Chapter 2. Sunt Aliquid Manes: Ancient Quotations of Homer

A multitextual approach to Homeric epic acknowledges and even embraces an expected amount of variation between performances of oral poetry. Because this multiformity was generated within a system, the attested variations enable us to appreciate the poetry of the Homeric epics on more than just the level of a single performance. By adopting a multitextual approach, we can, following Albert Lord, train our ears to hear the echoes of many past performances and appreciate the richness and complexity of the Homeric tradition as it evolved through time. [1]

One of the earliest allusions to the text of the Iliad in ancient literature contains a well-known and much discussed multiform. The text in question is Aeschylus, Suppliants 800–801, in which the chorus sings: The passage does not quote Iliad 1.4–5, but certainly seems to invoke it for an Athenian audience that would have been well versed in the Homeric epics. Here is the text of the Iliad passage as it is transmitted in the Venetus A manuscript: But Aeschylus seems to have known a different text. [2] There is no equivalent to πᾶσι in Aeschylus. And in fact we are told by Athenaeus (Epitome 1.12) that the Alexandrian editor Zenodotus read δαῖτα here at Iliad 1.5. It would seem that δαῖτα is an ancient multiform that was known as early as the fifth century BCE. Meanwhile, all other sources read the equally Homeric πᾶσι. [3] Gregory Nagy has written of this passage, “Both variants are traditional multiforms. In a multitextual format of editing Homer, we would have to take both forms into account” (1996a:134).

Indeed, as this and the following chapters will show, the earliest attestations of the Homeric texts are exceedingly multiform in comparison with the relatively uniform medieval manuscripts. A brief look at the first ten lines of the Iliad will provide a sense of the complexity of the transmission that we will be confronting: Numerous multiforms for these verses are attested, and they have a good deal of ancient support, both within the formulaic diction of the Iliad itself and in the debates of the Alexandrian editors (as preserved in the scholia of medieval manuscripts). The reading κεφαλὰς, for example, in line 3 was read by Aristophanes of Rhodes. At Iliad 11.55 we find powerful evidence that κεφαλὰς is perfectly formulaic: there this same verse is attested with κεφαλὰς in place of ψυχὰς. Meanwhile, the Venetus A scholia record that “some” (τινες) write κεφαλὰς instead of ψυχὰς at 1.3, “badly” (κακῶς) in the judgment of the scholiast. The scholia in the Venetus A also tell us that Zenodotus athetized lines 4 and 5, meaning he did not deem them Homeric, but Zenodotus is credited in Athenaeus with reading δαῖτα in the athetized verse 5. [4]

The multiformity does not stop there. An edition known to scholars in the ancient world apparently contained this single verse in place of Iliad 1.1–9: This was the edition of Apellicon known by Crates and Nicanor (Erbse 1969:3). Another version of the proem current in antiquity, according to Aristoxenus, consisted of three verses: The two variant proems come from a scholion found in a manuscript called the Anecdotum Romanum, or Ve1 (in the edition of Allen [1931]) or Z (in the edition of West [1998–2000]). Ve1 consists of two manuscripts of scholia on the Iliad which were once part of a single whole, one from Rome (Biblioteca Nazionale Vittorio Emanuele 6), and the other from Madrid (Biblioteca Nacional de España 4626). [5] The manuscript is as old as or possibly even older than the Venetus A, dating to the ninth or tenth century CE, but the information contained in it is much earlier, in that it preserves ancient scholarship dating back to Hellenistic and Roman times.

***

As the multiforms in the Iliad proem attest, we have a wealth of ancient material, including quotations in ancient authors, papyri, and the scholia of medieval manuscripts, from which we can gain an appreciation of the multiformity of the oral traditional system within which the Iliad and the Odyssey were composed. In this and the following chapters I propose to proceed chronologically and discuss each of these ancient sources separately, although there will be many points of contact between them. In the remainder of this chapter I seek to point out the possibility and fruitfulness of recovering more fluid stages of the Homeric Iliad and Odyssey from quotations (especially those of the fourth century BCE). Quotations in fourth-century authors such as Plato and Aeschines are among our very oldest witnesses to the Homeric texts. They are a crucial primary source for any attempt at assessing the state of the Homeric text in the fourth century BCE (Nagy’s period 4 in his evolutionary model, the time of our earliest potential transcripts). [6]

