¹ Notice that psychological egoism is a descriptive thesis about what motivations people actually have and are capable of having, not a prescriptive thesis about what should move people. (Egoistic prescriptive theses often go by the name ethical egoism or rational egoism.) Of course, such descriptive theses supply constraints for any prescriptive theses one might want to develop: if people are incapable of being motivated by any considerations other than those of self-interest, there is no sense developing an ethical theory demanding that they sometimes be motivated by other considerations.

² So Vlastos, G., Socrates: Ironist and Moral Philosopher (Ithaca, NY, 1991), 203 n. 14CrossRefGoogle Scholar: ‘[For Socrates] desire for happiness is strictly self-referential: it is the agent's desire for his own happiness and that of no one else. This is so deep-seated an assumption that it is simply taken for granted: no argument is ever given for it in the Platonic corpus.’

³ See Morrison, D., ‘Happiness, rationality, and egoism in Plato's Socrates’, in Yu, J. and Gracia, J.E. (edd.), Rationality and Happiness: From the Ancients to the Early Medievals (Rochester, NY, 2003), 17–34Google Scholar; Weiss, R., The Socratic Paradox and Its Enemies (Chicago, 2006)Google Scholar; Ahbel-Rappe, S., ‘Cross-examining happiness: reason and community in Plato's Socratic dialogues’, in Nightingale, A. and Sedley, D. (edd.), Ancient Models of Mind: Studies in Human and Divine Rationality (Cambridge, 2010), 27–40CrossRefGoogle Scholar; and Ahbel-Rappe, S., ‘Is Socratic ethics egoistic?’, CPh 107 (2012), 319–40Google Scholar.

⁴ That is true even of some who have drawn the opposite conclusion when it comes to Plato's Socrates: Morrison (n. 3) argues that Plato's Socrates is not committed to psychological egoism, but he concludes that Xenophon's Socrates clearly is. This is all the more striking because elsewhere Morrison reports having drawn the latter conclusion reluctantly, thinking psychological egoism an unfortunate but undeniable feature of Xenophon's Socratic works: see Morrison, D., ‘Remarques sur la psychologie morale de Xénophon’, in Narcy, M. and Tordesillas, A. (edd.), Xénophon et Socrate: Actes du Colloque d'Aix-en-Provence (6–9 novembre 2003) (Paris, 2008), 11–28Google Scholar, at 14. For other important ascriptions of psychological egoism to Xenophon, see Delatte, A., Le troisième livre des Souvenirs socratiques de Xénophon, Étude critique (Paris, 1933), 116–17Google Scholar; Dorion, L.-A., ‘Akrasia et enkrateia dans les Mémorables de Xénophon’, Dialogue 42 (2003), 645–72CrossRefGoogle Scholar, at 651; Denyer, N., Plato: Protagoras (Cambridge, 2008), 129Google Scholar and 185; Bandini, M. and Dorion, L.-A., Xénophon: Mémorables, Tome 2, 1^re partie [Livres II–III] (Paris, 2011), 347Google Scholar n. 5; and Johnson, D.M., ‘From generals to gluttony: Memorabilia Book 3’, in Stavru, A. and Moore, C. (edd.), Socrates and the Socratic Dialogue (Leiden, 2018), 481–99Google Scholar, at 493.

⁵ Except where indicated, all quotations and references are to the Memorabilia and follow the recent Budé editions: see Bandini, M. and Dorion, L.-A., Xénophon: Mémorables, Tome 1 [Introduction générale, livre I] (Paris, 2010 ³; 2000¹)Google Scholar; Bandini and Dorion (n. 4); and Bandini, M. and Dorion, L.-A., Xénophon: Mémorables, Tome 2, 2e partie [Livre IV] (Paris, 2011)Google Scholar.

⁶ The text of the first sentence is a well-known crux. The MSS are divided between reading τὸ … τὸ and τὸν … τὸν, while Heindorf (whose reading is accepted by the Budé) proposes τῷ … τῷ. Despite the hesitant defence of τὸν … τὸν in Delatte (n. 4), 113–15, we continue to think that it makes the combinations of participle and infinitive too difficult to explain. Both of the other readings are awkward; but if one adopts τὸ … τὸ and interprets the infinitival clauses introduced by the articles as accusatives of respect, Heindorf's emendation seems unnecessary. In any event, the general sense of the passage is not in dispute, and the proposed variants do not measurably affect the interpretation we go on to offer.

