I According to Stobaeus (n, 58) not all virtues are arts or sciences; but the traditional four virtues, ct>PÓll1/lTl� and courage and temperance and justice, all are. The Stoics distinguish the broad class of arts from the narrower class of sciences (which only the sage can have), but all of these virtues are sciences as well as arts. An art or science is always a state of the soul (more exactly, it is the ruling part of the soul disposed in a certain way); the Stoics (unlike the Pla- tonists, with their very different ontology), never consider a science "objectively" as a collection of theorems apart from a knower (sciences are ovra, while theorems are À(KTá). I abbreviate DL = Diogenes Laertius, AM = Sextus Adversus Mathematicos, SVF = Stoicorum Veterum Fragmenta (ed. von Arnim), DND = Cicero De Natura Deorum, TD = Cicero Tusculanae Disputa- tiones, DSR = Plutarch De Stoiconim Repugnantiis, DCN = Plutarch De Com- munibus Notitiis adversus Stoicos, PHP = Galen De Placitis Hippocratis et Pla- tonis, DG = Doxographi Graeci (ed. Diels).
2 The Stoics say not merely that virtue is an art of living, but also with special emphasis that it is an art (or a disposition) concerning the whole of life: so Stobaeus in defining virtue at n, 60, and again at it, 66-7, and Philo at SVF m, 202. 3 "Eù)(1i," AM xi, 208; the disappointment of this hope is crucial to Sextus' account of the motivation for scepticism, Outlines of Pyrrhonism I, 25- 30.
4 The Gorgias passage saying that things that can participate in either good or bad are themselves "neither good nor bad" is not, in context, saying that ev- erything other than knowledge is neither good nor bad. But Plato says this in the Meno passage cited, and more emphatically at Euthd. 281d2-e5. 5 Stoic sources routinely insist that there is no good without virtue, or that the good is "virtue or what participates in virtue" (this phrase DL vn, 94, Sto- baeus tt, 57 and u, 101). They also say that the good is "benefit or not other than benefit" (DL vu, 94, AM XI, 22); the connection between these two asser- tions is that what can exist apart from virtue (or from wisdom) can be abused, and what can be abused is not good (fairly explicit at AM xi, 61, DSR c.31). At least some Stoics cited Platonic precedent in disqualifying false "goods," for Antipater wrote a treatise in three books 7TEpt ro0 on Kara nÀ(ÍTwua fiovov T6 Kak6v aya86v (SVF Antipater 56, from Clement).
6 Plato develops the difficulty about the circularity of the epyov of virtue in different ways at Euthd. 291c4-292e5, Cleit. 408dl-410a6, Char. 171d1-175d5. Similar arguments are commonly used against Aristo (e.g., at Cicero De Finibus m, 12, m, 50), or against the Stoics in general, as at Plutarch DCN c.27, who cites the same phrase to mock the circularity of the opponent's answers ("6 At6T Kóp,v8oç," " apparently an endlessly repeated refrain in a children's song) that Plato had cited in the Euthydemus. Apparently Chrysippus had originally used these arguments (from Plato) against Aristo; Carneades then argued against the Stoics by using Aristo's arguments to show that the orthodox Stoic account of virtue must either reduce to Aristo's or else admit that things other than virtue are good, while using Chrysippus' arguments as a reductio ad ab- surdum of Aristo; Antiochus, Plutarch, and Sextus all use Carneades against the Stoics, while using him to support different positive positions. Aristo's own view seems to have been that virtue simply has no distinctive Epyov, and that the wise man is therefore unrecognizable: "for the sage is like a good actor, who, whether he puts on the mask of Thersites or of Agamemnon, will act ei- ther part 7TPOCíT/KÓVTl.IJÇ" (DL vim, 160). Aristo's view, here and on some other points, is reminiscent of some extreme predestinarian and antinomian views which keep surfacing on the fringes of Christianity (and of some other reli- gions) ; the orthodox Stoics' attempt to avoid these consequences has some things in common with Calvin's attempt to avoid antinomianism, and faces some of the same difficulties. I intend to return elsewhere to the theme of the circularity of the good in Stoic and anti-Stoic ethics.
