Aristotle’s Poetics 1461a30 contains a Homeric quotation neglected by modern editors.Footnote ¹ Poetics 1461a29–30 reads εἴρηται ὁ Γανυμήδης Διὶ οἰνοχοεύειν, ‘Ganymede has been said to pour wine for Zeus’. This is a partial quotation of Iliad 20.234, τὸν καὶ ἀνηρέψαντο θεοὶ Διὶ οἰνοχοεύειν, ‘that one [Ganymede] indeed did the gods carry off to pour wine for Zeus’.Footnote ² The phrase Διὶ οἰνοχοεύειν is distinct to and unique in Homer.Footnote ³ Nevertheless, modern critical editions of the Poetics are unified in their decision not to put quotation marks around Διὶ οἰνοχοεύειν or to demarcate the phrase as a quotation in any other way in their presentation of the Greek text.Footnote ⁴ While the Iliadic passage is sometimes cited in the accompanying commentary or index locorum, each critical edition presents the quotation differently from other literary quotations within the Aristotelian text itself, thereby obscuring its quotational status. In defending the status of Διὶ οἰνοχοεύειν as a Homeric quotation, I aim to demonstrate how triangulating the Greek manuscript evidence, the Arabic translation and Aristotelian stylistic data can help to establish the text of the Poetics. Modern critical editions of the Poetics recognize the importance of the Arabic and Latin translations for the constitution of the text.Footnote ⁵ In the case of Poetics 1461a30, the Arabic translation both illuminates the need to print a text that acknowledges the Homeric quotation and offers important insight on Aristotelian stylistics in the Poetics more generally.

Instead of the infinitive οἰνοχοεύειν, the Arabic evidence supports the variant οἰνοχοεύει, as it is also reconstructed in MSS Π and Σ.Footnote ⁶ Gutas points out that the Arabic translation has a verb in the perfect indicative, which normally indicates an original Greek indicative.Footnote ⁷ Greek infinitives are usually rendered by the Arabic translator with the particle an and a subjunctive verb.Footnote ⁸ Reconstructing the original Greek with the aid of the Arabic translation results in εἴρηται followed by ὁ Γανυμήδης Διὶ οἰνοχοεύει. The absence of a subordinating ὅτι or ὡς in the Greek manuscripts would require the reconstructed finite verb to be taken as a direct quotation and not as indirect discourse. The reconstructed quotation, ὁ Γανυμήδης Διὶ οἰνοχοεύει, as opposed to Διὶ οἰνοχοεύειν, is unmetrical and belongs to no extant source text, whether poetry or prose.

This unsatisfactory solution seems to have encouraged editors to disregard the indicative variant supported by the Arabic evidence. But the lack of quotation marks around Διὶ οἰνοχοεύειν in modern critical editions parallels the stumbling block that the Arabic translation reveals: when Aristotle uses εἴρηται to introduce direct quotations, he usually does so without syntactic integration of the quotation into his prose.Footnote ⁹ In other words, there is an assumption being made based on Aristotelian stylistics that, whenever εἴρηται introduces a quotation, whatever follows should be treated as a quotation syntactically independent from Aristotle’s prose. But ὁ Γανυμήδης Διὶ οἰνοχοεύειν is unmetrical and therefore cannot be a poetic quotation. The Arabic translation addresses the problem by emending the infinitive οἰνοχοεύειν to the indicative οἰνοχοεύει to create a grammatically coherent prose quotation. This unpoetic solution, however, is incoherent within the context of Aristotle’s defence of poetry in this section of the Poetics.

The problem of the Arabic solution throws into relief the interpretative choices that also inform modern editions. In manufacturing a prose quotation that is not, in fact, a real quotation, the Arabic translation appears to interpret ὁ Γανυμήδης Διὶ οἰνοχοεύει as a general quotation belonging to no particular source text. Just so, by neglecting to put quotation marks around Διὶ οἰνοχοεύειν, modern editors have operated on the assumption that εἴρηται introduces only direct quotations syntactically independent from Aristotle’s prose and have consequently treated the verb οἰνοχοεύειν as jargon used generally of Ganymede’s situation, not as a quotation, whether general or specific.Footnote ¹⁰ While both the Arabic translation and modern critical editions can be seen as unified in their treatment of οἰνοχοεύει(ν) as a generally spoken turn of phrase, there is a crucial difference. Despite the fact that the modern critical editions preserve the actual Homeric quotation Διὶ οἰνοχοεύειν in a way that the Arabic does not, their lack of quotation marks ironically strips the quotation of its quotational status where the Arabic’s misquotation of Homer demarcates οἰνοχοεύει as part of a quotation, real or not.