⁷ In the OCT (Xenophontis Opera Omnia, vol. 2 [Oxford, 1921²; 1901¹]), E.C. Marchant adopts ἀκρατεῖς, though he apparently changes his mind and prints ἐγκρατεῖς in his Loeb edition (Marchant, E.C., Todd, O.J. and Henderson, J., Xenophon: Memorabilia, Oeconomicus, Symposium, Apology [Cambridge, MA, 2013] [Marchant, E.C. and Todd, O.J. (edd.), 1923¹]Google Scholar). Oddly, Marchant still translates his OCT text in the Loeb edition (a translation retained in the recent revision). Despite the popularity of ἐγκρατεῖς among earlier editors (on which, see Delatte [n. 4], 115 n. 3), it has the thinnest of manuscript support, appearing only in S (Ambr. E 11 inf.), where it is also corrected.

⁸ Rather than follow the common practice of identifying the narrator of the Mem. with its author, we hold that the narrator is a fictional character—in effect, the ideal Socratic. Nothing in our argument turns on that point, but we shall subsequently avoid presupposing that Xenophon is the narrator.

⁹ See Dorion (n. 4), 658–9 and passim; and cf. Bandini and Dorion (n. 5 [Tome 2, 2e partie]), 170–1 n. 2. The point was made earlier by Vlastos (n. 2), 100–1, but his argument depended on the misjudgement that interpreting ἀκρασία to mean ‘incontinence’ would create a conflict between 3.9.5, where the possibility of knowledge-based incontinence is effectively denied, and 4.5.6, where it would be affirmed openly. The problem actually lies in Vlastos's translation of 4.5.6, where he took Socrates to be declaring that ἀκρασία ‘often so stuns men that, though perceiving both the good and the bad, it makes them do the worse instead of the better’ (πολλάκις αἰσθανομένους τῶν ἀγαθῶν τε καὶ τῶν κακῶν ἐκπλήξασα [οὐ δοκεῖ] ποιεῖν τὸ χεῖρον ἀντὶ τοῦ βελτίονος αἱρεῖσθαι;). What Socrates says is better understood as an assertion that ἀκρασία ‘often so stuns their perception of good and evil that they choose the worse instead of the better’ (Marchant). (Hence the immediately preceding remark in the same sentence: ἢ οὐ δοκεῖ [sc. ἡ ἀκρασία] σοι προσέχειν τε τοῖς ὠφελοῦσι καὶ καταμανθάνειν αὐτὰ κωλύειν, ἀφέλκουσα ἐπὶ τὰ ἡδέα … ;) Read that way, the passage no longer conflicts with 3.9.5 regardless of the interpretation of ἀκρασία upon which one relies. None the less, we think that the context of 4.5.6–7 best supports taking ἀκρασία to designate self-indulgence rather than weakness of will. For discussion of the relevant issues, see Bandini and Dorion (n. 5 [Tome 2, 2e partie]), 172 n. 6 and 173–6 nn. 8–9.

¹⁰ Aristotle distinguishes the ἀκρατής from the ἀκόλαστος at Eth. Nic. 7.8 (and concisely at 7.9.1152a4–6): both pursue bodily pleasure, but only the latter thinks this to be what he ought to pursue. Like Aristotle's ἀκόλαστος, Xenophon's ἀκρατής does not feel conflicted in pursuing pleasure. Xenophon himself infrequently uses ἀκόλαστος and cognates. In the several occurrences outside the Socratic works it means ‘unpunished’. The sole case in the Socratic works is Mem. 2.1.1, where the phrase ἀκολαστοτέρως ἔχοντα (‘rather undisciplined’) offers a polite way of describing the indulgent tendencies of Socrates’ friend Aristippus.

¹¹ Delatte's (n. 4) interpretation of the passage is based on exactly the misunderstanding we are cautioning against. Note especially his citation of Aristotelian passages (Eth. Nic. 7.3.1145b25 and [Mag. Mor.] 2.6.1200b25) as evidence in favour of retaining the first occurrence of ἀκρατεῖς (115–6 n. 3)—a correct decision (see n. 7 above), though one made on faulty grounds. Despite the accurate accounts of the meaning of ἀκρατεῖς given by Dorion, the mistake persists in Denyer's commentary on the Protagoras (n. 4), 122 (on 329c7–d1). For a similar misinterpretation, see Bevilacqua, F., Memorabili di Senofonte (Turin, 2010), 143–4Google Scholar.

