1. A BRIEF STATVS QVAESTIONIS

Ever since investigation into Aristotle’s lost dialogues picked up pace in the nineteenth century, scholars have been divided concerning the literary form of Protrepticus. Bernays includes Protrepticus in his work on the dialogues, but hesitates as to its literary form, whereas Bywater assumes that Aristotle’s lost work was a dialogue, something denied by Hirzel.Footnote ¹ Little more than a decade after the publication of Bernays’s book, the positions were exhausted (non-commitment, affirmation, denial), and so, it seemed, were the arguments, since all three scholars based their considerations on external evidence. The odd man out, Hartlich, did point to another route, but his suggestion was largely ignored until rather recently; I will return to him in a moment. The following survey offers only the main arguments.

The external evidence is clearly stated by Bywater and those following him, for instance Diels and Ross. They contend that Cicero modelled his Hortensius on Aristotle’s Protrepticus; Hortensius was a dialogue as is stated by ancient sources, so Protrepticus must also have been a dialogue. Also, they point out that in the ancient lists of Aristotle’s works Protrepticus appears among writings known to have been dialogues. Both arguments were already questioned, and rejected, by Hirzel and later again by Rabinowitz.Footnote ²

In the twentieth century, those following Hirzel changed the approach by arguing on the basis of the text of Aristotle’s Protrepticus itself. Jaeger and Düring, for instance, deny the dialogical form of Protrepticus, because they believe that the syllogistic–demonstrative procedure of Aristotle’s argument in the text is incompatible with dialogue (Jaeger) or that Aristotle supplanted ‘dialogical reasoning’ with the method of demonstration implying, as it were, that he also gave up the dialogue form (Düring). By proposing instead that the Cyprian discourses of Isocrates (Nicocles, To Nicocles, Evagoras) were the literary models of Aristotle’s Protrepticus, Jaeger and Düring articulated the alternative between dialogue or ‘Cyprian discourse’ that forms the backdrop to this article also.Footnote ³

Only a handful of scholars have argued, on the basis of internal linguistic features, that Protrepticus was a dialogue. Hartlich was first to propose that some of the awkward features in Iamblichus’ Protrepticus (a main source for Aristotle’s Protrepticus: Section 2), just as some of the vocabulary, were easily recognizable remnants of dialogue. But curiously his observation was made with respect to a part of Iamblichus’ text (chapter 7) that he did not ascribe to Aristotle’s Protrepticus.Footnote ⁴ His observation concerning the vestiges of dialogue was sound none the less. Allan, who was familiar with Hartlich’s work, noted that there are internal signs of dialogue in Aristotle’s Protrepticus, but did not elaborate.Footnote ⁵ So it was to be about a hundred years since Hartlich before Vendruscolo explored some of these stylistic and grammatical oddities further. He has recently been followed by Hutchinson and Johnson. I am indebted to these scholars and will give references to their discoveries in Section 3.

2. THE TEXTUAL BASIS

Any investigation of this kind must deal with one of the most controversial questions regarding Aristotle’s Protrepticus: which texts to include in an edition of this work? My own delimitation of the material requires a short survey of the answers to this question. Five phases can be distinguished.

In the first phase of rediscovery, Bernays worked on the basis of testimonia. He did consider Stob. Ecl. 3.3.25, but hesitated as to its ascription to Protrepticus.Footnote ⁶ Bywater ascribed parts of chapter 5 and all of chapters 6–12 of Iamblichus’ Protrepticus to Aristotle’s Protrepticus. In Rose’s 1886 edition of fragments ascribed to Aristotle, some of these findings were accepted (Stob. Ecl. 3.3.25 and chapters 6, 8 and 9 of Iamblichus’ Protrepticus), others rejected (chapters 5, 7, 10, 11 and 12 of Iamblichus’ Protrepticus), and new evidence added, most importantly the beginning and the end of Iamblichus’ De communi mathematica scientia (henceforth DCMS) ch. 26, a chapter whose central part is a virtually identical textual overlap with the main part of Iamblichus’ Protrepticus 6, and finally parts of Procl. In Eucl. 1.9.Footnote ⁷ Rose included further texts, but they have not found favour with later editors, so there is no reason to linger on them here.

The second phase was inaugurated by the publication of P.Oxy. 666 in 1904.Footnote ⁸ This text from the second century a.d. links up with Stob. Ecl. 3.3.25 and expands it (henceforth I shall refer to these texts as P.Oxy. 666). Walzer’s edition of Aristotle’s fragments, first published in 1934, included P.Oxy. 666. But that was not the only novelty, since, compared to Rose’s edition, Walzer enlarged the material further by including parts of Iamblichus’ Protrepticus ch. 5, all of chs. 6–12, as Bywater had originally argued, while accepting the parts of DCMS ch. 26 already included by Rose. Alongside this material, Walzer printed a significant amount of other texts which he did not clearly mark as testimonia or fragments.Footnote ⁹ It is indeed a delicate matter to distinguish testimonia from fragments and often impossible to do so in a meaningful way with respect to Aristotle’s dialogues. However, as for Protrepticus, it can and should be done.

