¹ The idea that emotions are judgements may at first glance seem less surprising to a modern than to an ancient philosopher: but crucial issues in Chrysippus' moral psychology – his conception of reason, his grounds for justifying moral responsibility, his reasons for believing the extirpation of the passions to be a good thing – are inescapably Stoic.

² The standard histories are: Marrou, H.-I., Histoire de l'Éducation dans l'Antiquité (Paris, 1950, tr. G. Lamb, London, 1956)Google Scholar; Clark, D. L., Rhetoric in Greco-Roman Education (N. Y., 1957)Google Scholar; Clarke, M. L., Higher Education in the Ancient World (London, 1971), pp. 28ff.Google Scholar; Bonner, S. F., Education in Ancient Rome (London, 1977)Google Scholar. General treatments of rhetoric in antiquity include: Volkmann, R., Die Rhetorik der Griechen und Römer (Leipzig, 1885, 2nd ed.)Google Scholar; Clarke, M. L., Rhetoric at Rome (London, 1953), esp. chs. 1, 3CrossRefGoogle Scholar; Norden, E., Die antike Kunstprosa (5th ed., Stuttgart, 1958)Google Scholar; Leeman, A. D., Orationis Ratio (Amsterdam, 1963)Google Scholar; Kennedy, G. A., The Art of Persuasion in Greece (Princeton/London, 1963)Google Scholar; id., The Art of Rhetoric in the Roman World (Princeton, 1972); Martin, J., Die antike Rhetorik: Technik und Methode (Munich, 1974)Google Scholar.

³ Style has been by far the most long-lived of all these rhetorical topics, of course. What distinguishes the rhetorical handling of style, however, are its continuing associations with persuasion and with the formal, institutionalised contexts in which it was displayed, associations which remained part of the method of teaching composition even beyond the end of antiquity. On the later history of rhetoric, see Kennedy, G. A., Classical Rhetoric and its Christian and Secular Tradition from Ancient to Modern Times (London, 1980)Google Scholar.

⁴ The stasis (or status) of a case, roughly speaking, is the ‘issue’ on which it turns, the point which if decided will determine the outcome of the whole case. Defining stasis and isolating its intricate subdivisions are major topics in the handbooks. Quintilian 3.6 is the clearest ancient treatment of this tricky subject. See also R. Volkmann (cited n. 2), pp. 38–92; Bonner, S. F., Education (cited n. 2), pp. 296ff.Google Scholar; Russell, D. A., Greek Declamation (Cambridge, 1983), ch. 3CrossRefGoogle Scholar.

⁵ Useful introductions to the teaching of style in antiquity can be found in all the texts cited in n. 2.; see also Russell, D. A., Criticism in Antiquity (London, 1981), esp. chs. 8, 9Google Scholar.

⁶ See Russell, , op. cit. n. 5, pp. 136ff.Google Scholar, on this tendency.

⁷ The chief bones of contention have been: Theophrastus' contribution to this area of stylistic theory, and the origins and ramifications of its commonest variant, the doctrine of the ‘three styles’ (viz. the grand, the plain, and the smooth or middle style), which was itself hotly debated; e.g. ad Her. 4.11 ff.; Cicero, de or. 3. 199, cf. 210ff., where different styles are discussed under the rubric of the virtue of appropriateness, and associated with different sorts of discourse; orator 20ff.; Quintilian 12.10.58ff.). Especially useful modern studies are: Kennedy, G. A., The Art of Persuasion (cited n. 2), pp. 278ff.Google Scholar; Russell, Criticism (cited n. 5), ch. 9; Innes, D., ‘Theophrastus and the Theory of Style’, in Theophrastus of Eresus: On His Life and Works, ed. Fortenbaugh, W. W. (Rutgers University Studies in Classical Humanities, vol. 2, 1985), at 251ff., esp. 260ffGoogle Scholar.

⁸ The fictitious deliberative or forensic orations, the suasoriae or controversiae, need not detain us: any contributions Stoic rhetoricians may have made to the ancient debate over the usefulness of such exercises (summarised by Quintilian, 2.10. Iff.) have not survived, and there is no point even looking for a purely aesthetic appreciation of them in the early Stoa. Later on, Posidonius may have entered into a related controversy about the θέδις: see n. 20. On declamatio, see Bonner, S. F., Roman Declamation (Liverpool, 1949)Google Scholar; Winterbottom, M., ed., Roman Declamation (Bristol, 1980)Google Scholar; Russell, Greek Declamation cited n. 4. On epideictic's historical origins, and its traditional association with a more ornate style, see e.g. Russell, D. A. and Wilson, N. G., Menander Rhetor (Oxford, 1981), xi ffGoogle Scholar.