The quotations are not only our earliest witnesses but also among the most multiform. It is not uncommon for quotations to contain additional verses (as compared to medieval manuscripts) or fewer verses, or to exhibit multiformity within a particular verse. As we will see in chapter 3, the quotations are not unlike the oldest surviving papyrus texts, the ones that have been deemed “wild” in relation to the medieval transmission. In fact, the further back in time we go, the more multiformity we find in our textual witnesses. This is what we should expect to find in the textual transmission of an oral poem, of which there can be no “original.” [7]

And yet, two recent critical editions of the Iliad and the Odyssey, those by H. van Thiel and M. L. West, explicitly disregard the variations presented by ancient quotations. Van Thiel deems ancient quotations and papyri “of minor importance” and treats any multiforms offered by a quotation “as if it were conjecture” (1991:xxi). These scholars are not interested in multiformity for its own sake; they seek rather to establish an authoritative text that can be presented in a printed edition. [8] West understands the multiforms that survive in early quotations to be “interpolations” (1998:vii). [9] Alternatively West and others explain the multiformity of early quotations as lapses of memory, under the common assumption that fourth-century authors like Plato and Aeschines did not have texts before them when quoting Homer. [10] Scholars typically disparage as “banal” [11] or even “inept” [12] (and therefore not worth our consideration) the types of variation that fourth-century quotations and early papyri present. [13] But not only does this criticism do nothing to solve the problem of what to do with this multiformity: it is also untrue.

I and my collaborators on the Homer Multitext are not seeking to establish a single definitive text, and therefore we approach the early quotations from a different perspective. What do the variations preserved in fourth-century quotations tell us about the early textual history of the Homeric poems? What Iliads and Odysseys were audiences at various points in antiquity actually hearing? What impact can the attested multiforms have on our understanding of the poetics of the Iliad and the Odyssey? I will discuss in this chapter two examples of the kinds of multiformity presented by the early quotations, in order to make a case for their value not only as witnesses to the oral traditional system in which the Iliad and the Odyssey were composed, but also for their intrinsic value within the history of literature. I would like to begin my discussion of these questions with what may seem to be an extremely, even perversely, minor textual variant, the difference of a single letter in Greek. I will argue that even a variation as small as a single letter has wide-ranging implications for our understanding of the Iliad and the transmission and reception of the poem in antiquity. We will get there by way of the Roman elegiac poet Propertius.

***

Propertius begins IV 7, a poem that alludes throughout to the dream of Achilles in Iliad 23.62–107, with a simple and emphatic declaration: sunt aliquid manes. The Shades are something. Commentaries point out that these opening words allude specifically to Iliad 23.103–104: The three words that begin Propertius’ poem, moreover, fit within an overarching structural allusion in which his lover Cynthia, like Patroklos, appears to Propertius in a dream after her death and reproaches him for neglect of her funeral rites. Sunt aliquid manes is a verbal echo of the exclamation of Achilles, who, after attempting in vain to embrace the shade of Patroklos, suddenly comprehends the nature of the ψυχή after death.

Propertius alludes directly to Iliad 23.65–104 in five more places in the poem: [15]

• Cynthia namque meo uisa est incumbere fulcro (3) ~ στῆ δ’ ἄρ’ ὑπὲρ κεφαλῆς καὶ μιν πρὸς μῦθον ἔειπεν· (Iliad 23.68)
• eosdem habuit secum quibus est elata capillos, / eosdem oculos; lateri uestis adusta fuit (7–8) ~ ἦλθε δ’ ἐπὶ ψυχὴ Πατροκλῆος δειλοῖο / πάντ’ αὐτῷ μέγεθός τε καὶ ὄμματα κάλ’ εἰκυῖα / καὶ φωνήν, καὶ τοῖα περὶ χροῒ εἵματα ἕστο· (Iliad 23.65–67)
• perfide nec cuiquam melior sperande puellae, / in te iam uires somnus habere potest? (13–14) ~ εὕδεις, αὐτὰρ ἐμεῖο λελασμένος ἔπλευ Ἀχιλλεῦ. / οὐ μέν μευ ζώοντος ἀκήδεις ἀλλὰ θανόντος· (Iliad 23.69–70)
• mecum eris, et mixtis ossibus ossa teram (94) ~ μὴ ἐμα σῶν ἀπάνευθε τιθήμεναι ὀστέ’ Ἀχιλλεῦ (Iliad 23.83)
• haec postquam querula mecum sub lite peregit, / inter complexus excidit umbra meos. (95–96) ~ Ὣς ἄρα φωνήσας ὠρέξατο χερσὶ φίλῃσιν / οὐδ’ ἔλαβε· ψυχὴ δὲ κατὰ χθονὸς ἠΰτε καπνὸς / ᾤχετο τετριγυῖα· (Iliad 23.99–101)