7 Aristotle, almost alone among ancient philosophers, admits that some- thing good can harm or can have bad instances (so he can admit that pleasure, health, wealth are goods). But on the supreme good (both on God as pure knowledge and on Cíocpia as our knowledge of God), Aristotle's position is close to Plato's, and is involved in the same difficulties of circularity, Epyov, and 7rpOK07r'l]. I come back to Aristotle's treatment of these problems (to con- trast him with the Stoics) in Section Iv below.
8 See the Stoic definition of art, cited by many authors in very similar forms, collected at SVF i, 73 and u, 93-97. The collections of definitions at SVF n, 93 [pseudo-Galen] and a, 95 [Philo] help to bring out the contrast be- tween rEXvrl and È7WTT�¡.r..,.,. Only the wise have i7TIO-T4/A77, whereas fools too have Ka;�X4*civ (AM VII, 151-2); and fools can combine their KaTa,\4*cts into T€)(Mtt. for the Stoics never doubt the claim of ordinary arts like carpentry and medicine to be rExvac: TiXV77 is a preferred indifferent (DL VII, 106), while È7I'¡UT�¡.r..,., is a good (DL via, 98 and Stobaeus it, 73). (See also AM xi, 207 and Simplicius In Categorias p.224 and p.284 on the jneo-at re)(Mu as dis- tinguished from the virtues; Stobaeus n, 73 adds that riyyat when they occur in the sage are "altered by virtue and become å¡.r.f.Tå7l'TWTO'," which otherwise they would not be.) For each individual kataleptic impression as TgXV77-like see AM va, 252. 9 When the Stoics say that xaT6X77*tv is the criterion of truth, and that it is something intermediate between 866a and È7I'IUT�¡.r..,." and shared equally by the fool and the sage, their aim is to show that fools have sufficient resources to progress toward wisdom; and it is precisely to deny the possibility of progress toward wisdom that the sceptical opponents of Stoicism try to shoot down the doctrine of KaraarpJrcs, and in particular to deny that it can be common to the fool and the sage (so Sextus, following Arcesilaus, at AM vn, 150-53).
10 Stobaeus says (B, 88) that all 6plAai are assents; but he adds that although the same act is both an assent and a 6plA�, it is not an assent and a 6pui to the same thing. Rather, the assent to the proposition "it is appropriate to do X," is a ópp.7Í toward doing X, where "doing X" is a KaT77YOP?7/Aa or predicate. Stobaeus says that there are also non-practical opuai: a probable example would be assenting to the impression "I ought to assent to this impression, or to this type of impressions," which leads me only to assent, and not to perform any external action.
11 On the Stoic theory of 6pA�, and on the distinction between 6p)Ai and ape iv, see Inwood 1985. , t As Martha Nussbaum points out, 6pe$ts- and OpiY(CT8at are technical terms in Aristotle but not in Plato (who uses 6piyeo-Oat but not 5peetv); but pseudo-Plato Definitions 413c8-9 shows that 5piEetv was technical as the most general term for desire in the Academy. Nothing here depends on whether the Stoics read Aristotle's ethical works (though Chrysippus at least did so, DSR c.15) or on how much they cared about responding to them; the Stoics certainly read and responded to Old Academic writers, and Aristotle must serve us as their representative.
13 Epictetus thinks we can have rational (/(/()ucn<; (the contrary of 5peets!) even in our present condition; this makes sense, since we should be able to per- ceive the evils which are present to us, although not the goods which are not. 14 'E1n8v�ía is defined as akoyos 5peets (DL vn, 113, also Galen citing Chrysippus at PHP iv.2.3, iv.4.2, and pseudo-Andronicus IIepi aa9wv at SVF in, 391); the Stoics "say that opposite to ¿7T¿8v�ía is 6o�,k?)Ttg, which is tvko- yos opeys " (DL vn, 116; same definition cited Stobaeus n, 87, pseudo-Andronicus SVF 1u, 432, cp. pseudo-Plato Definitions 413c8-9); and every gpeets must be one or the other. BovÀ7JEV7Tå8f.ta (DL vn, 116 and the pseudo-Andronicus passage) and therefore can exist only in the sage (this point made emphatically-in Latin, but the terminological equivalences are clear—by Cicero, TD iv,12), as (7T,8v�ía and other passions exist only in fools. So non-sages cannot have rational 5peetv, though they can have rational 0�?}. (Aristotle too distinguishes Po6X?lqtv as rational opeys from Em9v�ia as irrational opeys> but for him, although fJovÀ7JT for what falsely seems good, or is truly opined but not known to be good, is not really fJovÀT/C1w.)