⁹ In Rome representation by a ‘professional’ orator was far more common than under classical Greek law: see for example Dover, K. J., Lysias and the Corpus Lysiacum (Berkeley/Los Angeles, 1968), chs. 8, 9Google Scholar. (I put ‘professional’ in quotation marks because in law no fees could be received.) The Greek logographos has thus to be able to write himself into another's character. But writing ‘in character’ was no less important for the Roman orator, who needed to put across to his audience both the (moral) character he wished to project, and the character he wished his client to seem to possess; and there was also the task of introducing one's opponents, as well as fictitious or historical personages. The motive, of course, was to appear as plausible as possible. On Roman procedure, see Kennedy, , Art of Rhetoric (cited n. 2), pp. 8ffGoogle Scholar.

¹⁰ This definition of the orator, attributed to Cato the Elder by Seneca the Elder (contr. l. pr. 9), may perhaps suggest Stoic influence: but its totemic rôle may be another, small illustration of the happy coincidence between (some of) the principal tenets of Stoic ethics, and Roman ideology as developed in the face of Greek culture.

¹¹ The first definition of λόγος is presumably that of Diogenes of Bablyon, for he is the authority cited for the definition of λέξις which immediately precedes it. Diogenes' definition is slightly more sophisticated than the second (that at 7.57). His influence is perhaps confirmed by Galen: cf. PHP 2.130..13–15 De Lacy. That written language too comes under the scope of these definitions is shown, for example, by the fact that the Stoic ‘parts of λόγος’, the Stoic word-classes, are the constituents of all language, not just of significant talk.

¹² According to Posidonius' definition a poem is ‘metrical or rhythmical λέξις which deliberately avoids the form of λόγος (τὸ λογοειδές)’;. Language thus formally achieves poetic status by possessing rhythm or metre. Almost any rhetorician one could name stresses the vital importance of prose rhythm in oratory (e.g. Aristotle rh. 3.8.1408b21ff.; Cicero, de or. 3.173ff.; orator 168ff.; Quintilian 9.4.52ff.), and even Theophrastus thought polished prose at least should have rhythm (de or. 3.184). The main debate centred on how far oratory should reproduce the rhythms of poetry (an unorthodox view in Dionysius, c.v. 415–17). Lack of Stoic interest in rhythm is at least consistent with their disapproval of the figure Hyperbaton: see pp. 416. Chrysippus' attitude to euphony fluctuates intriguingly: see p. 420.

¹³ ‘The Principles of Stoic Grammar,’ in The Stoics, ed. Rist, J. M. (Berkeley/Los Angeles, 1978), pp. 27ff., at 39–41Google Scholar.

¹⁴ I would propose two further reasons why such articulate strings would be in question: (1) they are the bearers of ambiguity (Diogenes 7.62) and ambiguity will cause lack of clarity; and (2) they will also, presumably, be the primary bearers of what the Stoa would regard as undesirable euphonic and rhythmic properties: see further pp. 417–18. The use of the term φράσις in the definition of Hellenism is puzzling; I can only suggest that it is intended to exclude mere articulate strings of sounds or letters, for there seem to be only two Stoic offences against linguistic purity, barbarism, which attaches to single terms, and solecism, a fault of syntax. Yet, puzzlingly, the same term denotes one of the subdivisions of rhetoric, as already noted (Diogenes 7.43). The answer may simply be that φράσις is one of the commonest Greek terms for what the Roman rhetoricians called elocutio: Quintilian 8.1.1.

¹⁵ It is of course disputed how much of Diogenes Laertius' detailed report of Stoic dialectic at 7.49–83 is an excerpt from Diocles Magnes' Survey of Philosophers (7.48). The brief survey of dialectic at 7.43–4 is in fact very close, in both content and arrangement, to the longer account; the principal difference is that in the version of Stoic logical doctrine arguably known to Diodes epistemology is dealt with before dialectic proper begins, at 49–54. But, whether or not it is specifically Diodes whom Diogenes is excerpting, one must agree with Mejer that ‘the Stoic doxography cannot have been composed by Diogenes himself – nor can it go back to one of his usual sources. It must have been taken in toto from some other source’ (Diogenes Laertius and his Hellenistic Background [Hermes Einzelschriften, 40; Wiesbaden, 1978], p. 7Google Scholar). The information about Stoic stylistics must accordingly be granted special status. For a detailed,. discussion of the arguments for and against Dioclean authorship, see Mejer, op. cit., pp. 6ff.Google Scholar; see I also Holwerda, D., ‘De Dioclis Magnesii alterius operis vestigio neglecto’, Mnemosyne 15 (1962), 169–70CrossRefGoogle Scholar.

¹⁶ I strongly suspect that the remaining part, the ‘proof’, has simply dropped out of the text; it is impossible to believe that the Stoics simply neglected what was commonly held to be the vital component of any suit, proving one's case, especially in light of the Stoic reputation for sharpness and effectiveness in argument: see pp. 401. It is unclear why memory is absent, and why arrangement comes after and not before ϕράσις, its more usual position in the rhetorical textbooks. ϕράσις is itself a conventional rhetorical term: see n. 14.