These parallel passages are very close and present no difficulties. They provide a framework that allows Propertius to be in dialogue with the Homeric text and at the same time depart from it.

The allusion in line 1 by which Propertius signals this dialogue with the Iliad is not quite exact, however. If the larger context did not make it certain, we might question whether sunt aliquid manes should be considered an allusion at all. The Homeric text that I have printed above is that of the Venetus A, as well as that of most modern editions (with some differences of accentuation). [16] The crucial difference between the standard printed text of Homer and the wording that we might expect were we to translate backwards from Propertius is the reading of τις in line 103, rather than the neuter τι. The Homeric text can be translated for the moment as: “Alas there is indeed someone, even in the house of Hades / a soul and a likeness, but there is no real substance.” Whereas Propertius reads: “The Shades are something: Death does not end everything.”

What could account for such a discrepancy? It seems that Propertius indeed had in mind a different text. τις is the majority reading of the manuscripts in the Iliad passage, and Eustathius is an early witness. But τι is also a possible reading. It is found in twenty-seven manuscripts of the Iliad, and it has strong support in the form of a far earlier witness. This is Plato’s Republic (386 D). In the Plato passage Socrates discusses the stories that poets tell about the nature of the underworld, concerning which he gives seven citations of Homer in quick succession, including lines 103–104 of Iliad 23. Plato’s text is considerably closer to that of Propertius: ὢ πόποι, ἦ ῥά τι ἔστι καὶ εἰν Ἀΐδαο δόμοισι, / ψυχὴ καὶ εἴδωλον, ἀτὰρ φρένες οὐκ ἔνι πάμπαν. “There is something, even in the house of Hades, a ψυχὴ and an εἴδωλον…” or, without a sense break after δόμοισι (which is how Propertius must have read it) “the ψυχὴ καὶ εἴδωλον are something, even in the house of Hades.…”

Now τι is certainly a possible valid reading at Iliad 23.103, metaphysical questions aside, and Leaf in fact prints it. [17] There is support in the medieval manuscripts of the Iliad for this reading, although it is not the majority reading, nor is it in the oldest manuscripts. [18] Secondly, Plato definitely wrote τι in his citation of these lines. The only support for τις in manuscripts of Plato, the manuscript known as T, is itself an apograph of A, the ninth- or tenth-century manuscript on which modern texts are based, which reads τι. τις is therefore a corruption unique to the manuscript T in Plato. [19] Plato is our earliest witness to the text of Homer at these lines, far earlier than any papyrus and well over a thousand years before the medieval tradition of the text of Homer begins. Another very old witness, a first-century BCE papyrus (511), also reads τι. [20] Finally, Propertius, a first-century BCE witness, seems to have known the reading τι and translated it as aliquid in the line sunt aliquid manes.

When I first explored this variation in a 2001 article (Dué 2001b), I argued that Propertius may have known both readings. By translating the reading τι found in Plato (and possibly other sources) rather than τις (which may well have been in most texts of Homer, even in the fourth century BCE), Propertius, I argued, was engaging in an Alexandrian method of alluding to Homer. He displays Alexandrian learnedness by alluding to the variant readings of Homer known to him and to the textual problems, or z e ¯t e ¯mata, encountered in the work of scholars like Zenodotus and Aristarchus on the text of Homer. [21]