15 On "reservation" [ii1TE�aipEO"¿S'] see Inwood 1985, esp. 119-26. 16 For the denial of degrees of happiness or misery, see Stobaeus n, 98-100, Plutarch DCN c.33. Happiness and virtue are lha8iO"E'S' (DL vn, 89), and 8ia.dio-ets (as opposed to Zeciv) do not admit intensification or remission (Simplicius In Categorias p.237).
17 The last clause literally says "he would be a kind of king chosen by lot"-like the Athenian official called a j3aI, 345e5-346al, 346e7-347a6); but if he is already guaran- teed a sufficiency of goods, he has no motivation to override the motivation to exercise his art. If a true ruler is put in a position where he must rule, then he will rule well; he will not rule badly to benefit himself, even if he might prefer not to rule at all.
18 The Stoic definition of art (texts collected at SVF 1, 73 and u, 93-97) says not merely that it is a system of MtraX�tt? exercised together, but that they are exercised together "towards some riaos useful in life"; it is definitional of medicine that it is concerned with health, but also that it aims to produce or pre- serve health (cf. Galen De sectis ingredientibus c.l). An art, or reason in gen- eral, may be described as commanding the appropriate actions: an art "commands" what benefits the ruled (Republic 1346e3-7); ICa87íIColITa are "such things as X6yov convinces us to do" (DL vn, 105); vopos is "right aoyos com- manding what ought to be done and forbidding what ought not to be done" (SVF m, 332 [Clement] and parallels). 19 Phys. vm 4, 255a30-b29. The same idea underlies Chrysippus' compari- son of a person's disposition to the shape of a cylinder, which causes it to roll once it is given an initial push (Cicero De Fato 42-3, also SVF tt, 1000 [Aulus Gellius]): once the appropriate external circumstance sets it off, it is the thing's own internal gftv that determines its typical pattern of motion, whether the thing is a gravitating body or a rational animal acting according to a rational disposition. If the disposition is an art, it will lead us to assent to the appropri- ate impressions (and, if they are hormetic impressions, to act on them), when- ever we encounter the objects that the art is concerned with.
2o Metaph. O 2, 1046b'7-15. Here the contraries are not called good and bad, but only the positive (r6 1J7Tápxov) and its privation; Aristotle adds at Metaph. O 9, 1O51a10-15 that the positive contrary is good and the privation is bad. 21 The word "preferred" (used in contrast to "good"), does not by itself give any solution, since it does not explain how or why we should prefer these things; the considerations about the arts help to explain what kind of valuing is involved. Note that the Stoics give "the expert's appraisal" as one of the senses of "value" (DL via, 105, Stobaeus n, 83-4).
22 'E7UcrT1}P.11 is described as the kind of KaTåÀ111[1L<; which is dg, and/or a)ACT67rTWTOT (or a/ACT(i7TTWT09 into X6you), with one of these terms freely substituted for another, or two or all three conjoined, in the defini- tions of ('lTLCíT1}P.11 at DL vn, 47, Stobaeus IT, 73-4, Sextus AM vii, 151, SVF IT, 93 [pseudo-Galen], and SVF n, 95 [Philo]; whatever the sage believes he does da-(AaXCOT rcai f3(f3alw<; (Stobaeus n, 112); Zeno and Aristo and Chrysippus as- sert that virtue is X6yog onokoyovfievos Kai f3ÉfJaLo<; Kat ap,Eraarwros (Plutarch De virtute moraii c.3), and Cleanthes thought virtue could not be lost because of f3Éf3aLO& KaTaÀ1}1[Iu<; (DL vn, 127); the Stoics think that the greatest good is T6 al.I,ETQ.7frWTOY kv T0.CSKplO'EQL icat fJÉfJaLOlI, referred back to as dcr- 4XcLa Kat fJ(fJaLóT11<; (Plutarch DCN c.8); jmcrat riyyai (as opposed to the virtues) "fall short ofj3tj6cnM� EYEpyELY" (Simplicius In Categorias p.224), and Chrysippus says that the person who performs all the Ka81}Ko1ITa becomes happy only when these p.ÉcraL7rpd£«ty acquire TO j8t�8(nof (Stobaeus v,906-7). Compare Aristotle objecting, against what must have been a Platonic or Academic definition of ('lTL1171}P.11, that it is metaphorical (and thus improper in a definition) to say that cmoT7JjuT» is a'/AfTa'7rTWTO,v (Top. vi 2, 139b32-3). Even. earlier, Gorgias warns that 17 86�a, afeikepa Kat åf3ÉfJaLo<; oixra, a-q5akepaigKai åfJ(fJalo,<; (vT1rx.laL<; 7T«pij8dX\ee toot avril ypupAvovs (Helen, 11). ).