¹⁷ For an informative discussion of the earliest Stoic dialectic, see Long, A. A., ‘Dialectic and the Stoic Sage’, in The Stoics (cited n. 13), pp. lOlff., esp. 105ffGoogle Scholar. Long argues that it was Chrysippus who developed the whole of Stoic dialectic (not merely formal logic) beyond its question-and-answer origins into a science of all rational discourse.

¹⁸ See for example Long, , art. cit. (n. 17), 102–7Google Scholar.

¹⁹ For διέξοδος, cf. Clement, , Strom. 1 ch. 8, §39, vol. 2 p. 26. 10 Strahlin, vol. 2 p. 30.1 Klotz, cf. 8 ch. 4 §11, vol. 3 p. 203.28 KlotzGoogle Scholar; Plutarch (Fabius Maximus 16.5) applies the adjective to detailed, narrative history.

²⁰ There are other, less direct pieces of evidence that Stoic rhetoric has a narrower field of operations than all and any continuous discourse. Posidonius, who made at least one contribution to the technical side of rhetoric, in stasis theory (Quintilian 3.6.37), may have publicly defended the restriction of rhetoric to these narrow confines: this is implied by Plutarch's report (Life of Pompey 42.5) that he criticised the 2nd century B.C. rhetorician Hermagoras (usually credited with the ‘invention’ of stasis theory) on the topic of ἡ καθ' ὄλου ζήτησις, i.e. the rhetorical ‘thesis’. Cicero reveals (de inv. 1.8, de or. 2.65ff.) that Hermagoras and his followers had come under New Academic fire for failing to restrict rhetoric's field of activities – for poaching on philosophical and scientific territory – and this could well have been the point of Posidonius’ criticism too.

²¹ Striller saw the ghost of Aristotle stalking the Stoic ramparts at this point (de Stoicorum studiis rhetoricis, Breslauer, phil. Abhand, . 1 2, Breslau, 1886, p. 19Google Scholar); but ghosts, of course, are the products of a fevered imagination, and Aristotle could not have conceded that rhetoric and dialectic differ only in their style of discourse. It remains doubtful that in the area of rhetoric the Stoa owed anything specifically to Aristotle, and their debt to Theophrastus is certainly very limited: see pp. 419–20.

²² Schofield, M., ‘The Syllogisms of Zeno of Citium’, Phronesis 28 (1983), 31ff., esp. 51CrossRefGoogle Scholar.

²³ I am being intentionally vague, in order to skirt the difficult topic of what Zeno thought made an argument valid or probative: cf. Brunschwig, Jacques, ‘Proof Defined’, in Doubt and Dogmatism, edd. Schofield, M., Burnyeat, M., Barnes, J. (Oxford, 1980), pp. 125–60, esp. 155–60Google Scholar; and Schofield, , art. cit. (n. 22), 54–5Google Scholar.

²⁴ Cf. de or. 2.158: the Stoic dialecticians ‘ad extremum ipsi se compungunt suis acuminibus’.

²⁵ According to the theorists Quintilian is criticising, the natural officium of words is ‘to be slaves to sense’, servire sensibus. Things have their own names, and there is no need for circuitus or translatio (apparently blanket terms for periphrasis, tropes, and all rhetorical ornament); the oldest orators used to speak ‘maxime secundum naturam’, but later adopted a more ‘poetic’ language (12.10.41–2). Quintilian replies (43) that the orator has not only to ‘state the facts’, but to please and move his audience, and to do so he uses the ‘adiutoria quae sunt ab eadem natura nobis concessa’. The position he rejects could well form part of a (Stoic-influenced) Atticising diatribe, rather than being authentically Stoic (or Epicurean). The Stoic contribution to Roman Atticism is disputed; Kennedy, , An of Rhetoric (cited n. 2), 241–2, 338–9Google Scholar, argues convincingly that Atticism is, unusually, an original Roman contribution to stylistics, without Greek precedents.

²⁶ Philodemus, unfortunately in a very fragmentary context, refers to some people who ‘have taken their technical matter (τὰ τεχνικά) from other (τέχναι), e.g. certain eristic matters and the business of ambiguities from dialectic’ (rh. 2. 67.7–68.12 Sudhaus). This text shows at the very least that the question of shared τεχνικά – which presumably are or include θεωρήματα, the theorems of a science or expertise – was debated in Hellenistic times.

²⁷ Quintilian must mean ‘Atticisers’ rather than ‘Athenians’ his point is that the Atticists, with their concern for unadorned, emotionally restrained discourse, appealed to classical Athenian procedural law as providing a model for the structure of the oration which happily suited their antipathy to ornament and emotional stridency. Their appeal seems to have little historical basis: see Butler's, note ad loc. (Loeb tr., vol. 2, p. 386 n.l)Google Scholar.