¹² For an even stronger statement of the interchangeability of the two terms, see the introduction by Bandini and Dorion (n. 5 [Tome 1]), 87–9 n. 91, and cf. Dorion (n. 4), 652–3.

¹³ The treatment of those figures in Mem. 1.2 merits a thorough discussion, especially in view of the comments at 1.2.19–23 regarding the possibility that one may lose whatever virtue one has acquired. Yet, we take it as relatively clear that neither Critias nor Alcibiades is represented as having achieved anything close to complete possession of the virtue. Hence 1.2.39: Κριτίας δὲ καὶ Ἀλκιβιάδης οὐκ ἀρέσκοντος αὐτοῖς Σωκράτους ὡμιλησάτην ὃν χρόνον ὡμιλείτην αὐτῷ, ἀλλ' εὐθὺς ἐξ ἀρχῆς ὡρμηκότε προεστάναι τῆς πόλεως. (Compare 1.2.14–16.)

¹⁴ See, for example, 2.1.1. Compare also 1.2.24 (τῶν μὴ καλῶν ἐπιθυμιῶν κρατεῖν), which should be read with 1.2.18 (σωφρονοῦντε).

¹⁵ At 1.5.4–5 Socrates speaks in a way that has been taken to suggest clearly that ἐγκράτεια is a precondition for virtue but not itself a virtue: ἆρά γε οὐ χρὴ πάντα ἄνδρα, ἡγησάμενον τὴν ἐγκράτειαν ἀρετῆς εἶναι κρηπῖδα, ταύτην πρῶτον ἐν τῇ ψυχῇ κατασκευάσασθαι; τίς γὰρ ἂν ἄνευ ταύτης ἢ μάθοι τι ἀγαθὸν ἢ μελετήσειεν ἀξιολόγως; (See e.g. Seel, G., ‘If you know what is best, you do it: Socratic intellectualism in Xenophon and Plato’, in Judson, L. and Karasmanis, V. [edd.], Remembering Socrates: Philosophical Essays [Oxford, 2006], 20–49Google Scholar, at 34; and Morrison, D., ‘Xenophon's Socrates on sophia and the virtues’, in Rossetti, L. and Stavru, A. [edd.], Socratica 2008: Studies in Socratic Literature [Bari, 2010], 227–40Google Scholar, at 234–5.) However, the foundations of a building are just as much part of the building as the walls and the roof, and Socrates’ remark that one should pursue ἐγκράτεια ‘first’ may simply underscore the idea, implicit in the metaphor of a foundation, that without ἐγκράτεια no other virtue is possible. What's more, the idea that no valuable learning is possible without it does not imply that it is not a type of learning. Compare the case of someone who says: ‘You'll never learn anything (else) unless you learn your letters.’ (Other relevant passages, notably 4.5.1–2 and 4.5.7–9, can also be read along much the same lines.)

¹⁶ For some references to proponents of the view that Socrates here identifies the virtues with one another, see Bandini and Dorion (n. 4), 351 n. 12. Socrates’ contention that all the virtues are forms of wisdom is silent on the question of the unity of the virtues in so far as it does not indicate whether the possession of one relevant form of knowledge entails the possession of any other. Dorion makes a similar point, but he goes further and maintains that 4.6 (esp. 4.6.7) implies that the relevant forms of knowledge are not mutually entailing. (See Bandini and Dorion [n. 4], 351–4 n. 12 and [n. 5 (Tome 2, 2e partie)], 192–4 n. 6.) We do not think that 4.6 speaks clearly to the issue. Dorion's interpretation of that chapter is central to the position he takes concerning 3.9.4–5 in his (n. 4).

¹⁷ Morrison (n. 15), 228–9 agrees that justice and every other virtue are identified as kinds of wisdom at 3.9.5, but he thinks that σωφροσύνη in 3.9.4 is moral virtue as a whole and is being identified with moral wisdom as a whole. The other virtues are thus parts of σωφροσύνη/wisdom. Dorion, by contrast, thinks that σωφροσύνη is a particular virtue, coordinate with justice and the others. (See Bandini and Dorion [n. 4], 344 n. 13.) Yet, he argues that οὐ διώριζεν must mean ‘does not separate’ rather than ‘does not differentiate’, and he explains the phrase as asserting that σωφροσύνη and wisdom are mutually entailing.