A third phase is marked by Düring’s edition of 1961.Footnote ¹⁰ He suppresses the first part of DCMS 26, retaining only 79.15–80.1 as his fr. B 52, and suppresses the last part of DCMS 26 (83.6–22), which had been collected as fr. 8 by Walzer, since Düring believed that it was derived from both On Philosophy and Protrepticus. Further, his edition rearranges the fragments. The rearrangement is problematic for various reasons. But in one respect, at least, Düring’s edition marks an advance. He distinguishes between testimonia and ipsissima uerba in his edition, and he almost respects the distinction (his fragments B 1 and B 6 are, in fact, excerpts from testimonia in his A 1 and A 2 respectively).

An altogether new situation arose in the fourth phase with Gigon’s edition of Aristotle’s fragments in 1987. It seems that Gigon gave up on an edition of Protrepticus and offered instead what he called ‘τόποι προτρεπτικοί from various dialogues’.Footnote ¹¹ Among these he included the last part of Iamblichus’ Protrepticus ch. 5, almost all of chs. 6–12, almost all of ch. 26 of DCMS, most of ch. 27 of DCMS (up to 87.16), several pages from Procl. In Eucl. 1.9, P.Oxy. 666, some proverbs ascribed to Aristotle, and the Diuisiones in Diog. Laert. 3.80–109 as well as the Diuisiones Aristoteleae. To put it dramatically, Aristotle’s Protrepticus was dissolved and seemed to be lost once again in Gigon’s edition.

The fifth and present phase of research began in 2005 with the work of Hutchinson and Johnson on Iamblichus’ Protrepticus chs. 6–12.Footnote ¹² They present strong arguments for ascribing these chapters to Aristotle’s Protrepticus and, until their arguments are persuasively rejected, this conclusion stands. Only two editions, both with useful commentaries, take this result into account: an edition published by Van der Meeren in 2011, whereas the other, as yet unpublished but readily available digitally, is by Hutchinson and Johnson.Footnote ¹³ Today, three main questions remain to be settled: (a) whether to include parts of Iamblichus’ Protrepticus ch. 5 and parts of Iamblichus’ DCMS, (b) whether to include parts of Procl. In Eucl. 1.9, and finally (c) whether to include various pieces of papyrological evidence.

Van der Meeren, whom I follow, accepts P.Oxy. 666 and Iamblichus’ Protrepticus chs. 6–12, whereas Hutchinson informs me (in personal correspondence) that his and Johnson’s current position is that P.Oxy. 3659 and most of Iamblichus DCMS chs. 21–7 should also be included as being reliable citations from the lost work. They reject fragment B of P.Vindob. G 26008Footnote ¹⁴ and treat Procl. In Eucl. 1.9 as part of the reception tradition, not as a reliable citation. I am not sufficiently persuaded that the texts apart from P.Oxy. 666 and Iamblichus’ Protrepticus chs. 6–12 belong to Aristotle’s Protrepticus. But here is not the place for a detailed discussion of this.

My argument concerning indications of dialogue in Aristotle’s Protrepticus is based on six text corpora:

1. (1) Aristotle’s Protrepticus consisting of P.Oxy. 666 edited by Vendruscolo (accepted with some modifications) and Iamblichus’ Protrepticus chs. 6–12 in Pistelli’s edition.Footnote ¹⁵ Aristotle is said by Stobaeus to be the author of P.Oxy. 666, and I assume that he is the author of chapters 6–12 of Iamblichus’ text (except for the obvious bridge passages and the summaries and introductions at the beginning and end of the chapters) and that the text belongs to a single work (Protrepticus).Footnote ¹⁶ The reliability of the evidence from Iamblichus will be addressed when I have presented all my arguments (Section 8).

2. (2) The remains of Aristotle’s dialogues. I base the investigation on direct quotations exclusively: just 1,704 words by my count.Footnote ¹⁷

3. (3) Peripatetic dialogues. For purposes of comparison, I define P.Oxy. 3699 as a Peripatetic dialogue, that is, a dialogue of uncertain authorship that belongs to Aristotle’s school; thus also its most recent editor.Footnote ¹⁸ The comparison will strengthen my claim that certain features are particularly indicative of dialogue.

4. (4) Socratic dialogues. This group consists of Plato’s and Xenophon’s dialogues, whereas I have not taken the fragments of the minor Socratics into account. These texts form the immediate model of dialogue available to Aristotle.

5. (5) Isocrates’ Cyprian discourses (Nicocles, To Nicocles, Evagoras). Since these are contenders as to the literary form of Aristotle’s Protrepticus, they are drawn upon as contrasting material.

6. (6) Aristotle’s genuine esoteric writings, 792,763 words (based on the TLG).

