Julius Tomin

The belief that the late dating of the Phaedrus has been established on stylometric grounds became a dogma firmly held by the majority of Platonic scholars throughout the whole of twentieth century. In this chapter I shall argue that contrary to this belief stylometric data are not only consistent with the ancient dating of the Phaedrus, but that they can be best explained on its basis.

At the very beginning of the twentieth century G. Janell published the results of his research into Plato’s treatment of hiatus (1). He found that on this criterion the dialogues are divided into two groups and he maintained that within each group its use as a means of chronology is irrelevant. In the one group, comprising the majority of dialogues, hiatus varies from 23.9 times per page of the Didot edition of the Phaedrus to 46.0 times in the Lysis. In the other group, consisting of the Laws, Philebus, Timaeus, Critias, Sophist and Politicus, hiatus varies between 5.9 in the Laws and 0.44 in the Politicus. These results were welcomed as a confirmation of the late dating of the Phaedrus, for this dialogue exhibits the lowest frequency among the dialogues of the first group, and in this respect is the nearest to the late six, but as Cherniss later remarked, Plato consciously avoided hiatus in none of the first group and in all those of the second (2). The question is, how the elderly Plato succeeded with apparent ease in avoiding hiatus, when he made the decision to do so. No one appears to have considered this question except Thesleff, who remarks that the avoidance of hiatus was an Isocratean mannerism ‘unlikely to have been adopted by the aged Plato’ and therefore attributes it to ‘Plato’s secretary’ (3). However it is hardly likely that Plato’s secretary could have restructured each sentence so as to avoid hiatus while writing to Plato’s dictation, and even less likely that Plato would have permitted this person to rewrite the dialogues in an Isocratean manner. Yet Thesleff put his finger on a real problem, which requires explanation. The ancient biographic tradition offers us two pieces of information, which can help in finding a solution. For Diogenes informs us that before attaching himself to Socrates Plato wrote poetry, dithyrambs, lyric poems, and tragedies (iii. 5), which means that Plato in his youth cultivated the poet’s skill of avoiding hiatus, and this training, although not consciously exercised, left its traces in the Phaedrus, his first dialogue (iii. 38).
In 1904 W. Kaluscha published the results of his investigation of rhythmic clausulae, that is of rhythmic endings of sentences, in Plato’s dialogues. On this criterion the dialogues were divided into two groups as in Janell’s investigation; the late six in which Plato favoured and avoided certain clausulae, and the rest of the dialogues in which he had little concern for such rhythmic endings. Yet in one respect there is a marked difference; it is clear that in all of the late six dialogues Plato avoided hiatus with ease, for the figures are uniformly low, but concerning the rhythmic clausulae this was not the case. In the Timaeus and Critias he appears to be in search of rhythmic clausulae with which he could be satisfied, but he is still only searching. He begins to show a clear preference for certain rhythms in the Sophist, and becomes proficient in achieving the desired rhythms in the remainder of the late six, that is in the Politicus, Philebus, and Laws. As Brandwood puts it, Plato’s new rhythm ‘did not spring perfect from his head, but developed through succeeding works’ (4). For this task Plato was not prepared by his early literary exploits.

Instead of referring to the biographic tradition in order to make sense of the stylometric data obtained by Janell and Kaluscha, Platonic studies took a very different course. In 1920 L. Billig argued that Kaluscha treated the Sophist as a rhythmically homogeneous work, which in his view it was not. He drew the dividing line between the late dialogues and the rest of Plato’s works through the middle of the Sophist. He claimed to have found that the first and the final sections of the Sophist, devoted to dichotomy, exhibited none of the prose-rhythms characteristic of Plato’s late style, whereas the digression devoted to Being and Not-Being did (5). On his data, the rhythms of the digression are the same as in the Politicus, whereas those of the main part of the work resemble the rhythm of the Timaeus; the Timaeus in its turn is characterized by equability of rhythm which it has in common with the Republic (6).

In the 1950s G. E. L. Owen published his thesis that Plato in his most mature stage abandoned the Forms in favour of a conceptual analysis somewhat akin to the analytical philosophy then cultivated at Oxford and Cambridge. Unhappy about the position of the Timaeus, in which the theory of Forms is prominent, among Plato's late dialogues, and relying on the correctness of Billig’s investigation concerning the rhythmic clausulae, he separated the Timaeus from the late six, attaching it closely to the Republic, and placed the Phaedrus, Parmenides, and Cratylus together with the Sophist, Politicus, Philebus and Laws in the group of dialogues belonging to the last, critical, and most mature stage of Plato’s philosophy (7). As will be seen, Owen was mistaken in his reliance on Billig.