While modern critical editions are right to preserve the actual Homeric phrase Διὶ οἰνοχοεύειν, the Arabic translation’s treatment of οἰνοχοεύει(ν) as part of a quotation suggests the need to examine Poetics 1461a27–30 more closely to elucidate the apparent deviation from Aristotle’s usage of εἴρηται in connection with direct quotations:

This passage occurs in ch. 25 (1460b6–1461b25), a section devoted to defending poetry,Footnote ¹² particularly Homer, from criticism. This subsection addresses criticism against phrases that are factually inaccurate yet customary in speech. Here, Aristotle presents two pairs of evidence. The first constituent of each pair is a phrase in the vernacular. The parallel structure is evident in the lacuna that requires the reader to take φασιν εἶναι from the first constituent of the first pair and resupply it in the first constituent of the second pair, that is, χαλκέας τοὺς τὸν σίδηρον ἐργαζομένους φασὶν εἶναι. Both pairs comprise a phrase in the vernacular, φασίν X εἶναι Y, and a Homeric quotation introduced by the particle ὅθεν, expressing how the poetic phrase follows the same logic underlying the phrase in the vernacular. For subsection 1461a27–31 to cohere with rest of ch. 25, it must address poetic evidence, and ὅθεν indicates that this is indeed the case. On the one hand, κνημὶς νεοτεύκτου κασσιτέροιο (Il. 21.592) has been unequivocally accepted as a Homeric quotation most likely because of the red flag waved by the Archaic genitive κασσιτέροιο. On the other hand, the confusion posed by εἴρηται in the Aristotelian context has combined with the lack of distinct Homeric linguistic markers within the quotation and with the interpolation of the nominative ὁ Γανυμήδης in place of the original τόν to obscure the rightful status of Διὶ οἰνοχοεύειν as a Homeric quotation.Footnote ¹³

Nevertheless, the context of Aristotle’s argument and the structural parallelism between the constituents of each pair of evidence clearly indicate the need to mark the phrase Διὶ οἰνοχοεύειν explicitly as a poetic quotation. The status of the Homeric quotation is supported not only by the parallel ὅθεν construction but also by the fact that both quotations contain features that are unique in the extant Homeric corpus. To disregard the Homeric status of these quotations is to neglect how Aristotle, by incorporating these quotations with minimal emendation, highlights the features that make them unique in Homer and, moreover, relies on those very features as building blocks of his own argument on the levels of both form and content.

With regard to form, the syntax of each quotation allows for its incorporation with minimal emendation into the parallelism that Aristotle establishes with the perfect passive indicative verbs πεποίηται and εἴρηται, respectively. Iliad 21.592 is the sole instance of the noun κνημίς in the nominative and, moreover, the sole instance of the nominative when accompanied by the genitive of material κασσιτέροιο.Footnote ¹⁴ The nominative κνημίς is incorporated easily into Aristotle as the subject of πεποίηται. It is not given the expected article by Aristotle when incorporated into his prose, however.Footnote ¹⁵ Since the lack of the article does not preclude a reader from understanding Aristotle’s argument, the interpolation is evidently deemed unnecessary and is rejected in favour of preserving the original syntax of the Homeric quotation. In this same vein, οἰνοχοεύειν is readily incorporated into Aristotle’s prose as an infinitive dependent on εἴρηται,Footnote ¹⁶ and subsequent emendations made to the quotation are minimal. The subject of οἰνοχοεύειν must be clarified by the nominative ὁ Γανυμήδης for Aristotle’s εἴρηται construction to be felicitous, but where further interpolation could have been made, Aristotle does not emend the quotation, as we can see in Aristotle’s treatment of οἰνοχοεύειν. In the Homeric corpus, the root οἰνοχοε- appears in two verbal forms, οἰνοχοεύω and οἰνοχοέω. The former can only be used intransitively, and the latter can be used transitively or intransitively in its indicative and participial forms but only intransitively when it appears in the infinitive form, which is extant only in the aorist tense.Footnote ¹⁷ Notably, elsewhere in the context of the divine symposium, the verb is only used transitively.Footnote ¹⁸ Paraphrasing Il. 20.234 with οἰνοχοέω in the indicative would allow Aristotle to use the form more frequently connected to divine contexts and to clarify his argument more concisely, for example νέκταρ οἰνοχόει, denoting that nectar is the actual substance being poured, hence the accusation of factual inaccuracy.Footnote ¹⁹ With the absence of a direct object, we see how Aristotle instead includes without emendation the direct quotation, Διὶ οἰνοχοεύειν, and appends to it a genitive absolute to provide the clarification that the gods do not drink wine.

