¹ An earlier version of this paper was read to the Australian Society for Classical Studies Conference at Melbourne in February 1985. I am grateful for helpful criticism to Dr. R Jackson of Melbourne University and Dr. H.B. Gottschalk of Leeds University.

² ‘Something like epic’ is the view held by most writers who attempt a decision. Among exceptions are Gigon, O., Gegenwärtigkeit und Utopie (Zürich 1976), 263–6Google Scholar who opts for a dramatist, as apparently does Koller, H., Die Mimesis in der Antike (Bern 1954), 19:Google Scholar ‘also ist der … der Schauspieler, “Darsteller des Schicklichen”’. An entirely non-dramatic poet is the option of Gadamer, H.G., Dialogue and Dialectic (New Haven 1980), 45,Google ScholarWarry, J.G., Greek aesthetic theory (London 1962), 58,Google Scholar and Vicaire, P., Platon, critique littéraire (Paris 1960), 223.Google ScholarAnnas, J., An introduction to Plato’s Republic (Oxford 1981), 99Google Scholar thinks it unclear whether any of the accepted poetry will be imitative. There is something to be said for the verdict of Nettleship, R.L., Lectures on the Republic of Plato (London 1958), 105Google Scholar that no clue is given as to genre.

³ 3380cl–2 and 7–8, 383a3, 387c9, 390al–2, 392al3, b5–6, d2, 394b9, 397c8.

⁴ Of course originality is not claimed for every part of this scheme. Every commentator gets the first set right, if often with misleading terminology, and in general the second set is clearly separated from it, though without exact explanation of the role of the ideal style of speech. The function and indeed the existence of the third, as I see it, have never been properly recognised. It is impossible to record credits adequately in the extensive literature on this topic, so that I am conscious of a tendency not to show appreciation of valuable discussions and to refer to some of them mainly to exemplify errors.

⁵ The tragedian imitates Oedipus in writing the part and so does the actor in playing it. Epic poet and rhapsode and teller of prose nursery tales imitate whenever direct quotation occurs in the narrative. The word sometimes, but not always, implies the independent existence of a model which is copied or a subject which is represented; sometimes it has an internal object, as when an actor imitates Oedipus, where ‘Oedipus’ is the content of the imitation. See Cross, R.C. and Woozley, A.D., Plato’s Republic, a philosophical commentary (London 1966), 271Google Scholar and Sörbom, G., Mimesis and art (Uppsala 1966), 119.Google Scholar

⁶ Koller (n.2), 17 believes that Plato leaves it open whether the guardians will personally perform tragedies, but that the behavioural sense of following a model or ‘doing like’ becomes prominent from 395 a. The view that spectators also imitate seems to me to be a short cut to giving the argument modern relevance. It is stated by Adam, J., The Republic of Plato² 1 (Cambridge 1963), 150,Google Scholarfollowed by Cross and Woozley (n.5), 272, and rightly criticized by Sörbom (n.5), 122–24 so far as the meaning of imitation is concerned, although he is overgenerous in allowing that there may be something in the general idea on the basis of Laws 643sqq. and656a-b. It is remarkable, if Plato thinks that spectators imitate, that the term is avoided when spectator reactions are discussed e.g. at Rep. 605d.

⁷ For a survey of the prevalence of music and dance in Greek life see Morrow, G.R., Plato’s Cretan city (Princeton 1960), 302–3,Google Scholar and for a discussion of the importance of poetry in the ‘song-culture’ of Greece and the conditions of performance of all types of poetry, see Herington, J., Poetry into drama (Berkeley 1985), 3–76.Google Scholar

⁸ A similar move occurs at Laws 654b–c: the agreement that the well-educated man will be able to dance and sing well needs to be supplemented by the condition that the songs and dances will be good.

⁹ Rep. 603c; Poet. 1448al.

¹⁰ Where might A come on the list? It does not fit easily, in that while it is intermediate as less desirable than B (i), it does not show increased amount and range of imitation. But what manner of man would prefer no imitation at all, if the good prefers little and the worse progressively more? The only reason implied for restricting imitation is moral, so that in practice A cannot figure as a preferred speech style, and Socrates’ scale from B (i) down to C is acceptable.

¹¹ 397b6–cl1 does not receive much attention from critics. Koller (n.2), 19 regards it as an incidental indication of the connection of Plato's new evaluation of poetry with Damon’s ideas about the moral effect of music. Else, G.F., ‘ “Imitation” in the fifth century,’ CP 53 (1958), 83–4Google Scholar regards it as an anticipation of the discussion of music, designed to show that speech and music are treated according to identical principles. Neither critic therefore sees it as integral to the argument leading to the selection of poets. The inclusion of metre as a subdivision of rhythm by Aristotle at Rhet. 1408b28 and Poet 1447a22–9 shows that ‘rhythm’ in our passage may refer to metre.

¹² with the dative, ‘to meet with’ or ‘chance upon’, cannot overlap in meaning with for it is not intensional, although the same verb with the genitive of the object aimed at conceivably might. The absence of the verb ‘imitate’ in this passage is decisive for the interpretation of 398b2, where doubt has arisen about the identity of the good man imitated.

