begun the complex διαίρεσις of διαίρεσις itself found in late philosophical sources, for which see Talamanca, op. cit. in n. 2, 55 ff.

⁷² Kessels, A. H. M., ‘Ancient Systems of Dream-Classification’, Mnemosyne xxii (1969) 389CrossRefGoogle Scholar; in Latin, Macrobius, , Comm. in Somn. Scip. 1.3.10Google Scholar has the same pattern. Stoic influence is possible, but some think it older: Schryvers, P. H., ‘La Classification des rêves selon Herophile’, Mnemosyne xxx (1977) 13CrossRefGoogle Scholar.

⁷³ Nat. Quaest. II 56 1Google Scholar.

⁷⁴ Id.ibid. II 39, and 48–50.

⁷⁵ Oppenheim, A. Leo, Ancient Mesopotamia (1964) 211 ffGoogle Scholar.

⁷⁶ The Making of the Middle Ages (1953) 179Google Scholar.

⁷⁷ Instit. V 14 32Google Scholar.

⁷⁸ Op. cit. in n. 62, 68: ‘the various subjects, so far as they are not connected by more or less vague associations, succeed each other without connection or plan’. Cf. Letts, M., ‘The Sachsenspiegel and its Illustrator’, Law Quarterly Review xlxix (1933) 535Google Scholar.

⁷⁹ Driver, and Miles, , The Babylonian Laws I (1952) 41 ff.Google Scholar; a rough division by classes of person, transitions sometimes due to association of ideas. There is no division of the text into sections in the stele, no discussion (naturally enough) of organisation, and connections of any kind are often hard to see. The form of the provisions is particular and hypothetical: ‘if a man does X, he shall be put to death, or given the ox back’, or whatever. Watson's use (op. cit. in n. 56, 194) of these laws to show that the Romans could have learned to classify without Greek aid seems to me misconceived; there is a profound difference between grouping by sorts of person, and the abstract distinction of different types of theft.

⁸⁰ ‘There is no attempt at logical arrangement, civil, cultic and ethical laws appearing indiscriminately’, Hasting's Dictionary of the Bible² (1963) sv. Law (in OT)Google Scholar.

⁸¹ The Mishnah, ed. Danby, H. (1933) xxivGoogle Scholar.

⁸² As in the Koran and the Buddhist Tripitaka.

⁸³ L.'Année Sociologique vi (1901–1902) 1Google Scholar. Cf. Lévi-Strauss, C., La Pensée Sauvage (1963) chap, iiGoogle Scholar: ‘La logique des classifications totémistiques’, esp. 54 on the Navaho.

⁸⁴ Granet, M., La Pensée Chinoise (1934) Livre IIGoogle Scholar: Les Idées Directrices.

⁸⁵ The Etruscans probably already had a system (of oriental origin?) by which different gods were associated with 16 sectors of the heavens and probably of other spaces, particularly the liver of sacrificial animals; what further correspondences were accepted is unclear, though certain trees are associated, in rather general terms, with the di inferi avertentesque in a tantalising fragment of Tarquitius Priscus' Ostentarium Arborarium, one of the works making known the Etrusca disciplina in Latin (Macrobius, , Sat. III 20 3)Google Scholar; see Pallottino, M., The Etruscans (1975) 144Google Scholar.

⁸⁶ I am grateful to Professor P. Stein for drawing my attention to, and lending me, some of the literature on Roman Law; and to Dr. L. Jordanova, and more particularly Mr. R. F. Tannenbaum, for references to some of the non-classical material.