This analysis of the syntactic structure of Aristotle’s argument has an important implication: εἴρηται in Poetics 1461a29 should be treated in comparison not to other instances of εἴρηται in Aristotle’s treatises but instead to its parallel within this particular passage, namely to πεποίηται. When the two perfect passive indicative verbs are taken syntactically in parallel, we can see more clearly how exactly the partial quotations that depend on these verbs function within the parallelism of Aristotle’s argument on the level of content. The quotation introduced by πεποίηται is uttered by the omniscient third-person Homeric narrator: ἀμφὶ δέ μιν κνημὶς νεοτεύκτου κασσιτέροιο | σμερδαλέον κονάβησε, ‘and the greave of newly wrought tin rang terribly around him’ (Il. 21.592–3). With his incorporation of the partial quotation, Aristotle redirects our attention from the action of the original scene, how the greave is clanging, to the action of the poet; replacing κονάβησε with πεποίηται recalls the topic sentence of this subsectionFootnote ²⁰ and foregrounds the genitive of material, which is now explicitly dependent on πεποίηται. This manipulation of the quotation shifts the focus to how the poet has chosen to construct the greave, that is, from the material of tin as opposed to bronze. This alleged material inaccuracy in Homer’s poetic construction is Aristotle’s primary concern. πεποίηται has the dual effect of capturing what physical material the unreal object ‘has been made of’ within the poetic reality of the Iliad and how that poetic reality ‘has been made’ by the poet with particular words.

Just so, Aristotle diverts our attention in the second quotation from the action of the original scene to the action of the poet. Iliad 20.234 relays how the gods have taken Ganymede away (ἀνηρέψαντο), but Aristotle’s quotation instead emphasizes how Ganymede’s story has been told by the poet (εἴρηται), thereby bringing Homer’s diction, particularly the intransitive οἰνοχοεύειν, to the fore. The absence of a direct object to clarify the factual reality of the scene makes this quotation a uniquely ideal exemplum for Aristotle’s argument, which hinges on the alleged inaccuracy of the term when applied to the nectar-drinking gods. Just like its parallel πεποίηται, the verb εἴρηται has a dual effect: it captures not only the words that have been spoken by the poet but also the poetic reality of the Iliad in the immediate context of the quotation. At 20.234, Διὶ οἰνοχοεύειν occurs within the passage in which Aeneas is telling Achilles his lineage. Thus Διὶ οἰνοχοεύειν not only ‘has been spoken’ by Homer but also ‘has been spoken’ by a character within his poem.

Overall, restoring the quotation marks around Διὶ οἰνοχοεύειν is crucial for understanding Aristotle’s argument, which is fundamentally an assessment of inaccuracies found in poetic evidence. Neither πεποίηται nor εἴρηται should be taken as a formulaic Aristotelian marker of direct quotation, but rather should be understood as working intimately with its respective partial quotation to incorporate with minimal emendation the full Homeric context into Aristotle’s discussion of poetic realities. Both verbs fulfil the dual function of referring to the reality simultaneously of the poem and of the poet. Aristotle’s defence of poetry and the nature of its realism is as much about the poetic construct as it is about poetic construction.

In its disagreement with the critical editions of the Greek, the Arabic translation of Poetics 1461a29–30 offers three important insights: first, οἰνοχοεύειν occurs within a direct quotation; second, this passage is an outlier in view of Aristotle’s usage of the verb εἴρηται in connection with direct quotations elsewhere in his treatises; and third, despite the complicated history surrounding its publication, transmission and emendation,Footnote ²¹ readers of Aristotle’s Poetics have historically approached the text with the assumption that it exhibits stylistic coherence. It is this last point that has significant implications for our understanding of the Poetics more broadly. A statistical analysis of εἴρηται in the Poetics and Rhetoric corroborates an approach that assumes on the macroscopic level at least a degree of stylistic coherence in Aristotle’s works,Footnote ²² while Poetics 1461a29–30 demonstrates the need to pay due attention to textual coherence on the microscopic level. Close examination of the syntactic parallelism resolves the apparent contradiction between the typical Aristotelian usage of εἴρηται and the incorporation of the Homeric quotation in this particular instance. Equally important to the result of this study is the analytical method behind it, namely the stylistic analysis of Aristotle informed by the oblique insight that the Arabic translation affords in its reception of the Greek. This method, which seeks to elucidate the conversation between witness and source, particularly at points of disagreement between the witnesses, has applications that extend far beyond Poetics 1461a29–30.