²⁸ Graeven, ad loc. (Cornuti Artis Rhetoricae Epitome, edition and commentary by Graeven, J. (Berlin, 1891Google Scholar, repr. Dublin/Zurich, 1973), §207, p. 41.10ff. = 1.454.1.Iff. Sp.) doubts whether Quintilian is referring to Stoic (and Platonic) views, but he gives no reason for this scepticism. (Graeven's identification of Corhutus as the author of the original version of this handbook has not been accepted: see Kennedy, , Art of Rhetoric [cited n. 2], 616 n. 5Google Scholar). For Stoic influence on the Atticisers, see n. 25.

²⁹ Philodemus refused to admit two of the three conventional types of rhetoric, forensic and deliberative, as τέχναι, while claming special status for ‘sophistic’ rhetoric, which apparently extends to literary prose in general; but he continued to stress the importance of clarity (and solecism is classed as a cause of obscurity: rh. 1, pp. 157–8, col. 15.6–24 S.). He voices disapproval of the (panegyric) orators' lack of concern for the truth, and of their obsession with ‘mere sound, and periods, and parallel constructions, and antithetical ones, and homoioteleuta’ by which people are ‘persuaded’ (ψυχαγωγούμενοι) (1, p. 33 = 2, pp. 257–8, col. 4^a4–col. 5^a.4). Philodemus is a useful but limited source for Diogenes of Babylon's teaching on rhetoric (see p. 421); unfortunately the text is badly preserved and fragmentary, and it is not always possible to be sure who is being quoted and criticised.

³⁰ Chrysippus defined rhetoric as a kind of τέχνη, and τέχναι are not confined to the wise (cf. SVF 2.393). What Chrysippus meant by this is far from clear. See further, pp. 420.

³¹ Nature and the things in accordance with nature are ‘the principle of duty and the material of virtue’ according to Chrysippus (Plutarch, comm. not. 1069E)Google Scholar. See Inwood, B., Ethics and Human Action in Early Stoicism (Oxford, 1985), pp. 201ffGoogle Scholar.

³² This idea of advancement to virtue through practice seems to be implied by e.g. SVF 3.500, 510, and esp. by Cato, at de fin. 3. 17ffGoogle Scholar.

³³ E.g. SVF 3.124, 128; Diogenes 7.104–5; Cicero, de fin. 3.50. Withou t this difference between objects – not a difference in mora l value, but in terms of preference – there would be nothing for wisdom to do, nothing on which it could act.

³⁴ The sage and the friend are classified as ‘no t other than benefit (ὠθέλεια)’: Sextus, M. 11. 22ff.; cf. Diogenes 7.103, SVF 3.104. The virtues an d virtuous actions are classed as ‘good’ because benefit happens to result from them: M. 11.26.

³⁵ For this distinction, see e.g. Plutarch, St. rep. 1048A, a quotation from Chrysippus’ περίἀγαθῶν Bk. 1.

³⁶ See especially SVF 1.179, 552, 186, 357a, 566; 2.1170; 3.214, 216, 234.

³⁷ The sage will not, admittedly, always and automatically take part in politics. This is implied by the principle ‘the sage will enter politics, if nothing prevents him’ which has already been quoted (Diogenes 7.121). He will prefer to do so (SVF 3.611, 686, 690), just as he will prefer to marry an d have children. ‘<for> these things are in accordance with the <nature> of the animal which is rational, sociable, and reciprocates affection’, SVF 3.686, p. 172.19–20. Stobaeus mentions one ground above all for the wise man's abstaining from politics, ‘if he is going to do no benefit to his homeland’ (SVF 3.690, p. 173.20–1). Indeed the question ‘will the wise man enter politics?’ became a favourite topic for declamation (Quintilian 3.5.6; Cicero, , top. 82Google Scholar. Under the later Republic an d the Empire commitment to political service became a hotly-contested issue, as Cicero an d Seneca bot h attest (see Griffin, M., Seneca, a Philosopher in Politics (Oxford, 1976), ch. 10.Google Scholar; the Stoic position is analysed too, at pp. 340–4, and Cicero's at pp. 344–5). But my concern here is rather with how the sage will behave once he has made the decision not to abstain from the life of politics, an d not with his reasons for abstaining or quitting.

³⁸ For Zeno, ordinary people are ‘enemies and foes and slaves and foreigners, even parents to their children and brothers to brothers and kin to kin’ (Diogenes 7.32), whereas the good alone are ‘citizens and friends and kin and free’ (33; cf. 124, and SVF 1.262, 264). In Zeno's ideal republic there would be no lawcourts: Diogenes 7.33; forensic rhetoric would presumably wither away. (In Posidonius' Golden Age the wise were rulers and apparently there were n o laws: Seneca, ep. 90.5.)

³⁹ Stobaeus (SVF 3. 686) reports that the sage will by choice take part in politics, and that being a king or supported by one, and being a sophist, are acceptable options too. This is strikingly similar to Chrysippus' account of the sage's three best methods of making a living: Plutarch, , St. rep. 1043EGoogle Scholar.