The authenticity of the dialogue On Noble Birth, a main source to Aristotle’s dialogues, was questioned in antiquity.Footnote ¹⁹ The content is certainly Aristotelian (the definition of εὐγένεια as ἀρετὴ γένους appears in Pol. 3.13 and Rh. 2.15) and, seeing no reason why it should not be genuine, I treat it as a dialogue written by Aristotle.

3. INTERNAL INDICATIONS OF DIALOGUE

Before listing the internal indications of dialogue in Protrepticus, let us consider briefly what is meant by ‘dialogue’ here. Following Jaeger, Allan insisted that a distinction between two types of Aristotelian dialogue should be kept in mind, one type similar to those of Plato’s dialogues consisting of shorter, lively and rapid exchange of question and answer, the other type similar to Cicero’s dialogues consisting of longer speeches poised against each other for and against some controversial question and with Aristotle as the main speaker. More recently, Janko has urged a complete reconsideration of what Aristotle’s dialogues were, more or less along the same lines.Footnote ²⁰ My investigation shows that, while it is important to keep the distinction between shorter exchanges and longer speeches in mind, Aristotle’s Protrepticus seems to harbour both types of dialogue. Hence we should distinguish between two patterns of dialogue (shorter exchange and longer speech), which may stand side by side in the same dialogue.

The claim that Protrepticus was a dialogue is based on the following observations:

• asyndeton (considered by Aristotle to be indicative of spoken language)

• use of δήπου (indicative of spoken language)

• use of οὐκοῦνFootnote ²¹ (indicative of dialogue)

• interlocutors (there are two persons represented in the fragments, an ‘I’ on several occasions and a ‘you’ in a couple of places; the frequent use of οἶμαι is particularly indicative of dialogue)Footnote ²²

I will review these phenomena individually in three steps: first, as they are found in Socratic dialogues, the immediate literary model of dialogue available to Aristotle; second, as they are found in Aristotle’s dialogues; third, as they are found in Aristotle’s Protrepticus. Material from Peripatetic dialogues, Isocrates’ Cyprian discourses and Aristotle’s esoteric writings will be drawn upon for corroboration or contrast.

4. ASYNDETON

A collapse in syntax (asyndeton) is not uncommon in philosophical dialogues which reflect, as they do to an extent, spoken language. It is helpful to distinguish, with Denniston, between formal and stylistic asyndeton.Footnote ²³ In philosophical dialogues, formal asyndeton often results from the linguistic frame of question and answer. It is an inconspicuous phenomenon. Stylistic asyndeton is different; when employing it, the author of a dialogue wants to point something out. According to Denniston, stylistic asyndeton is used for emotional effect and to signal that a speaker or writer is too engaged in his argument to care for logical coherence.

There is a considerable amount of this phenomenon in Plato. In most cases, the asyndeton results from the linguistic dynamic of question and answer.Footnote ²⁴ Thus most cases of asyndeton in Plato will not be stylistic, which makes the instances of stylistic asyndeton all the more interesting. The following example comes from Denniston (translation here and in the following passages are mine) (Pl. Prt. 339e6–340a1):

The asyndeton here signals that Socrates has been put slightly off his feet by Prodicus (339e1–3) and that he is stalling to find them again (339e3–5). Plato employs stylistic asyndeton here to convey something about Socrates’ mood. Further examples can be found in Callicles’ longer interventions in Gorgias.Footnote ²⁵ Stylistic asyndeton suits a heated argument, such as might occur in a dialogue, where a speaker easily stumbles over his words and thoughts, because he is agitated or stressed or for whatever reason. It adds livelihood and pace.Footnote ²⁶

Among Aristotle’s dialogues, Eudemus offers the following instance of stylistic asyndeton. I follow Ross in placing commas around it in the Greek (φημί) and dashes in my translation (Arist. Eudemus fr. 45 R³ [Philoponus, In De an. 144.21]):

It is difficult to know precisely what this asyndeton conveys. Perhaps it is no more than a reminder to the reader that there is an interlocutor, who takes a strong position towards the claim that soul is harmony. For my purposes, the asyndeton is important simply because it is there, since, as it happens, stylistic asyndeton occurs in Aristotle’s Protrepticus as well, in a part of P.Oxy. 666 fraught with problems. I print my preferred text, focussing on the emendation of the asyndeton suggested by Diels and subsequently adopted by most editors of Protrepticus (P.Oxy. 666, Vendruscolo with modifications):

Diels suggested inserting ὅπερ or ὅ to emend the asyndeton (τὰ κτήματα· <ὅπερ> πάντων αἴσχιστον: ‘they are themselves less valuable than their possessions, which is the worst of all’).Footnote ²⁹ Two considerations speak against this. First, Aristotle tells us that asyndeton is characteristic of spoken language (Rh. 3.12.1413b17–21). Second, Aristotle was famous in antiquity for his skill as a writer of dialogues, where he is said to be masterly in conveying character by way of his interlocutors’ utterances.Footnote ³⁰ In the face of these observations, the asyndeton in Protrepticus should be retained; it is an intended stylistic feature, indicative of spoken language and used for emotional effect. The text that results contains language which is enthusiastic, even rough, but which is acceptable in a dialogue, as the parallels from Plato and Eudemus demonstrate. Finally, Denniston notes that Isocrates rarely employs stylistic asyndeton.Footnote ³¹ His point is valid for the Cyprian discourses in which I have not been able to discover a single instance.