A cursory glance at the same section of the Poetics reveals how this method can illuminate the mangled Homeric quotations of Iliad 10.1–2, 11 and 13 in 1461a16–19, which cannot be fully resolved to restore an accurate quotation of the Homeric text as we have it.Footnote ²³ All manuscripts and translations of the Poetics, including the Arabic, agree with the Homeric evidence in reading ἄλλοι, the ‘other’, rather than πάντες, ‘all’, ‘gods and men were sleeping all night’. This rendering of the Homeric verse causes two points of confusion. First, ἄλλοι cannot be applied appropriately to both men and gods given the original Homeric context in which Agamemnon, a man and not a god, is being distinguished from the ‘others’. Second, Aristotle goes on to say in his assessment of the quotation in 1461a19–20 τὸ γὰρ πάντες ἀντὶ τοῦ πολλοί κατὰ μεταφορὰν εἴρηται, ‘for all has been said metaphorically instead of many’.Footnote ²⁴ Because of these problems, modern critical editions replace ἄλλοι with πάντες in the quotation despite the overwhelming contrary evidence of the witnesses. This is where a method that strives to reconstruct the conversation between witness and source can be useful. The Arabic translation appears to address the first problem by transposing the two nouns, gods and men, rendering the verse as ‘the other men and the gods’, with the Arabic adjective ‘other’ immediately following and therefore modifying the noun ‘men’.Footnote ²⁵ The fact that the Arabic solution to this first problem is to transpose the two nouns rather than to emend ἄλλοι to πάντες demonstrates its insistence on ἄλλοι as the correct rendering of the Homeric quotation. Indeed, the Arabic translation’s neglect of the second problem, namely the absence of a quotational referent for τὸ πάντες, is reason for pause. Perhaps ἄλλοι does not pose the problem that modern editors think it does.

Turning to the Aristotelian stylistic data, we know that εἴρηται, when connected to direct quotations, usually introduces a syntactically independent quotation. This is not the case here, as πάντες is subordinated to an article to form a substantive that is the subject of εἴρηται. This kind of incorporation, however, is not the same kind of syntactic integration of direct quotation that we saw in 1461a29–30, which would require subject–verb agreement between the plural noun πάντες and a corresponding plural verb. Surveying the Aristotelian evidence demonstrates how this subordination of the plural masculine noun to a singular neuter article need not indicate direct quotation but rather notional quotation,Footnote ²⁶ statistically the most common function of εἴρηται in the Poetics.Footnote ²⁷ By ‘notional quotation’ I mean speech content that is framed as having been previously uttered but is not explicitly reiterated through direct quotation. The notional quotation, which is to say the previously addressed topic or concept denoted by εἴρηται, may appear as the nominative subject of the verb.Footnote ²⁸ If we consider the Aristotelian evidence in light of the witnesses’ insistence on τὸ πάντες despite its dissonance with the quoted ἄλλοι, the logical conclusion is that τὸ πάντες may be treated not as a direct quotation but rather as a notional quotation. In this passage, Aristotle discusses the implications of the poetic phrasing. Stating that those ‘other’ than Agamemnon are asleep is equivalent to saying that ‘all’ except for Agamemnon are asleep. ‘Others’ suggests the concept of ‘all’ in place of the literal ‘many’, leaving the problem that no one is left to play the pipes and syrinxes that Agamemnon hears.Footnote ²⁹ Thus it is the inaccuracy of the implied ‘all’ that Aristotle is addressing here, not necessarily the actual word πάντες. Neither its function as the subject of εἴρηται nor its subordination to the article τό necessitates that πάντες in Poetics 1461a16–19 be treated as a direct quotation. Instead, τὸ πάντες can be considered within the paradigm of notional quotation, as a way to render the concept of ‘all’ that only appears notionally in Aristotle’s reconstruction of Homer. Therefore the Aristotelian text does not require that the Homeric quotation itself be emended to read πάντες in correction of ἄλλοι, the word found in Iliad 10.1 and attested in Aristotle by all of the witnesses.

Producing a critical edition of such a poorly transmitted text is a difficult task. My hope is that the analytical methodology behind these emendations will further the important work that is being done to give full consideration to the witnesses that we do have.