⁴⁰ This is suggested by the anecdote recounted by Seneca, ep. 116.5; cf. also perhaps Cicero, de off. 1.46.

⁴¹ Panaetius does claim that the wise man needs health, money, and strength (Diogenes 7.128), but that this is almost certainly the familiar Stoic doctrine that he must have a sufficient store of ‘preferred’ things on which to exercise his virtue has been ably argued by Rist, J. M., Stoic Philosophy (Cambridge, 1969), pp. 7ffGoogle Scholar.

⁴² The focus of the debate is whether de off. 1.71–2 is borrowed from Panaetius (as argued by Pohlenz, M., Antike Führertum (Teubner, 1934), pp. 46–7Google Scholar); on the other side Rist, J. M. (op. cit.(n. 41), pp. 193, 200Google Scholar) argues that de off. 1.152–60 is a veiled attack by Cicero on Panaetius' scholastic ideal.

⁴³ ‘The Stoic Doctrine of the Affections of the Soul’, in The Norms of Nature, edd. Striker, G., Schofield, M. (Cambridge, 1986), pp. 93ff., at 104Google Scholar.

⁴⁴ On Dionysius, see Russell, , Criticism (cited n. 5), p. 137Google Scholar, and Bonner, S. F., The Literary Treatises of Dionysius of Halicarnassus (Cambridge, 1939), pp. 18ffGoogle Scholar.

⁴⁵ The pretty well universal ancient assumption in literary criticism (and teaching of composition) that any discourse has content independent of language, is clearly shared by the Stoa; and thus can win no place in an attempt to isolate and explain crucial differences between Stoic and professional rhetoric.

⁴⁶ Cf. orator 71–2: ‘Quod et in re de qua agitur positum est et in personis et eorum qui dicunt et eorum qui audierunt. Itaque hunc locum longe et late patentem philosophi solent in officiis tractare – non cum de recto ipso disputent, nam id quidem unum est…’. The philosopher most closely associated with the topic of duty is Panaetius, and it looks likely that Cicero has at least been inspired by Panaetius' work in his own analysis of the oratorical qualities appropriate to each character and each station in life; Panaetius ethical scheme for adapting one's behaviour to one's role is outlined at de off. 1.107ff. Whether Panaetius adapted his notion of individual appropriateness to discourse is, as far as I know, not recorded, although he was a teacher of the famous Stoic Rutilius Rufus, whose oratory was so unsuccessful: see pp. 426–7. The unorthodoxy of the sort of appropriateness the Stoics may have had in mind is perhaps illustrated by Zeno's claim that ‘nihil esse obscenutn, nihil turpe dictu’ (SVF 1.77): since there is nothing naturally obscene, language used to denote it cannot be obscene either. Hence there will be no need for polite circumlocutions, whatever the audience or circumstances of utterance.

⁴⁷ Art. cit. (n. 22), 56.

⁴⁸ Art. cit. (n. 13), 55ff.: cf. his ‘The Origins of Traditional Grammar’, in Historical and Philosophical Dimensions of Logic, Methodology, and Philosophy of Science, edd. Butts, R. E., Hintikka, J. (Dordrecht/Boston, 1977), pp. 51ff., esp. 60, 68, 71, 72–4Google Scholar; and also Egli, U., ‘Stoic Syntax and Semantics’, in Les Stoiciens et leur logique, ed. Brunschwig, J. (Paris, 1978), pp. 135ff., esp. 138Google Scholar.

⁴⁹ A vaguer, but still powerful factor will I believe have been the Stoic assumption of the fundamental rationality of language, which reflects – albeit imperfectly, given its corruption over time – the rational ordering of the cosmos, down to the level of individuals. On the one hand, the structure of sentences, as noted above, is largely determined by the syntax of λεκτά, which in turn appropriately articulates externally obtaining relations between particular objects with reference to their actions, passivities, and states; Chrysippus’ interest in anomaly probably testifies to an acceptance of a ‘basic regularity’ in the relationship between language's formal and semantic characteristics: see Frede, (art. cit., n. 48), 68–9Google Scholar. Stoic Hellenism is the expertise which extracts rules and underlying principles from linguistic data on the assumption that how we speak is ultimately determined, in content and structure, by reason; and will even go so far as to modify usage where the primitive correspondence has been unacceptably blurred or distorted. On the other hand, Chrysippus' reliance – on a vast scale – on everyday usage, in his de anima and On the passions (see n. 61) as support for Stoic (moral) psychology, rests on the assumption that language can be a reliable guide to the nature and quality of things in the world, such as the seat of the ruling part of the soul. Stoic etymology (reported e.g. by Augustine, de dialectica, ch. 6) attempts to provide rules for these correspondences at the level of the individual word.

⁵⁰ See also Russell's, D. A. note ad loc, ‘Longinus’ On the Sublime (Oxford, 1964), p. 151Google Scholar.