Hence the asyndeton should be retained, indicating—as it does—spoken language, which makes sense on the assumption that Protrepticus was a dialogue. The asyndeton occurs in a part of Protrepticus that seems to consist in a longer speech. The topic in Protrepticus is how to live, which seems well-suited to cause a speaker to become agitated, and for this reason a small collapse in syntax makes sense here. Also, Aristotle might wish to remind his reader that this is indeed a spoken and lively argument, even though it has for a while had the flavour of a monologue.

5. USE OF ΔΗΠΟΥ

The particle δήπου is of particular interest as yet another feature indicative of dialogue. In Denniston’s words, ‘the certainty of δή is toned down by the doubtfulness of που’ and δήπου therefore translates ‘presumably’, ‘I believe’, ‘I imagine’ and might even be equivalent to ‘of course’.Footnote ³² Denniston’s examples are predominantly taken from Plato or Greek comedy and tragedy. That δήπου in Plato’s dialogues indicates colloquial style is well documented.Footnote ³³ For some occurrences in Xenophon, see Mem. 1.2.41, 3.4.11; Symp. 3.3. One example from Plato should suffice (Euthyd. 276a4–7):

Here δήπου underscores that the example offered is something obvious that merits no further discussion.

Aristotle employs δήπου twice in his dialogues. In the first case, by δήπου the speaker invites, and soon enough receives, assent from his interlocutor. In the second, the speaker uses δήπου to add weight and credibility to his statement:

That δήπου is particularly indicative of dialogue also in Aristotle is suggested by the distribution of this particle across the corpus Aristotelicum. There are five occurences in genuine esoteric writings (δήπου: Metaph. 1.9.992b21, 5.5.1010b12; δή που: Mete. 1.14.352a28; Pol. 3.17.1288a25; Rhet. 1.1.1354a30), whereas it occurs twice in his dialogues, as just observed. A reader, then, should expect to read 852 words in order to encounter δήπου in one of Aristotle’s dialogues, but 158,553 words in his esoteric writings before doing so.Footnote ³⁴ It is not impossible that Aristotle used δήπου in a letter (it occurs in the letters ascribed to Plato at 3.317e, 4.320c, 7.324d and 7.329a), but it is much more common in dialogue.

There are emphatic, confirmational and connective particles evenly distributed throughout Protrepticus (γε, δή, μήν, καίτοι, μέντοι, τοίνυν and combinations). However, these are all fairly common and not particularly indicative of dialogue. As just noted, δήπου is an exception. It is therefore significant that it occurs twice in Protrepticus:

In both passages, δήπου adds emphasis to the statement, whereas only the last instance clearly also invites assent. I take these occurrences as indications that Protrepticus was a dialogue. By contrast, δήπου occurs nowhere in Isocrates’ Cyprian discourses.Footnote ³⁵

6. USE OF ΟΥΚΟΥΝ

οὐκοῦν has an interrogative aspect (οὐκ) and a connective aspect (οὖν), which makes it well suited for use in philosophical dialogue. In fact, as Denniston and others have observed, οὐκοῦν belongs primarily to the Socratic tradition of philosophical dialogue with Plato and Xenophon as its main representatives. It occurs also frequently in Aristotle’s esoteric writings, but, as will be made clear below, not with the same density as in Aristotle’s dialogues.

Plato uses οὐκοῦν in both questions and statements, but οὐκοῦν, no matter its use, always retains an interrogative force.Footnote ³⁶ So even a claim put forward by use of οὐκοῦν will have the implied force of a question. Moreover, Socratic questions with οὐκοῦν are not open or inquisitive. They invite the speaker to either confirm or deny a premise, as intended by the questioner, but do not leave any real choice. Occasionally, a respondent denies a premise that he was intended to affirm, but that only leads to refutation and, in the end, to confirmation of the original premise.

Plato employs οὐκοῦν primarily to make it clear to Socrates’ interlocutors how a statement relates to what has already been accepted or how it will fit into what is to come later in the argument.Footnote ³⁷ Questions with οὐκοῦν are marks of a tightly structured and targeted type of argumentation in dialogue form. They invite assent, or dissent as the case might be, and spell out the argumentative route taken in the discussion so far. Here is one example from Plato (Grg. 475b4–c9):

This text is structured around a cluster of inferential particles, the most important of them being οὐκοῦν (see Xen. Mem. 3.4.8–11 for a similar use). Abstracted from the text, these particles yield the following sequence: οὐκοῦν, ἆρα, οὐκ ἄρα, οὐκοῦν, οὐκοῦν, οὐκοῦν. The same, or a rather similar, cluster of particles recurs in Aristotle’s dialogues. Aristotle’s use of οὐκοῦν is particularly conspicuous from On Noble Birth, but there is an instance in On Philosophy also (col. ii 7 Auffret), and a potentially problematic one in On Poets (fr. 72 R³) salvaged from Athenaeus. Some editors of Athenaeus (Gulick following Jahn) correct οὐκοῦν into οὔκουν. The most recent editor of Athenaeus, however, and all editions of Aristotle’s dialogues (treated above, Section 2), which include On Poets, read οὐκοῦν.Footnote ³⁸ I accept this, but do not base anything on it. For corroborative purposes, οὐκοῦν occurs in a fragment of Peripatetic dialogue in P.Oxy. 3699 (col. iii 8–12 Verhasselt).