I made this argument in part because the first line of Propertius’ poem alludes to much more than a minor textual discrepancy. The meaning of the Homeric lines was in fact a major topic of discussion for Homer’s readers in antiquity, and the scholia that survive in medieval manuscripts show us that Alexandrian scholars struggled to provide possible explanations. [22] Any interpretation of the lines has to take into account that single sigma that is the difference between τις and τι. First of all, translating τι, how can the εἴδωλον be said to be something anywhere other than in the house of Hades? The same question could be applied to τις: there is a ψυχὴ and an εἴδωλον even in the house of Hades? In both cases the force of the καὶ is awkward. But if we translate τι and take ψυχὴ καὶ εἴδωλον in apposition (a grammatical construction very natural to Homeric Greek), Achilles seems to be wrestling with the exact nature of what remains of life after death. Something remains of Patroklos’ former self in Hades, that is a ψυχὴ and an εἴδωλον, but then he qualifies it—ἀτὰρ φρένες οὐκ ἔνι παμπαν.

This qualification is just as difficult to interpret. What exactly is the distinction being made between ψυχὴ καὶ εἴδωλον? Patroklos has just given a long speech with no apparent lack of reason, as is noted in the scholia of the Venetus A (ἐμφρόως καὶ συνετῶς διείλεκται πάντα ὁ Πάτροκλος). The intelligence displayed in Patroklos’ speech troubled Aristophanes enough for him to suggest that φρένες refers to the physical container of reason—that is, corporeality. [23] Achilles has after all just tried to embrace Patroklos and failed. Aristarchus on the other hand commented that for Homer the souls of the unburied dead still have their intelligence, as opposed to the buried souls whom Odysseus encounters in Hades in Odyssey xi. [24]

There are still more explanations to be found in the scholia on these lines in other manuscripts. One is that Patroklos shows lack of φρένες in rebuking Achilles for neglecting his funeral rites. [25] Achilles has indeed already completed the greater part of the preparations for burial and performed most of the rituals on that very day, and the shade of Patroklos visits Achilles on the eve of the planned burial. A final possibility is found in the word πάμπαν, which seems to further qualify Achilles’ statement. The shades have sense or intelligence, but it is somewhat diminished.

I do not intend to try to solve the interpretive difficulties here, which have confounded Alexandrian and modern scholars alike. [26] It was sufficient for the purposes of my 2001 article to point out that the meaning of these lines was controversial in antiquity, and there I explored what the poetic significance of that Homeric debate means for our interpretation of Propertius IV 7, which is steeped in poetic traditions about the afterlife and philosophical questions about the nature of the soul. But in that article I was also making a larger point about the Homeric text, namely that both readings, τις and τι, have authority, that both may have been performed in antiquity, and that a multitextual approach to Homer allows us to consider both. If we do not have to privilege one reading over the other, how does that change how we approach the meaning of the passage and its reception at various points in history?

In that same 2001 article I suggested that a poem by Ovid might just prove that Propertius was in fact engaging a known textual issue—that is, he was confronting the multiformity of the Homeric textual tradition—by translating τι. Amores III 9, a poem in which critics have found connections with Propertius’ fourth book, [27] is a formal lament for the death of Ovid’s fellow elegist Tibullus. The thematic connections with Propertius IV 7 (and Lucretius, also discussed in the article) are obvious: the same questions about the immortality of poetry and the soul are explored. [28] Like Cynthia, Delia and Nemesis will live forever, according to Ovid (31). Homer plays an important role in this poem as well: Ovid’s lament is compared to laments for Achilles and Memnon (1–2), and Ovid notes that even Homer had to die, but now his poetry lives on (25–30). Towards the end of the poem, Ovid speculates as to whether a soul lives on after the death of the body. He formulates his question twice, each time as an indirect question within a conditional sentence. The first begins: si tamen e nobis aliquid nisi nomen et umbra restat… (“If nevertheless something does remain of us beyond a name and a shade…”). With this formulation Ovid questions whether there is something even beyond the existence of a shade after death. The second formulation is somewhat different: siqua est modo corporis umbra (“if only there does exist a shade of the body…”). Here he seems to question the existence of a shade at all after death. And once again, mixed up in the whole philosophical question is the key textual difference between the two formulations: the adjective aliquis versus the pronoun aliquid. Or, to put it in Greek, τις versus τι. [29]

Whether or not Propertius knew what he was doing when he read τι (and I have tried to show that he did), he set in motion a complex web of allusion and intertext with long-lasting effect. [30] We should wonder then if there is a deeper significance to Propertius’ allusion beyond a mere display of Alexandrian learning. Dorothy Lange has suggested that IV 7 is not merely a playful “twist on the amatory elegy” but rather a poetic statement that marks Propertius’ farewell to erotic poetry. [31] Achilles’ final vision of Patroklos, so full of regret and so tender, is a wonderful metaphor for such a farewell. But by making the shades something, Propertius has given Cynthia (and erotic poetry) the power to come back.