⁵¹ Death may look like the ‘odd man out’ in this list of approved objects, but no doubt Cato's rhetorical skills would be deployed to represent death as no evil, and even as a duty in certain circumstances, and to evoke the (unusual) emotional responses appropriate to this Stoic perspective.

⁵² 2nd century: OCD s.v.; 1st century: RE s.v., vol. v^a pp. 2037–8.

⁵³ On the rhetorical exercises as a genre, see: esp. Kroll, W., R.E. Supp. 7Google Scholar, ‘Rhetorik’, coll. 1118–19; Reichel, G., Quaestiones Progymnasmaticae (Leipzig, 1909)Google Scholar; Kennedy, , Art of Persuasion (cited n. 2), 270Google Scholar; Clark (cited n. 2), pp. 177ff.; Bonner, S. F., Education (cited n. 2), pp. 250ffGoogle Scholar. There is a new translation of Theon by J. R. Butts (Claremont, 1986). The progymnasmata will be referred to by the Spengel (Spengel, L., ed., Rhetores Graeci, 3 vols., Leipzig, 1856Google Scholar), volume, page, and line numbers; it begins at 2.59ff. The Stoic classification is at 81.30–83.13. All translations are my own.

⁵⁴ On linguistic sophisms, περὶ τῶν παρὰ τὴν λέξιν σοϕισμάτων, vol. 14.582–98 Kühn; references by the page and line number of the edition of Gabler (Diss., Rostock, 1903) and the page number of Ebbesen's, S. edition (Commentators and Commentaries on Aristotle's Elenchi, Sophistici (3 vols., Leiden, 1981), ii. 1ff.Google Scholar). The Stoic classification is at ch. 4, 12.10ff. G, 21–5 E. The ‘more sophisticated Stoics’ are mentioned at 12.18 G 21E. Theon's wording at 81.30–1 might lead one initially to believe that the only sort of ἀμϕιβολία isolated by the dialecticians is that παρὰ τὴν κοινήν, and that therefore the rest of the classification is Theon's own, or is drawn from some other source; but the closeness of the resemblance between this list and the one reported by Galen rules out either possibility – Theon's clumsiness of expression is simply misleading. For a full discussion of both classifications, see my The Stoics on Ambiguity (Cambridge, forthcoming).

⁵⁵ ‘Obscurity in word-complexes’ is said to occur ‘when what is said can be understood in several ways, nothing being either added or taken away’ (76.24–6). But what Theon is offering here is almost certainly a definition of ambiguity. His brief list here of cases of obscurity in single terms obviously anticipates the longer and more detailed classification at 81.6ff., immediately before the dialecticians' classification of ambiguity types. Theon is surely calling on the same sources for both his treatments of the topic of linguistic obscurity. If so, the ‘obscurity in word-complexes’ he describes will in fact be a condensed version of the dialecticians’ ambiguity: Theon has simply failed to note that they are classifiying equivocation in particular, not obscurity in general. And in that case the dialecticians' ambiguity will by definition have excluded Elleipsis.

⁵⁶ Theon's clumsy introduction to the classification has already been observed (n. 54). Most revealing, perhaps, is Theon's account of the last equivocation type (82.26ff.), which is significantly dissimilar from the certainly genuine description of this kind in Galen, and may very well not merely reflect differences in the original (though admittedly that could well be the case with some other discrepancies between the two sources).

⁵⁷ Olympiodorus uses the term ‘interpolation’ where, in one instance, the parenthesis contains an explanation of what Aristotle is saying, in another, an illustration of it: in meteor. 41.23ff., 204.18ff. and 24ff.

⁵⁸ Elsewhere Seneca expresses a dislike of long an d complex sentence-structures (ep. 114.16). But it would be rash to employ Seneca straightforwardly either as an authority for or as an example of approved Stoic style. On the one hand, he can be starkly critical of Stoic style; on the other, the question of the sources for Seneca's style is a vexed and complicated one. It is rather a matter of saying that such-and-such a piece of Senecan writing or Senecan stylistics betrays signs of Stoic influence. His criticisms of the Stoa sometimes remind the reader forcefully of Cicero's strictures on the same subject: references in Griffin op. cit. (n. 37), p. 15, n. 1, who also has an informative survey of possible models for Seneca's own style (13ff.).

⁵⁹ Fronto, , ad M. Antoninum Imp. de eloquentiae liber II, ed. Hout, (vol. 1, Brill, 1954), §§16–17, pp. 139–40Google Scholar (= (part) SVF 2.27). The figures of thought in question seem to be: anticipation (prolepsis), question (interrogatio), personification (prosopopoeia), mimicry (ethopoeia), and perhaps vivid description (evidentia). Amplification is not so much a figure as a generic term for what figures enable the speaker to do: so e.g. Cicero, , de or. 3. 104ffGoogle Scholar. Chrysippus was in fact notoriously prolix: Diogenes 7.180, SVF 2.21, 883.