¹⁸ See 4.3, passim; 4.5.1; 4.5.11–4.6.1; and cf. 4.6.15, 4.8.11.

¹⁹ See Delatte (n. 4), 118 with n. 1. Delatte argues against another reading of the construction, whereby it means ‘equally F and G’. For that interpretation, see Döring, A., Die Lehre des Sokrates als sociales Reformsystem. Neuer Versuch zür Lösung des Problems der sokratischen Philosophie (Munich, 1895), 182Google Scholar. (For illustration of the two competing readings of the οὐ μᾶλλον-construction among the pre-Socratics, see DeLacy, P., ‘οὐ μᾶλλον and the antecedents of ancient scepticism’, Phronesis 3 [1958], 59–71CrossRefGoogle Scholar, at 60.) Delatte is correct to reject Döring's reading, since it would have the implication that the person under discussion is equally wise and unwise. While one could possibly make sense of such a thought—say, by supposing that the person is wise in the sense of knowing what should be done but unwise in the sense of none the less preferring what he takes to be in his interests—Socrates will go on at the end of 3.9.4 and in 3.9.5 to deny wisdom of any sort to a person who divorces his interests from his sense of what should be done. We discuss his reasons in what follows. For recent treatment of later philosophical uses of the οὐ μᾶλλον-construction, see (in addition to DeLacy) Bett, R., Pyrrho, his Antecedents, and his Legacy (Oxford, 2000), 30–3Google Scholar.

²⁰ See Delatte (n. 4), 116–17. Note that this emendation concerns a different use of ἀκρατεῖς from the one discussed in n. 7.

²¹ It is in fact dismissed as ‘francamente assurdo’ in Bevilacqua (n. 11), 253.

²² See, for instance, the translation by Tredennick, H. and Waterfield, R., Xenophon: Conversations of Socrates (New York, 1990)Google Scholar, where Socrates responds as follows to the question whether he would think the people under discussion ‘wise and weak’ (note the mistranslation of ἀκρατεῖς): ‘No more than I think them unwise and weak … . So I think that those who act wrongly are neither wise nor prudent.’

²³ Dorion's translation in the Budé edition (n. 4) treats the remark as if the adjectives ἀσόφους and ἀκρατεῖς were being compared with one another: ‘Ils ne sont pas moins dépourvus de savoir que de maîtrise de soi.’ We cannot see how the grammar warrants such an approach. Bevilacqua (n. 11) translates: ‘Non li ritengo niente di più che ignoranti e incapaci di dominarsi.’ Yet, it is difficult to explain how rendering οὐδὲν μᾶλλον ἤ as ‘nothing more than’ connects the thought adequately with the preceding question, represented by εἰ … νομίζοι. (A translation similar to Bevilacqua's is offered by Natali, C., ‘Socrates’ dialectic in Xenophon's Memorabilia’, in Judson, L. and Karasmanis, V. [edd.], Remembering Socrates: Philosophical Essays [Oxford, 2006], 3–19Google Scholar, at 8: ‘I think that they are nothing more than both unwise and lacking in self-control.’)

²⁴ Seel (n. 15), 38.

²⁵ For such a meaning, see, for instance, Thuc. 2.87.2 (cited in LSJ): καὶ οὐχὶ ἐς ναυμαχίαν μᾶλλον ἢ ἐπὶ στρατείαν ἐπλέομεν (‘we made the voyage not to fight a sea battle but for a land-engagement’).

²⁶ Seel (n. 15), 38 ignores this point and thus condemns a reading like Marchant's as awkward.

²⁷ Bandini and Dorion (n. 4): ‘je crois en effect que tous les hommes choisissent, parmi les actions possibles, celles d'ont ils s'imaginent qu'elles leur apportent le plus d'avantages, et que ce sont celles-là qu'ils font …’ . Tredennick and Waterfield (n. 22): ‘I presume that everyone acts by choosing from the courses open to him the one which he supposes to be most expedient.’ Bevilacqua (n. 11): ‘Credo infatti che tutti gli uomini, quando scelgono tra le varie possibilità, fanno ciò che ritengono più utile per loro.’ (Cf. also p. 139 of her edition.) Boys-Stones, G. and Rowe, C., The Circle of Socrates: Readings in the First-Generation Socratics (Indianapolis, 2013), 86–7Google Scholar (adapting a translation by K. Sanders): ‘I think that everyone chooses what he takes to be the most expedient among his available options and acts accordingly.’