In 1958 Brandwood defended his ‘Thesis’, in which he surveyed and tested stylometric works of the past, finding in them serious flaws, such as mistakes concerning the simple calculation of frequency of chosen particles, on which the allotted chronological position of Plato’s dialogues was based. He found mistakes even in Janell’s investigation of the frequency of hiatus and in Kaluscha’s investigation of rhythmic clausulae, but in these two cases the mistakes found were minor, and did not impede the correctness of the obtained results as far as the stylometric proximity of investigated dialogues to each other was concerned. But he himself made a curious mistake when he wrote in reference to Owen: ‘Because his interpretation of Plato’s philosophy required a considerably earlier position for the Timaeus than that traditionally allotted to it, Owen accepted the results of Kaluscha’s and Billig’s investigations.’(8)
This statement is misleading, for Kaluscha had firmly maintained that the Timaeus belonged to the group of the late six dialogues (9). Moreover,  Brandwood found that the Timaeus was even closer to the Laws than Kaluscha had maintained. For although Kaluscha noticed that in the Timaeus as in the Sophist three of the five highest rhythmic frequencies coincide with those of the Laws, he failed to observe that in the Laws Plato avoided hexameter endings, which according to Brandwood are in the Timaeus slightly less common than in the Sophist (Brandwood 1990, p. 173). Concerning the Phaedrus, Theaetetus and Parmenides,  Brandwood asks whether these dialogues belong to the late group on the criterion of the rhythmic clausulae, and replies: ‘No, Kaluscha declared, “since they have absolutely nothing in common with it; they belong to the earlier period when Plato had little concern for the rhythmic clausulae”.’ (Brandwood 1958, p. 306)

‘having built up the hypothesis of a late insertion [of the ontological digression in the Sophist] on the basis of an observed jump in the frequency of the main rhythm after 236c, Billig was compelled to fabricate a similar break to make the end of the inserted part. Nor, it seems, was this the only thing that he fabricated’. (Brandwood 1958, p. 328)

These harsh words are omitted from Brandwood’s published version of the ‘Thesis’. Instead, we find there the following words: ‘It seems reasonable to accept Billig’s statistics in general as a trustworthy foundation on which to base conclusions about the chronological sequence of the works involved’. (Brandwood 1990, p. 186)

And so one must ask: was it not because of Brandwood’s well based criticism of Billig that his ‘Thesis’ had to wait for thirty two years for its publication? Was the ‘re-evaluation’ of Billig a condition for its publication? In his Preface Brandwood writes: ‘Professor M. F. Burnyeat first invited a revision of the thesis, then contended unstintingly for its publication.’ (Brandwood 1990, p. x)

Brandwood’s ‘Thesis’ remained unpublished for three decades, and during this time a myth appears to have spread far and wide that his stylometric investigations supported Owen’s thesis. It can be found in a comprehensive review of stylometry by H. Thesleff, a Finnish Platonic scholar, who was led to believe that Brandwood claimed to have reinforced Owen’s view ‘beyond doubt’ (10).  It infected J. B. Skemp’s authoritative survey of Platonic studies (11). It guided M. C. Nussbaum's hand when she wrote that ‘various [stylometric] criteria used independently converge, consistently placing it [the Phaedrus] just before, but close to, the group Sophist, Statesman, Philebus, Laws' and added that ‘some would include Timaeus - but L. Brandwood’s work suggests doubt in this case’ (Nussbaum, p. 470, n. 5; the italics are mine, J. T.). It led Platonic scholars to believe that Brandwood’s ‘Thesis’ contained ‘definitive’ results (12), yet nothing can misrepresent it more than such a view. His aim was to clear the ground for future stylometric research, having demonstrated that the stylometric research in the past was very unsatisfactory. He concludes his work with the words: ‘There is no need to despair yet of the efficacy of the stylistic method, since compared with what can be done what has been done is very insignificant’. (Brandwood 1958, p. 425)

The dialogues in the second group Brandwood ordered chronologically as follows: Republic, Parmenides, Theaetetus, Phaedrus, for ‘the order shown seems to be most probable’(Brandwood 1976, p. xviii). For evidence he refers to his ‘Thesis’, but he omits to say how unsatisfactory he had then found the data concerning the middle group: ‘the evidence relating to the middle dialogues is insufficient to allow any deductions to be made about the order in which they were written’ (14). Nevertheless, Brandwood in his ‘Thesis’ did place the Phaedrus in the middle group, and so I must ask myself, how can I square this with my dating of the Phaedrus?