In a continuous stretch of dialogue from On Noble Birth, οὐκοῦν (written οὐκ οὖν and once in reverse word order) appears three times. First, the respondent has just answered a question (On Noble Birth fr. 92.13–21 R³ [Stob. Ecl. 4.29.15–23]):Footnote ³⁹

Here οὐκοῦν introduces a new premise into the argument (to define ‘noble born’ through ‘once rich’ is just as wrong as to define it through ‘once good’) and invites assent. As Denniston notes, even such a statement as this has interrogative force. The dialogue continues in this way (On Noble Birth fr. 92.21–3 R³ [Stob. Ecl. 4.29.23–5]):

The reversed word order here is of no importance; Aristotle employs οὖν οὐκ to introduce a conclusive question inviting an affirmative answer. The dialogue continues (On Noble Birth fr. 92.23–8 R³ [Stob. 4.29.5]):

In this last stretch, οὐκοῦν introduces an additional premise and invites assent to it. A similar use of οὐκοῦν seems to be found in On Poets (fr. 72 R³). The argument in On Noble Birth concludes thus (On Noble Birth fr. 92.28–32 R³ [Stob. 4.29.30–5]):

The sequence of inferential markers here at first seems muddled (καί, γάρ, οὐκ οὖν, γάρ, ἆρα, οὖν οὐκ, ἐπεί, γάρ, οὐκ οὖν, ἀλλὰ μήν, γε φαμέν, ὥστε, ἄρα), but, when stripped to the bone, the structure of the argument is: οὐκοῦν, οὖν οὐκ, οὐκοῦν, ἄρα.

A similar picture seems to emerge from On Philosophy. The text is somewhat difficult to interpret, because it is incomplete; none the less, the tendency is clear (On Philosophy, col. II.7–17 Auffret):

οὐκοῦν here appears in surroundings familiar from On Noble Birth. There the main interlocutor obtains a premise from the respondent by using φαμέν (fr. 92.30 R³ σπουδαῖον δέ γε φαμὲν ἕκαστον κατὰ τὴν αὐτοῦ ἀρετὴν εἶναι) paralleled in On Philosophy by the main interlocutor’s γὰρ … φαμέν and the respondent’s φαμὲν γάρ. We seem to enter the text somewhere within a longer argument. οὐκοῦν is paired with [οὖ]ν before the respondent has a chance to reply. Perhaps the argument was offered as a whole packet for the respondent to accept, perhaps a clearly marked conclusion (ἄρα) was introduced once the respondent conceded the οὐκοῦν, [οὖ]ν clause. At least, the tendency to have οὐκοῦν paired with a conclusive particle of some sort is clear.

These occurrences reveal that οὐκοῦν for Aristotle is a convenient way to structure the argument while staying true to the fact that he is writing a dialogue. He employs οὐκοῦν in non-dialogical texts also but, as will be clear below, not quite in the same way as in his dialogues. οὐκοῦν makes the inferential structure of the argument clear by inviting the interlocutor to assent as the argument moves ahead towards its clearly marked conclusion (ἄρα) or towards securing a premise needed for this conclusion, if that is how οὖν works in On Philosophy. The argument is structured by a sequence of inferential markers: οὐκοῦν, ἄρα, οὖν with a number of other particles, conjunctions, verbs and so on to support it (ἀλλὰ μήν, ἆρα, γάρ, φαμέν, ὥστε, etc.).

With this is mind, let us turn to Protrepticus. There is no need to look in detail at all nineteen occurrences of οὐκοῦν here.Footnote ⁴⁰ They are remnants of the original dialogical form of the work, retained by Iamblichus because they help structure the argument. Most conform to the sequence of inferential markers in Aristotle’s dialogues in the sense that one or more instances of οὐκοῦν are paired with an explicit conclusive adverb, conjunction or particle (ἄρα, οὖν, ὥστε). Only one instance is a self-contained inference (9.50.15–16 οὐκοῦν τό γε καλῶς, ὀρθῶς). The following variations are found:

• οὐκοῦν … ἄρα (7.43.1–7, 7.43.14–20, 9.52.4–8, 11.59.3–13)

• οὐκοῦν, οὐκοῦν … ἄρα (9.51.20–52.4, 11.57.23–58.9)