¹³ Herington (n.7) 52–4 well imagines the impact of early poems regarded as utterances of the vividly characterized poet. If poetry is so regarded, it is natural for Socrates to think that all poetry coincides with the preferred speech-style of good men, worse, or bad. Nor is it hard to understand the awe with which the dramatist is viewed — he becomes all these others.

¹⁴ Poet. 1448al8, b25. See Else, G.F., Aristotle’s Poetics: the argument (Cambridge, Mass. 1957), 71–8CrossRefGoogle Scholar and Lucas, D.W., Aristotle, Poetics (Oxford 1968), 63–4Google Scholar for discussions of the social and moral dichotomy and its application to literature. It is worthwhile reminding ourselves when reading Rep. 3 that ‘good’ and ‘bad’ have strong overtones of ‘admirable’ and ‘low’.

¹⁵ One of a pair of co-ordinate adjectival ideas may follow the noun, cf. Adam (n.6) on 532b–c. In several of the passages quoted by Adam and those to whom he refers the deferred adjective contains a defining idea (e.g. Aristotle.4.P. 51.3 ‘that grain in the market which is unground’) or picks up an earlier idea (e.g. Demosthenes or. 18.270–71 [I will now talk about ‘the fortune of apparently all men, which is common’).

¹⁶ Among previous critics Nettleship (n.2), 105 comes closest to what I see as the truth, but his view needs to be modified to incorporate the degree of truth in the majority view. As with the discussion of speech, the subsequent account of song and dance applies to these as they figure in all genres but also to the genres particularly characterized by them.

¹⁷ Else, F., Plato and Aristotle on Poetry (Chapel Hill, N.C. 1986), 35–36,Google Scholar suggests that the man rejected is a ‘juggler’, a ‘seedy music-hall character’ whose general popularity serves as a warning of what indulgence in drama leads to, even though in himself he has nothing to do with tragedy or comedy. This suggestion, implausible in itself, is ruled out by the passage to which Else refers for support (Laws 658 c-d), which explicitly states that most people prefer the tragedian, not the conjurer, whose admirers are ‘the very little children’.

¹⁸ As is done by Annas (n.2), 99 and Belfiore, E., ‘A theory of imitation in Plato’s Republic,’ TAPhA 114 (1984), 121–46Google Scholar on p. 132.

¹⁹ Where does dithyramb fit in this scheme? It seems to have changed in the fifth century from a form characterized by extensive simple narrative to a later form which is remarkably mimetic. On the present scheme presumably the latter would be mixed, with a strong imitative element, but the simple form should lie outside the range B (i) to C. However, the tentative wording of 394c3 implies not that any dithyramb is entirely non-imitative but that dithyramb is the best place to find long stretches of unbroken simple narrative. In fact, it might be suspected that the old dithyramb would approach pretty well to the style recommended, but there are good reasons why Socrates does not use it. First, he has already allocated it as exemplar of non-imitative poetry; second, dithyramb as a choral lyric form is less easy to handle as an example of a style of speech, and it is poetry in which speech predominates that Socrates is concerned with here (see n.16); and third, the later dithyramb was notorious for mimetic profusion such as Socrates eschews, apart from being proverbially high-flown to the point of meaninglessness.

²⁰ Various sorts of ambiguity have been detected, e.g. Finsler, G., Platon und die Aristotelische Poetik (Leipzig 1900), 19Google Scholar finds it between a general sense in which all art is imitative and the special dramatic sense; Tate, J., ‘Imitation in Plato’s Republic,’ CQ 22 (1928), 18Google Scholar between a good and a bad sense; Vicaire (n.2) 223 between dramatic imitation and realistic copying; Belfiore (n.18), 126–7 between ‘versatile’, ignorant imitation and imitation with knowledge. Koller (n.2), 17–18 finds three: dramatic representation, following a model, and musical expression.

²¹ cf. 399d5.

²² The background is explored by Koller (n.2), Else (n.ll), Sörbom (n.5) and Halliwell, S., Aristotle’s Poetics (London 1986), 109–116.Google Scholar

²³ The oddity has been variously assessed, but the attempt to reconcile the two passages by taking the later one into account in interpreting the earlier, as is done by Tate (n.20) and Belfiore (n.18) seems to me to preclude proper understanding of Book 3, because it obliges ‘imitation’ to be taken in more than one sense within the compass of the topic for which the special, dramatic sense is devised. Nor can the contradiction entirely be removed by the correct observation that someone who is is not necessarily cf. Menza, V., Poetry and the Techrte Theory (Ann Arbor 1972), 162.Google Scholar All attempts founder on the rock accurately charted by Moss, L., ‘Plato and the Poetics,’ Philological Quarterly 50 (1971), 539:Google Scholar Plato ‘extends his original rhetorical sense of the term, denoting a narrative manner involving dialogue, to a far broader sense denoting all narration’.