²⁸ Perhaps the fullest expression and most vigorous defence of it occurs at Pl. Grg. 466a–481b. It is arguably also the main thesis of the Republic.

²⁹ See here Mem. 4.1.5, 4.2.38–9; Oec. 1.9–15.

³⁰ Rejecting the preference of Bandini and Dorion (n. 4) as well as Bevilacqua (n. 11) for Reiske's emendation of ἀρετῇ to σοφίᾳ. Dorion argues in the former work (350 n. 11) that the received text is at odds with the suggestion of the preceding lines that everything just and estimable is due to wisdom. Yet, the fact that things just and estimable are accomplished by means of virtue (ἀρετῇ πράττεται) is precisely what allows Socrates to go on to infer in the final clause of the sentence that virtue is identical to wisdom. (Dorion doubts that the latter conclusion really follows, but this is because he wrongly views the conclusion as being intended to follow directly from the first clause of the sentence, rather than from the first clause together with what has been said immediately before: ‘Or comment pourrait-on conclure, à partir de l’énoncé que le juste et toutes les choses belles et bonnes se font grâce à la vertu [ἀρετῇ πράττεται], que la justice et les autres vertus consistent en sophia?’) What's more, there is no tension between what is said here and the earlier contention that τά τε γὰρ δίκαια καὶ πάντα ὅσα ἀρετῇ πράττεται καλά τε καὶ ἀγαθὰ εἶναι. Socrates regards ὅσα ἀρετῇ πράττεται and καλά τε καὶ ἀγαθά as extensionally equivalent expressions, and that is why he can be casual about the order in which they occur.

³¹ Taking the referent of ταῦτα to be καλά τε καὶ ἀγαθά. Others interpret differently. For instance, Delatte (n. 4), 121 and Dorion (Bandini and Dorion [n. 4]), 349 n. 9 take it to be τά τε δίκαια καὶ πάντα ὅσα ἀρετῇ πράττεται. Greenwood, J. (Xenophontis Memorabilia Socratis [London, 1823], 140)Google Scholar takes it to be the whole proposition τά τε γὰρ δίκαια καὶ πάντα ὅσα ἀρετῇ πράττεται καλά τε καὶ ἀγαθὰ εἶναι.

³² On the translation of 4.5.6, compare n. 9 above.

³³ For a similar assessment in Plato, concerning the thesis that the one who knows what is right will never unjustly harm another, see Jones, R. and Sharma, R., ‘The wandering hero of the Hippias Minor: Socrates on virtue and craft’, CPh 112 (2017), 113–37Google Scholar, at 130–3.

³⁴ καὶ γὰρ βασιλεὺς αἱρεῖται οὐχ ἵνα ἑαυτοῦ καλῶς ἐπιμελῆται, ἀλλ' ἵνα καὶ οἱ ἑλόμενοι δι' αὐτὸν εὖ πράττωσι· καὶ στρατεύονται δὲ πάντες ἵνα ὁ βίος αὐτοῖς ὡς βέλτιστος ᾖ, καὶ στρατηγοὺς αἱροῦνται τούτου ἕνεκα, ἵνα πρὸς τοῦτο αὐτοῖς ἡγεμόνες ὦσι.

³⁵ Morrison (n. 4), 14–15 advances an egoistic reading along both of the lines just mentioned.

³⁶ Contrast how laboriously Plato works the point at Resp. 340c–347d.

³⁷ Of course, since Socrates thinks that doing what is right is at the same time in one's own interest, he has no need to portray good leaders as sacrificing themselves for the sake of their men. In focussing on others they also achieve what is good for themselves, a theme that is in fact sounded explicitly in Socrates’ conversation with Aristippus (2.1.17–19).

³⁸ See Bandini and Dorion (n. 5 [Tome 1]), xcix–cxviii.

³⁹ For written comments or helpful conversations about the issues discussed in this paper, we are grateful to audiences at Western University and the Tahoe Workshop in Ancient Philosophy, as well as to Adam Beresford, Don Morrison, Bryan Reece, Marco Romani Mistretta and an anonymous referee for this journal.