⁶⁰ The fragments of a book of this work are preserved as PHerc 307. It seems to be a more or less random collection of problems in dialectic (so Sedley, D. N., ‘The Negated Conjunction in Stoicism’, Elenchus 5 (1984), 311ff.Google Scholar, at 314: ‘The Logical Questions appears to consist in a series of aporetic problems in logic without any attempt at definitive solutions’). Form, content, and style all suggest that it may have been little more than a private notebook. (See my unpublished Cambridge Ph.D. thesis, ‘The Stoics on Ambiguity’, pp. 203ff.).

⁶¹ Galen's, de placitis Platonis et Hippocratis (ed. Lacy, P. De, 2nd. ed. with tr. and commentary, Berlin, 1981)Google Scholar preserves considerable fragments of these works; those of the de anima were assembled by von Arnim in what he believed to be their original order at SVF 2.911; those of the On the passions are collected at 3.461ff.

⁶² Cf. R.G. 3.11.18ff. Sp.

⁶³ For a survey of Theophrastus’ debt to Aristotle in matters of diction, delivery, and almost certainly style as well, see Innes, art. cit. (n. 7).

⁶⁴ Unfortunately the text is problematic; the most plausible reconstruction has Chrysippus refer to rhetoric's subject-matter as κόσμον εἐρομένου λόγου καὶ τάξιν order and arrangement of continuous discourse’. (In passing I would point to a possible connection with Gorgias 504b.) The phrase ‘continuous discourse’ (if it is authentic) recalls both the λόγοι ἐν διεξόδῳ of Diogenes 7.42 and the Stoic definition of rhetoric preserved by Sextus (M. 2.6). The description itself seems at once too vague and too restrictive: not sufficiently technical, yet excluding such matters as vocal modulation, gestures, and facial expressions, which Plutarch asserts Chrysippus did indeed discuss.

⁶⁵ Zeno's definition of ‘expertise’, ‘a system of apprehensions organised together to some useful end in life' (SVF 1.73), became standard in many later texts (e.g. Sextus, M. 2.10; SVF 2.93, 94, 95). Chrysippus' and Cleanthes' definitions (Olympiodorus, in Gorg. 12, pp. 69.26–70.3Google Scholar Westerink) are similar. Definitions of ‘science’ stress its stability and unchangeability, but never mention an ‘end’: Sextus, M. 7.151; Diogenes 7.47; SVF 2. 117. Stoic sapientia is not like ‘the other expertises’ because its end lies in activity, as with dancing or acting (de fin. 3.24–5); but this only confirms that expertises all have some sort of end. The sage is said to have an ‘expertise of life’ (e.g. Sextus, M. 11.199), but this synoptic expertise will not be on all fours with ordinary-range expertises, which each comprise a narrow set of theorems directed to a particular goal within the context of life as a whole.

⁶⁶ The most comprehensive survey of arguments against rhetoric's technical status (strictly, against its existence) is Sextus, M. 2; Quintilian presents and attempts to refute a fair selection at 2.15ff. and 12. Iff. For a favourable review of Sextus’ criticisms, seeBarnes, J., ‘Is Rhetoric an Art?’, DARG newsletter 2/2 (Fall 1986), 2ffGoogle Scholar.

⁶⁷ Sextus, , M. 2. 48–59Google Scholar sets out to prove that rhetoric has no subject-matter, ὔλη. At 56–7 Sextus gives three grounds on which the orator might be said to frame ‘fine language’, all of which he goes on to reject; the second and third together almost cover the area of the Stoic stylistic virtues (appropriateness alone is missing). Sextus concludes that ‘therefore rhetoric's job is not to produce fine language and speaking well (τὸ εὖ λέγειν)’ (58). One might at first glance think this a poor sort of argument against Stoic rhetoric; but Sextus might well regard it as a good tactic to attack the Stoa for paying even the feeblest attention to matters of style, and for restricting their ‘speaking well’ by any stylistic criteria at all. If this is Sextus’ main individual criticism of Stoic rhetoric, it could confirm that it did not have an ‘end’, and in particular that the Stoa did not associate rhetoric with the goal of persuasion. It is unclear to me whether Chrysippus' definition can be usefully linked with a passage from Sextus (M. 2.43), where ‘certain people’ are reported as holding that there are two sorts of rhetoric, one for the wise, one for ‘middling’ or ‘intermediate men’ (ἐν μέσοις ἀνθρώποις). I do not see how μέσος could be used by a Stoic as co-extensional with ‘ordinary, non-wise’: for it typically designates actions just insofar as they are performable by wise and non-wise alike, not such actions as performable or actually performed by just one of those jointly exhaustive and mutually exclusive groups.