• οὐκοῦν … οὖν (7.41.29–42.3, 8.45.11–17, 9.53.10–18)

• οὐκοῦν, οὐκοῦν … οὖν (11.58.21–59.3)

• οὐκοῦν … ὥστε (6.41.2–5, 7.44.17–26, 9.50.24–51.6)

• οὐκοῦν, οὐκοῦν … ὥστε (12.59.26–60.8)

These all conform neatly to the cluster of inferential particles, adverbs and so on, found in Aristotle’s dialogues and in Plato’s and Xenophon’s dialogues, except that Aristotle in his dialogues and in Protrepticus more directly links οὐκοῦν with a clearly marked conclusive particle. Aristotle’s use of οὐκοῦν suggests a more detailed attention to the logical structure of the arguments.

I offer, first, a fairly easy example from Protrepticus, then a difficult example. I have supplied the text with question marks and inserted the answers of the respondent in square brackets. The answers are taken from On Noble Birth. I place them in the text to make clear how easy it would have been for Iamblichus to delete the answers while retaining unaltered the questions posed by the main interlocutor. Of course, we do not know whether Protrepticus was narrated, as I pretend it to have been here. The fairly easy example is this (Protrepticus 6.41.2–5):

There is no need to linger on this text. A commitment to certain claims is signalled by οὐκοῦν and a consequence of this commitment is teased out by ὥστε. Here οὐκοῦν introduces a new question, which is how it is used on several occasions in Plato (Euthyd. 276a4; Grg. 475b5, c4, c6, c7, treated above) and in Aristotle’s dialogues On Noble Birth (fr. 92.26–8 R³) and On Philosophy (also discussed above).

The fairly difficult example is found at Protrepticus 9.51.16–52.8. I add Latin letters in parentheses for ease of reference (Protrepticus 9.51.16–52.8):

Syntactically and logically speaking, the text from (a) to (j) is a bumpy ride and Vendruscolo may be right to suspect that Iamblichus has interfered with it in some way.Footnote ⁴¹ His main point is that it is unclear what the protasis to the apodosis introduced by εἰ τοίνυν (9.51.16) is. The most natural reading would be to take (c), (d) and (e) introduced by οὐκοῦν (9.51.20–1) to be the protasis to (a) and (b), in which case οὐκοῦν has conclusive force (‘so, then’). However, (c), (d) and (e) do not follow from (a) and (b); therefore, unless the argument is completely derailed, οὐκοῦν cannot have conclusive force here.

Perhaps Iamblichus did obfuscate the argument in some way. Perhaps something has gone wrong in the process of transmission. Be that as it may, οὐκοῦν at 9.51.20–1 and 9.51.24 only make sense if we take them as confirmatory. οὐκοῦν in these cases does no more than introduce further premises, (c), (d), (e) in the first case, (f), (g), (h) in the second.

If so, we have to understand, with Düring and Vendruscolo, all the premises in (a)–(i) as forming apodoseis to the protasis in (j) φρόνησις ἄρα τις … κτλ. (9.52.2–4). This is the least unsatisfactory solution. In this case the argument that one has to salvage from the text is this:

• Premise 1: τοῦτο (= φρόνησις) ὕστατον τοῖς ἀνθρώποις (h)

• Premise 2: κατὰ φύσιν ὕστατον τέλος (b)

• Conclusion: φρόνησίς τις κατὰ φύσιν ἡμῖν ἐστι τὸ τέλος (j)

I have reversed the word order in (b) to ensure that the logic adds up. Strictly speaking, the premises do not yield φρόνησίς τις (‘some sort of thinking’), but this is no serious problem. ἡμῖν (‘for us’) in (j) is warranted by τοῖς ἀνθρώποις in (h). The real problem is getting the premises (a)–(i) sorted out so as to yield this or a similar argument. I can do no better than Vendruscolo here and refer to his reconstruction of the logic.

Düring sorted out the syntax, yet he had reservations about three occurrences of οὐκοῦν in such close vicinity (9.51.20–1, 9.51.24, 9.52.4) and noted that there was no parallel in Aristotle’s esoteric writings for οὐκοῦν thrice in such close vicinity.Footnote ⁴² Had he looked at Aristotle’s dialogues, however, he would have found such a parallel in On Noble Birth (fr. 92 R³), and had he looked at Plato, he would have found it in Grg. 475b3–c9 also (both treated above). On the assumption that Protrepticus was a dialogue, Düring’s worry dissolves. Düring misses this, because he interprets Protrepticus as based on the literary model of Isocrates’ Cyprian discourses.

To sum up on this quite difficult example of οὐκοῦν in Protrepticus, the syntactical and logical obscurity remains, but is not aggravated by assuming dialogue.

In (f), (g), (h), οὐκοῦν is followed by γάρ, but on the analogy of On Noble Birth (fr. 92.13–21 R³), such a construction is permissible in an Aristotelian dialogue.