24 Note that on Chrysippus' view it always can be changed non-rationally, due to an attack of melancholy or the like (DL vn, 127). 25 To see why there might be a problem if we follow the dictates of oph- thalmology, consider the fragment of Herophilus On Eyes: "for those who can- not see in the daytime, twice daily rub on an ointment [composed of] gum, the manure of a land-crocodile, vitriolic copper, and the bile [gall] of a hyena made smooth with honey; and give the patient goat-liver to eat on an empty stomach" (Herophilus Fr. 260 von Staden, von Staden's translation). 26 As DL VII, 109 shows, all laia-a KaO�KOVTa have exceptions (axaflfjicov avtv 7repicrTa? is not a MtKa9nKOV that holds except under exceptional circumstances). A doctor who withholds the truth so that the patient will not flee from treatment does something (an action- type) that is not /(0.811/(0/.1, but does it Seovrws (SVF m, 513 [Philo]).
21 The Charmides is claiming (or rather, Socrates in the Charm ides repre- sents the fictional "doctors of Zalmoxis" as claiming) that the eyes themselves cannot be correctly treated without treating the whole body, and that the body itself cannot be correctly treated without treating the soul. The idea that dis- eases of the body can (and should) be cured by directly treating the soul is pre- Platonic (see Claus 1981) and Plato echoes this tradition here and elsewhere; but the present context is not entirely serious, and Plato may be verbally playing with the earlier claim that the health of the soul produces health in the body, while really meaning only that the health of the soul is intrinsically more important than the health of the body (because the soul uses the body, and better a sound artisan with damaged tools than an incompetent artisan with efficient tools). Here as elsewhere Plato contrasts the merely empirical doctor, who treats only the immediately affected part, with the ideal of the scientific doctor, who considers the whole body and understands how diseases are caused and why the appropriate treatments cure them (cp. Phdr. 270b-d).
28 Compare Plato reminding us that "all yevems comes to be for the sake of this, that the life of the whole [universe] should have a blessed existence; it does not come to be for the sake of you, but you for the sake of it. For every doctor and every skillful craftsman does everything for the sake of the whole: striving towards what is best overall, he produces the part for the sake of the whole and not the whole for the sake of the part" (Laws x, 903c3-dl, reading o-vvreivwv at c7). The word "craftsman" [Õ7lJ.uovPYó�] reminds us of the divine craftsman of the universe; we are urged to replace our own partial and self- centered perspective with his perspective, which is the objective perspective of the art of world-making or world-doctoring, and to judge the state of our own bodies by their contribution to the overall health of the world-body.
29 The parallel text of the pseudo-Plutarchan Placita gives vopos instead of X6yov in the second definition; both texts given by Diels, DG p.323.
3o So Cicero at DND II, 37 quotes Chrysippus as saying that "man has arisen for contemplating and imitating the world."
3� But, contra the usual Aristotelian position, cf. PA t 1, 640al-10, denying that physics is a BEW(J'nTLK7f ¿7UCTT�J.l.T/, and apparently implying that it is irot- 77TtK�.