Earlier in this chapter I took recourse to the ancient biographic tradition on the dating of the Phaedrus to explain its proximity to the late six dialogues concerning the frequency of hiatus on the one hand, and its distance from them concerning the rhythmic clausulae: Plato's poetic beginnings meant that he was well trained in avoiding the hiatus and this skill left its trace in his first dialogue. The choice and cultivation of certain rhythmic endings of sentences and the avoidance of other such endings was a new development, and it required a new skill to which Plato applied himself in his late six dialogues. But why did he take the trouble? This problem may be solved if we take account of the stylometric proximity of the Phaedrus to the Republic, Theaetetus, and Parmenides against the background of the ancient biographic tradition on the one hand, and of Plato's reflections on the art of imitation in the Republic on the other.

Diogenes Laertius recounts that Plato was about to compete for the prize with a tragedy (mellôn agônieisthai tragôidiai), but when he listened to Socrates in front of the theatre of Dionysus he consigned his poems (ta poiêmata) to the flames, becoming a pupil of Socrates from then on (Diogenes Laertius iii. 5-6). The term ‘his poems’ presumably refers to a set of three tragedies accompanied by a satirical piece, which were a regular feature of dramatic competitions throughout the fifth century. This information relies on a trustworthy source, namely Dicaearchus (iii, 4). We may therefore safely assume that Plato had developed a style of his own prior to introducing Socrates into his writings. It is equally plausible to assume that in his first dialogue Plato did not succeed in liberating himself fully from his original style on the one hand, and in fully appropriating Socrates’ voice on the other, especially since the theory of Forms, so that in presenting it in the Phaedrus, he was presenting his innermost thoughts through Socrates’ mouth.

If Plato at a certain point decided to give up on imitating Socrates’ voice, we may suppose that he quite naturally reverted to some aspects of his pre-Socratic style, which left its mark on his first dialogue. In the third book of the Republic we can see Plato giving vent to his despondency concerning imitative writing. When he says that a decent man, who in his narrative (en têi dihgêsei) arrives at some saying or action of a good man, will be willing to communicate these by impersonating that man’s voice (ethelêsein hôs autos ôn ekeinos apangellein) and not be ashamed of this kind of imitation (kai ouk aischuneisthai epi têi toiautêi mimêsei, 396c5-8), it sounds like an apology for having impersonated Socrates for so long. That he is indeed serious about turning his back on imitation as such he makes clear in the tenth book, where the writer of tragedies is denounced as ‘being thrice removed from the king and the truth’ (tritos tis apo basileôs kai tês alêtheias pephukôs, 597e7), and where together with him all the other imitators (kai pantes hoi alloi mimêtai, 597e7-8) are in the same way dismissed. Since Plato began the Republic with Socrates as the main speaker, he could not but persevere with him throughout the dialogue; but we can see from stylometric data that separates this dialogue from the earlier ones, that he began to free himself from the style in which he was true to Socrates’ voice. The same considerations apply to the remaining two dialogues put by Brandwood in the middle group together with the Phaedrus and the Republic, that is the Thaeatetus and the Parmenides, for they were written either concurrently with the Republic, or slightly later. Furthermore, the Theaetetus is dramatically associated with two dialogues in which Socrates is pushed into the background and plays a role of a mere listener and learner, that is the late Sophist and Politicus, and in the Parmenides Socrates speaks only in the introductory part.