Furthermore, it might seem bold to place διό at the beginning of a question or remark (i). I do so to avoid two causal clauses (γάρ, διό) in one question. There is no precedent for διό at the beginning of a question in Aristotle’s dialogues; but it does appear in this position on a single occassion in Plato (Plt. 294e5–7), so this construction is not impossible even if it is unusual.

There are many interrogative adverbs and pronouns such as πῶς, τί or τίς spread out across Protrepticus. They could easily be construed as means for introducing questions in a dialogue, but since they are common enough in non-dialogical texts also, there is no reason to look at them for my present purposes. As should be clear, on the other hand, οὐκοῦν is frequent in both Protrepticus and Aristotle’s dialogues. The frequency of οὐκοῦν in Protrepticus is best explained on the assumption that it was originally a dialogue. In Isocrates’ Cyprian discourses, by contrast, there is not a single use of οὐκοῦν.

7. THE INTERLOCUTORS

Interlocutors are a prerequisite for dialogue, at least in the sense of the term which is relevant here. An ‘I’ and a ‘you’ might occur in a letter, as in Isocrates’ Cyprian discourses; even so, however, the presence of interlocutors in Protrepticus does not weaken the claim that this text was a dialogue, even if it does not prove it either.

It needs no demonstration that there are two or more interlocutors in Plato’s and in Xenophon’s dialogues. The narrator, if there is one, and the main interlocutor, usually the questioner, often appear in the first person singular (‘I’), whereas the respondent usually comes forth either as ‘you’ or as ‘he’ (‘he said’, and so on). Much the same can be said about Aristotle’s dialogues. I focus first on the main interlocutor (the ‘I’) and on Aristotle’s use of οἶμαι, before turning to the respondent. Both interlocutors make an appearance in the following passage ( On Noble Birth fr. 91 R³ [Stob. Ecl. 4.29.24]):

On Noble Birth is a narrative dialogue in which the narrator takes part himself. This means that the respondent also occasionally speaks in the first-person singular as above. I will return to him in a moment.

In Protrepticus, there are two interlocutors, an ‘I’ and a ‘you’, both of whom make an appearance in the following text (Protrepticus 9.49.14–17):

There is nothing particularly dialogical about the use of λέγω in this passage. It appears often in Aristotle’s extant esoteric writings (431 matches in the TLG). But the parallelism between Protrepticus and On Noble Birth fr. 91 R³ is striking; in both texts a statement is addressed to a ‘you’ and elaborated by adding λέγω δ᾿ ὅσων or λέγω δὲ τοῦτο.

If λέγω is not particularly dialogical, the use of οἶμαι (‘I believe’, ‘I think’, and so on) is. This verb conveys personal commitment, caution, reservation, emphasis or similar sentiments. In Plato, there are 165 occurences in Republic, 43 in Laws and, significantly, only one in Timaeus, which for the most part consists of monologue. There is no point in citing an example from Plato or Xenophon.

Aristotle employs οἶμαι once in his dialogues in a fragment where he takes Euripides to school for misrepresenting how Aetolians usually wear their shoes in war (On Poets fr. 74 R³ [Macrob. Sat. 5.18.19]):

This fragment demonstrates that interlocutors in Aristotle’s dialogues would express caution or reservation by using οἶμαι in this case when explaining a strange phenomenon. Aristotle uses οἶμαι only three times in his genuine esoteric writings (An. post. 1.1.71a6; Metaph. 1.8.990a18, 7.16.1041a1);Footnote ⁴³ this makes its occurrence in one of his dialogues all the more noticeable.

Now, there are four instances of οἶμαι in Protrepticus. It is not inconceivable that Aristotle used οἶμαι in an exoteric work, which was not a dialogue. But there is no evidence that he did, whereas he does employ οἶμαι in his dialogues. These occurrences differ from οἶμαι in On Poets (fr. 74 R³), which appears in connection with a tentative explanation of a curious fact, in that they are all paired with some controversial claim: that thinking is the most choiceworthy of all goods (7.43.20–5); that phronêsis is necessary for eudaimonia (8.46.22–6), whereas exterior goods are irrelevant (8.47.18–21); or, as in the following text, that thinking (the virtue Aristotle is talking about) is the key element in human identity (Protrepticus 6.41.29–42.4):

In Protrepticus, then, οἶμαι signals commitment to a claim that most people would probably dismiss. It conveys strong personal conviction of a kind we rarely encounter in Aristotle’s esoteric writings.

None the less, Isocrates also employs οἶμαι in his Cyprian discourses, six times. In three, οἶμαι is used in connection with a claim that all or most people will admit, believe or agree to something that Isocrates has said (To Nicocles 9; Nicocles 14, 29). So these claims are intended to be uncontroversial. In two cases, Isocrates signals by οἶμαι that he is aware that he could have said more than he does (Evagoras 33, 70). Here οἶμαι is used self-consciously to identify a speaker in control of his subject matter, but these are also not controversial claims. Only one instance is similar to Aristotle’s use in Protrepticus. Isocrates claims that Evagoras is worthy of becoming immortal and cautiously adds ‘οἶμαι’ (Evagoras 73). That is controversial. But this aside, the tendency is clear. Isocrates appeals to consensus, to what everyone believes, whereas Aristotle’s use of οἶμαι reveals a confrontational attitude, a contempt for what everyone believes well known from the Socratic tradition where an opinion is never validated just because many people hold it (for example Pl. Cri. 46c8–47a5). A confrontational attitude sits easier in a dialogue than in a letter of admonition.