3z DND n, 57-58, claiming to be following Zeno, contrasts other natures, which are merely arteficiosae, with the nature of the whole world, which is it- self an artifex. 33 So Cleanthes is reported as saying that ` just as all the parts of any one thing arise [0�eTa&] from the seeds at the appropriate [lCaÐ77lCova-,] times, so too the parts of the whole [universe], including animals and plants, arise [q>vETa,] at the appropriate [lCa9�lCova-,] times. And as certain k6yot of the parts are mingled together in the seed, and then are separated out when the parts are gen- erated, so all things [in the universe] are generated from one [the preceding conflagration, or the moisture into which it is transformed], and from all things they are collected into one [in the succeeding conflagration], methodically and harmoniously traversing the circuit [the life-cycle of the universe from one con- flagration to the next]" (Stobaeus 1, 45); compare Cicero, DND n, 81-86.
36 See Cicero TD m, 2 for the ideal of a course of education laid down by nature (which, in the world as it is, is frequently interrupted and corrupted by the false opinions, about the good or about what human actions are appropriate, that we pick up from human society). It is this ideal course of education, lead- ing up to virtue and to knowledge of the good, that Cicero is sketching in De Finibus m, 16-23. We must first use the impulses that nature has given us to acquire dispositions to appopriate actions, i.e., to selection of ra Kara (A�0-tV:
only then, once we have extended and stabilized these dispositions so as to live consistently with nature, does "what can truly be called good begin to be pre- sent in us, and begin to be understood," namely in the "the order and harmony of action" (m, 21) consequent on e7NOTij/xi7 of ra /(ŒTà CPVCTlv. By strengthen- ing our /(ŒTŒÀ.�"'EI� of ra Kara S56a-tv (which are non-goods) until we have (7TlITT�P.71, and then reflecting on the knowledge thus attained, we recognize that the knowledge itself, and the actions consequent on it, are good; this gives the Stoics the psychological mechanism that Plato lacks in Republic vn to ex- plain how we can make the transition from ordinary knowledge of non-goods to an extraordinary knowledge of the good. The o&Ketwo-tg-accounts are not ar- guments designed to show that a certain conclusion (the supremacy of virtue) is necessary, but narratives designed to show how a certain process (becoming virtuous) might be possible. I intend to discuss the structure and function of the 04KEtw(rtg-accounts at greater length elsewhere.
3� Aristotle distinguishes the two senses of "for the sake of which" [ro ov 9veKa], namely "to benefit whom" [Tb �)] and "to attain which" [Tb ovl, at Metaph. A 7, 1072bl-3; something unchangeable, like a god, cannot be TO but only TO ov. In ttjv rov 8(ot) dvapiav at EE VIJI 3, 1249bl7, 8(ot) is objec- tive genitive. The emendations adopted by Walzer and Mingay in the OCT (ri7v tov 8(ot) 8(wpíav into rilv TOO Octov 8(wpíav at bl7, and TbV 8(ov 8(p- a1T(t}(&V Ka't Oewpeip into T6 iv �1J.Îv 8(Îov 8(pa1Tdmv leal 8(wpfÏv at b20) are bizarre and must be rejected. 38 When Posidonius says that the riaov is "to live contemplating the truth and order of the universe, and cooperating in constructing it according to our ability, not being led in any respect by the irrational part of the soul" (Fr. 186 Edelstein-Kidd [from Clement]), he is echoing this passage of Aristotle (Aristotle goes on to say, just after the sentence cited, "it is thus in the soul, and this is the best standard for the soul, when it least perceives the irrational part of the soul as such"). When Posidonius says that in addition to contemplating the cosmos, we must also "cooperate in constructing it" [avyKaTaa-.Kev4Cttv], he may be deliberately modifying Aristotle; but it looks as if he has simply misun- derstood what Aristotle meant by 8(pa1T(V(¡V, taking it to be more than merely Oewpeiv.
39 I would like to thank Rachel Barney, Brad Inwood, Martha Nussbaum, and Gisela Striker for comments on earlier versions of this paper; also Rachana Kamtekar, for many discussions of what became major themes of the paper; and a class at McGill and audiences at Wellesley and Yale for letting me try out my ideas on them. I apologize for any incongruities between this paper and Martha Nussbaum's comments, due to changes I have made (in part responding to Martha's suggestions) since the paper and comments were originally deliv- ered.