This hypothesis can be further strengthened by taking into account the nature of stylometric data that link together the Republic, Theaetetus, Parmenides, and Phaedrus. In the ‘Conclusion’ to his ‘Thesis’ Brandwood writes:

‘The middle group of dialogues, Republic, Parmenides, Phaedrus, Theaetetus, is distinguished from the early dialogues either by the occurrence of new expressions (mainly reply formulae) or by the use of these expressions becoming regular for the first time. Examples of these are ti mên, alêthê, pantapasi men oun, dêlon, kai mala, alêthestata legeis, orthôs, orthotata legeis, pôs, pêi, poion’. (Brandwood 1958, p. 402)

Most of these expressions are expressions of certainty, and are thus characteristic of dialogues in which Plato aimed at setting aside Socrates’ philosophic ignorance and the philosophic uncertainty accompanying it. Aristotle’s account of Plato’s conception of the theory of Forms as a direct result of the impact of Socrates’ fixing of mind on definitions on the young Plato steeped in Heraclitism (Met. 987a32-b8) makes it clear that Plato was convinced that with his conception of Forms he achieved certainty and acquired true knowledge. In his first dialogue, the Phaedrus, Plato quite naturally used language that expressed his philosophic certainty, in contrast to his following early dialogues, in which he became better at impersonating Socrates stuck in the uncertainties of his not-knowing.

In 1989 G. R. Ledger published Re-counting Plato, in which he based his stylometric investigations on three groups of variables, 37 in total, all of which were letters of the Greek alphabet. The first group consisted of 19 members, which he called ALETs, and these consisted of the 24 letters of the Greek alphabet; six of the letters occur so infrequently, that he comprised them all into one variable. These nineteen variables recorded the percentage rate of the occurrence of words according to their overall letter content; a word is counted only if it contains the specified variable; if the letter occurs twice or more times in a word, the count for that word is still only considered to be one. The second group consisted of the nine most frequent word endings, which he called BLETs, and the third group of the nine most frequent penultimate letters.(15) The work was very badly received. P. Keyser ends his review with the conclusion ‘that despite a noble attempt, the work is carved from rotten stone and crumbles at the touch’ (16); T. M. Robinson is less harsh, but equally dismissive (17). But pace these two reviewers, I find Ledger’s work extremely interesting; one would think that percentages of word counts based on letters the words contain would ‘merely exhibit random noise’ (Ledger, p.9), but cluster analysis of these variables, which is an entirely neutral approach, making no assumptions about the provenance of the samples, assembles the samples into groups that to a surprising degree correspond to works from which they are taken. Furthermore, the data obtained by Ledger dramatically reaffirmed the division of Plato's dialogues into two stylometrically different groups of dialogues, and for this reason alone deserves serious consideration.

The method Ledger used allowed him to compare works of seven Attic prose writers: one speech by Aeschines and Isaeus, five speeches by Isocrates, one by Lysias, all works of Plato longer than 1000 words, three books of Thucydides’ Histories, Xenophon’s Memorabilia, Oeconomicus and the first book of his Histories (Ledger, p. 41). The cluster analysis misclassifies certain samples, which is the fact on which Keyser founds his main objection against Ledger’s work. The misclassification affected in particular certain works of Xenophon and certain dialogues of Plato.  Keyser quotes as the worst case the test in which Ledger in his ‘Preliminary Survey’ subjected to cluster analysis 8 samples from Plato’s Apology, 15 from Protagoras, 8 from Republic I, 30 from Xenophon’s Memorabilia and 15 from his Oeconomicus. Ledger says that apart from the Apology, the 8 samples of which are neatly grouped together right in the middle,

‘there is no clear evidence of individuality in any of the other works. Broadly speaking the Platonic samples are grouped on the right of the diagram and those of Xenophon on the left, but there is evidently much room for confusion, with 14 of the Xenophon samples appearing with the Plato cluster, or nearly one-third. On the other side the confusion is not so widespread, with only four out of the 31 Plato samples being placed amidst the main Oeconomicus and Memorabilia cluster’. (Ledger, pp. 68-70)

Keyser remarks ‘My conclusion would be - Ledger has clearly proven that his simple letter-counting scheme does not work’.(Keyser, p. 3). The matter looks very different, if we pause to think what all these works of Plato and Xenophon have in common: Socrates is the main character and the main speaker in all these works.

In addition, Keyser holds it against Ledger that ‘Xenophon’s Hellenica is sufficiently different from Xenophon’s Memorabilia and Oeconomicus that the two sets appear to be different authors’ (Keyser, p. 3). But the fact that the stylometric data obtained by Ledger compelled him to ascribe the Memorabilia and Oeconomicus to an author called Xenophon 1 and Hellenica to Xenophon 2, and that Plato's works also formed two groups so stylometrically different, as if written by two different authors (designated by Ledger as Plato 1 and Plato 2), is by far the most important finding in Ledger's investigation. Ledger's Plato 2 comprises Plato's six late dialogues Critias, Laws, Philebus, Statesman, Sophist, Timaeus, plus Clitopho, Epistles 3, 7, and 8, whereas all the other dialogues belong to Plato 1 (Ledger, pp. 106-7). The most surprising, and for Ledger deeply annoying and inexplicable feature of his work was that his discriminating analysis could not meaningfully differentiate between Plato 1 and Xenophon 1 (Ledger, p. 162). The current scholarly atmosphere in Platonic academic circles did not allow him even to contemplate Socrates as the obvious factor that Xenophon 1 and Plato 1 have in common.