⁶⁸ Quintilian's formulation of the ‘end’ of rhetoric adopts what was originally probably a Stoic scheme, but I doubt that his application is itself Stoic. Quintilian distinguishes between the ‘end’ of rhetoric, ‘speaking well’, and persuasion, which is only a desired consequence, the σκόπος (2.17.23). This does not seem to conform to the authentic Stoic end/σκόπος scheme: contrast esp.SVF 3.16, Plutarch, , comm. not. 1070fGoogle Scholar, Cicero, de fin 3.22. On the whole issue, see Alpers-Gotz, R., Der Begriffe σκόπος in der Stoa und seine Vorgeschichte (Hildesheim, 1976)Google Scholar.

⁶⁹ Of course the sage is not infallible in any straightforward sense. It is significant that Chrysippus’ formulation of the ‘end’ of life is ‘living in accordance with experience of what happens by nature’ (e.g. Diogenes 7.87). Even the wise cannot always predict what is fated to happen, and thus direct their choices accordingly (cf. Inwood, , op. cit. (n. 31), 203ff.Google Scholar); hence they too will sometimes assent to its being merely ‘reasonable’, εὔλογον, that such-and-such will be the case, and their infallibility actually lies in never failing, where circumstances require it, to preface the propositions to which they assent with this ‘operator’. Similar allowances may perhaps be built into expertises and sciences alike: some at least of their theorems will be guidelines, not substantive and invariable laws. Augustine (SVF 2.106) reports that the Stoics defined dialectic as the sollertia disputandi which, like Diogenes 7.48, confirms that it necessarily has an experiential element. Rhetoric might be subject to the same sort of limitations: for example, the sage cannot be sure, only be sure it is reasonable to expect, that this speech will be clear to this audience. But this provision does not help fix the distinction between rhetoric as expertise and as science.

⁷⁰ Quintilian may perhaps be reacting to the quietist movement in Stoicism that received a hostile response from Cicero and Seneca, and perhaps Panaetius. See further nn. 37, 42.

⁷¹ In describing the passions as weak and corrupt judgements 1 am of course following the Chrysippean moral psychology. A Chrysippean orator would see part of his task as extirpating excessive impulses and irrational judgements (the passions), and his (justified) appeals to the passions as incitements to (bad and vicious) judgements, whereas a Posidonian would rather call for the passions to be controlled by the rational part of the soul, or, in the special case under consideration, urge them against the dictates of reason. But the appeals to reason and the strict demands of morality would remain substantively the same; and it is therefore I think doubtful that this theoretical difference would show itself in the style of rhetoric.

⁷² One might be tempted to add that the wise cannot even be said to attempt persuasion, since no virtue can be exercised with varying degrees of success. (Many conventional definitions of the goal of the orator or of oratory employ such a ‘failsafe’: what the orator does is, not persuade, but ‘find persuasive material’ or ‘treat a political question as persuasively as possible’ or the like: cf. e.g. Quintilian 2.15.12,13,23ff.). Failure to persuade in any given case would not be damning, since it may only be reasonable, εὔλογον, not fixed and predictable, that such-and-such a speech or certain components of it will or will not bring success in these particular circumstances (cf. n. 69). I think a more formidable objection to this formulation of the Stoic orator's goal is that introducing the notion of the ‘plausible', πιθανόν, would grossly distort our apprehension of his true purpose, benefiting others. The use of falsehood and of appeals to emotions – that is, of mere persuasiveness – has a limited role in his programme, which as a whole is ethically impeccable. Further, Chrysippus stresses that the sage is in no way responsible for our being deceived (Plutarch, , st. rep. 1055E–1056AGoogle Scholar), and he surely could not be described as having the intention to deceive, that is, of trying to lead us to assent to what he knows to be false: what the sage wants is not assent, but action and impulse (1057B).

⁷³ So Antony, de or. 1.231, cf. Xenophon mem. 4.4.4; elsewhere (Tusc. Disp. 1.71), Cicero asserts that Socrates had magnitudo, greatness of mind, not superbia, haughtiness. For other noble Romans in the Stoic mould, see Griffin, , op cit. (n. 37), 248–9Google Scholar. Antony's account of the trial is fascinating. Rutilius and his fellow-pleaders have to toe the party-line (almost literally: no foot-stamping permitted), or else be denounced to the Stoic authorities(de or. 1.230). The tone is mocking – Antony's Stoa functions an an alternative state, the implacable and authoritarian guardian of morality to which offenders can be ‘denounced’ – but would lose its bite if Stoic orators did not have the ideal of calm and unemorive delivery and, I would suggest, language as well. On Rutilius, cf. Hendrickson, G. L., ‘The Memoirs' of Rutilius Rufus’, Classical Philology 28 (1933), 153–75CrossRefGoogle Scholar.

⁷⁴ A version of this paper was read at the meeting of the Southern Association for Ancient Philosophy in September 1987, and I am grateful to the participants for their comments on that occasion. I would also like to thank Jonathan Barnes and Donald Russell for their constructive written comments, and Michael Frede and the Editors for many invaluable suggestions and criticisms.