There are two further rather unusual instances of the ‘I’ in Protrepticus. At 6.40.19, Aristotle writes μοι δοκεῖ (‘it seems to me’). This is not found in his dialogues and is attested only in one instance outside of Protrepticus (Mete. 1.3.339b23). The reverse word order, δοκεῖ μοι, occurs nowhere. This way of writing is, in other words, extremely rare in the corpus.

The same is true for the second instance. At 6.40.13, the text reads νομίζω (‘I believe’), a hapax legomenon in Aristotle. It occurs in what appears to be a bridge passage between blocks of texts. Even if νομίζω is compatible with dialogue, as instances from Plato and Xenophon demonstrate (Pl. Euthphr. 3a7; Ap. 26c2; Xen. Mem. 1.2.45, 1.6.11), I base nothing on it here, since I suspect Iamblichus’ hand behind it, not Aristotle’s.

The respondent in an Aristotelian dialogue seems not to be a complete philosophical novice. In the sample from On Noble Birth fr. 91 R³, cited above, he admits puzzlement (ἐγὼ ἀπορῶ) concerning the theme at hand. This suggests that the respondent is philosophically advanced enough to be aware that there are things he does not understand. A dialogue by Aristotle, then, is not designed to force this awareness unto the interlocutor, it would seem. Perhaps we should imagine the respondent as similar to Theaetetus in this respect (Pl. Tht. 148e): cooperative, benevolent, non-eristic. Be that as it may, in the extant fragments of dialogue, the respondent contributes in a modest way (On Noble Birth fr. 92 R³ [Stob. Ecl. 4.29.25]):

This response could be removed from the text without any loss of meaning. In Protrepticus, the respondent (the ‘you’) appears only on two occasions. The first was cited above (9.49.14–17), the second is as follows (Protrepticus 9.50.20–3):

The respondent seems to be directly addressed, but we do not know his answer. My suggestion, as noted already, is that Iamblichus could easily and without loss of meaning remove interventions by the respondent, if they were similar to what we find in Aristotle’s dialogues (see the instances from On Noble Birth and On Philosophy cited above). Iamblichus’ modus operandi with respect to removing dialogical elements from the ‘Plato chapters’ of his Protrepticus has been investigated by Hutchinson and Johnson, and he could easily have done the same to the ‘Aristotle chapters’.Footnote ⁴⁴

In Protrepticus and in the dialogues the ‘you’ is never addressed in the second-person singular imperative (βιάζεσθε in Eudemus fr. 44 R³ τί με βιάζεσθε λέγειν should be read as second-person singular indicative in the middle voice). Isocrates, on the other hand, employs the second-person singular imperative often in the Cyprian discourses, particularly when he lists ὑποθῆκαι for Nicocles. Aristotle does not imitate this grammatical feature in Protrepticus, where there is hardly any trace of ὑποθῆκαι, the argument about honouring one’s parents coming closest to such traditional admonitions (8.46.11–15).

Finally, the fragments of Protrepticus allow no reliable conclusions about how many interlocutors appeared in the text, or who they were.Footnote ⁴⁵

8. IAMBLICHUS’ RELIABILITY

It cannot be excluded that the occurrences of δήπου, οὐκοῦν and οἶμαι in Aristotle’s Protrepticus were inserted into the text by Iamblichus. In his ‘Plato chapters’, Iamblichus twice inserts δήπου (13.66.5, 15.79.21) and once οἶμαι (14.76.3) into Plato’s text (he retains οἶμαι faithfully five times: 13.69.17, 15.80.4, 15.80.11, 15.81.5, 18.87.16). Likewise, he twice inserts οὐκοῦν (13.62.17, 13.63.8), suppresses it six times (mainly in ch. 13, once in ch. 17; see Pl. Phd. 64e4, 66b1, 67b7, 67d4, 68c8; Grg. 494b3) and retains it faithfully seven times (13.62.17, 13.66.6, 13.69.11, 15.79.21, 18.87.3, 18.87.17–18, 18.87.25). In all cases, he does this to mask dialogue.

However, it is unlikely that all occurrences of these words in the ‘Aristotle chapters’ were thus inserted. Not all occurrences in the ‘Plato chapters’ were inserted by Iamblichus, after all. But if he did insert them to mask dialogue in Protrepticus, as is the case with the ‘Plato chapters’, that would be a welcome result for the present analysis. The real worry seems to be that Iamblichus could have inserted these words into some non-dialogical text by Aristotle (for whatever reason).