Deeply steeped in modern theories of Plato's development, Ledger finds it disconcerting that the Clitopho belongs to the group of the late six, for in this short dialogue Clitopho criticizes Socrates' unwillingness or inability to tell him what justice is (ê ouk eidenai se ê ouk ethelein autês emoi koinônein, 410c5-6), and what is the true product that a just man can produce (ho dunatai poiein hêmin ergon ho dikaios, ti touto phamen, 409b8-c1), while this is precisely  what Plato’s Socrates did accomplish in the Republic. Ledger complains that ‘in this context Clitophon appears to be rather meaningless’, that is if written after the Republic, but he points out that ‘the stylometric evidence is quite decisive in placing Clitophon with the late dialogues’ (Ledger, p. 207). Let me here note that Ledger was fully justified in trusting his stylometric data concerning the affinity of the Clitopho to the late six; long before him, a German scholar Heinrich Bruennecke investigated the frequency of hiatus in this dialogue and found that on this criterion the dialogue belongs to the group of the six late dialogues (18). But Ledger was equally justified to feel uneasy about it. The Clitopho is a conundrum that cannot be solved within the accepted interpretative framework that rejected the ancient biographic tradition according to which Plato began to write dialogues prior to Socrates' death. A solution of this conundrum will be attempted in a chapter devoted specifically to the Clitopho, so let me merely note here that my acceptance of the ancient biographic tradition allows me to propose a different dating of the Clitopho, one fully consistent with Ledger’s stylometric data. Socrates reacted to Plato’s picture of him in the Phaedrus by emphasizing his ignorance, and Plato in the subsequent dialogues written prior to Socrates’ death struggled more and more desperately against Socrates’ ignorance. This struggle culminated in the Clitopho (19). Since Clitopho in the dialogue expresses Plato’s own criticism of Socrates, Plato wrote it in his own style, that is the style to which he reverted when he decided later in life to finally turn his back on imitating Socrates’ way of speaking. It is worth noting that the list of the eight works closest to Clitopho in descending order in Ledger’s Table 8.21 is as follows: Laws, Ep. 7, Philebus, Phaedrus, Timaeus, Republic, Sophist, Epinomis. The table shows that apart from the late Laws, The Seventh Letter, and the Philebus, the closest work to Clitopho on Ledger’s data is the Phaedrus, which accords with my hypothesis that Plato in his late six works reverted in many ways to his early style, the style which he developed before he began to impersonate Socrates in his dialogues, and of which we can see traces in the Phaedrus, that is his first dialogue, and in the Clitopho, in which Socrates speaks merely six lines, while the remainder of the dialogue contains Plato’s critical assessment of the limitations of Socrates’ approach to philosophy, face to face with the living Socrates.

In 1990 Leonard Brandwood finally published his revised doctoral ‘Thesis’, enriched by a chapter devoted to D. Wishart’s & S. V. Leach’s study of Plato’s prose rhythm. These two authors subjected ten works of Plato to their analysis: Phaedrus, Symposium, Republic, Timaeus, Critias, Sophist, Statesman, Philebus, Laws, and The Seventh Letter. Their analysis allowed them to compute the distances of these dialogues from one another, which they interpreted as  relevant chronologically; on their results the Phaedrus is by far the earliest of the ten works under investigation:

‘We find no stylistic evidence to support the present theory that the Phaedrus is a homogeneous late work. On all our diagrams it completely opposes the Laws, which we take to be the latest work considered, and since the other samples fall into a chronological order between the two, we are obliged to place the Phaedrus early.’(20)

Brandwood invalidates Wishart’s and Leach’s early dating of the Phaedrus on the following ground:

‘Regarding the chronological order arrived at by the authors there is a serious objection to the position allocated to the Phaedrus. All five samples of this work were taken not from the dialogue section, but from speeches: one from that attributed to Lysias, one from Socrates’ first speech and three from his second. However, it has already been mentioned (p. 160) that Socrates’ two speeches are specifically indicated by Plato as poetical in character, including their rhythm, a fact born out by observation. It is not surprising, therefore, if in a comparison with the late works the Phaedrus shows less similarity than do the Symposium and Republic, and in these circumstances it is a mistake to infer that it must be earlier than they. The failure of the authors to take account of the multiplex nature of the Phaedrus has also led to their absurd conclusion that the Lysias speech was a later composition than those of Socrates, although the latter stem from and are a criticism of the former.  This aberration on their part does not necessarily invalidate their findings in respect of the sequence of the later works ...’ (Brandwood 1990, p. 247)

Describing the early dating of the Phaedrus as an aberration, Brandwood joins the chorus of twentieth century Platonists; so let me examine his arguments. Firstly, to justify his allegation that Wishart and Leach did not take into account the poetical character of the Phaedrus Brandwood refers back to p.160 of his book, but all that he says on p.160 concerning this point is to the contrary of what he wants:

‘If the attribution of the low frequency of hiatus in the Phaedrus to its rhetorical-poetical tenor were correct, one would expect the incidence of hiatus in Socrates’ two speeches to be somewhat lower than in the dialogue section, since they are by definition rhetorical as well as being specified by Plato himself as poetical. Yet most of the passages noted above in which hiatus appears to be avoided more carefully than usual occur in the dialogue section, and hiatus in general, however calculated, can hardly be said to be significantly lower in the two speeches. The ‘poetical’ character of the Phaedrus, then, does not seem an adequate reason for the comparatively low frequency of hiatus.’

If the stylometric proximity of the Phaedrus to the late six dialogues concerning the frequency of hiatus cannot be attributed to its poetic character, in spite of what may have been alleged by many scholars without any proof, as Brandwood showed, how can he now discredit Wishart’s and Leach’s investigation on the grounds that the stylometric distance of the Phaedrus from the Symposium and the Republic, which Wishart and Leach interpret as indicative of chronological distance, is due to the ‘poetical’ [I retain Brandwood’s inverted commas] character of the dialogue? Let me note that Brandwood’s finding – that most of the Phaedran passages in which hiatus appears to be avoided more carefully than usual occur in the dialogue section – can be best explained on the basis of the ancient tradition according to which Plato in his early days wrote tragedies.

In addition, Brandwood accuses Wishart and Leach of ‘their absurd conclusion that the Lysias speech was a later composition than those of Socrates, although the latter stem from and are a criticism of the former’. Had the authors arrived at this conclusion, it certainly would have been an ‘aberration’, as Brandwood puts it. But the reader will look in vain for any such conclusion in Wishart’s and Leach’s article. In the abstract prefaced to the article they say: ‘it is further suggested that the “Lysias speech” from the Phaedrus could have been a genuine “epideiktic paignion” by Lysias, rather than Platonic parody (Wishart and Leach, p. 90).’  And in the ‘Conclusions’ they write ‘There is some evidence to suggest that the “Lysias speech” is not closely related to the other passages from the Phaedrus, and it is possible that this passage may have been written by Lysias rather than Plato’ (Wishart and Leach, p. 98). Brandwood was well aware of this, for on p. 241 of his book he wrote that concerning the Lysias speech the authors

‘considered that in view of the uniformity of rhythm in the four Symposium samples [that is speeches by Aristophanes, a writer of comedies, Pausanias, an aristocrat, Eryximachus, a physician, Diotima, a wise prophetess], despite their being parodies, imitation of Lysias’ style by Plato would not account for its deviation from the other four Phaedrus samples; they therefore concluded tentatively that it was a genuine speech of Lysias.’(21)

The widespread prejudice concerning the authorship of Lysias’ speech, which is involved in Brandwood’s accusation, is worth examining, for it has a bearing on our view of the dialogue as a whole and of Plato as its author. Championing Plato’s authorship of Lysias’ speech, Hackforth writes ‘that Plato would thoroughly enjoy exercising his powers of imitation, and would have disliked incorporating extensive material from another’s pen’. In the accompanying note he adds: ‘Plato has admittedly taken great pains (228a-e) to convince us that the speech is authentic; the question is whether he has not taken too great pains’ (Hackforth, p.18). In other words, if Plato insists that something is A, a clever interpreter will know that it is not-A.