***

I turn now to a more complex example of multiformity, Aeschines’ quotation of Iliad 23.77–91 in his speech Against Timarchus (149). [32] The quotation shows three plus verses and significant internal variation in three separate lines. [33] As we will see, these multiforms are arguably just as “Homeric” as the text preserved in medieval manuscripts such as the Venetus A, and they illustrate, even more so than my first example, the necessity of taking a multitextual approach to Homeric epic.

Throughout his prosecution of Timarchus Aeschines provides carefully selected citations of Homer and the tragedians to support his claim that Timarchus has led the kind of life that, according to Athenian law, precludes him from speaking in the democratic assembly. The following is Aeschines’ text of Patroklos’ address to Achilles in Iliad 23, which the speaker cites in an attempt to refute the defense’s claim that Homer highly approved of sexual relationships between men. Aeschines’ point is that Homer never explicitly says that Achilles and Patroklos were lovers; rather, the speaker argues, they possessed the noblest of friendships (τὴν ἀρετὴν καὶ τὴν φιλίαν ἄξιον αὐτῶν 146). [34]

Compare the text of the medieval Venetus A manuscript, including line 92 [≈ 83b], which was athetized by Aristarchus and not quoted by Aeschines: Among other variations, this version is shorter than Aeschines’ quotation by at least three verses. We do not know if Aeschines’ text contained line 92 and its reference to the golden amphora of 83b, because his quotation breaks off at 91. [36]

The arguments typically used to dismiss similar kinds of multiformity in the Ptolemaic papyri will not work here. Aeschines’ quotation is a witness older than any papyrus or manuscript, and has not undergone any scholarly editing by the Alexandrians. It has been assumed that this text was provided to a court reporter, who read aloud from it when called upon by Aeschines to do so, starting and stopping where he indicated. [37] Whether that is indeed the case, or some other text was read aloud and included in Aeschines’ published speech, a lapse in memory on the part of Aeschines is not a solution to the variation within lines. The official nature of the situation in which the quotation took place makes it unlikely that the passage is an eccentric private copy of the sort that a collector might possess, but rather the contrary. A standard wording is what is called for here, and one that was familiar to the jury if it was to carry any weight.

To confront this situation, in which a fourth-century BCE witness challenges the medieval transmission, a scholarly consensus seems to have been reached which asserts that verses 83a and 83b are an interpolation “effecting the equation of the σορός of 91 with the golden amphora which Thetis provides for Achilles’ bones at Odyssey 24.73–77” (Haslam 1997:76). [38] In this passage from the Odyssey, the ghost of Agamemnon in the underworld recalls the funeral of Achilles: The golden amphora referenced here would appear to be the same one depicted on the so-called François Vase, a large Archaic black-figure vase that prominently features the wedding of Peleus and Thetis, among other scenes connected to the life of Achilles. [39] On that vase, Dionysos can be seen carrying an amphora on his back as he proceeds with the other gods to the wedding (see Plate 7A). Rather than view these multiple attestations of the golden amphora as confirmation of its traditional place in Greek myth and in Homeric poetry, some scholars assert that the Odyssey passage actually calls Iliad 23.92 into question. Michael Haslam has argued in connection with the Odyssey passage: “discussions of the amphora [on the François Vase] generally fail to realize that the verse [23.92] is an interpolation, and that the jar’s only Homeric occurrence—provided we define Od. 24 as Homeric—is in Od. 24” (Haslam 1991:36). Proponents of this explanation, including Haslam, point to Aristarchus’ athetesis of line 92, a verse that is also missing in papyrus 12, a mid-third-century BCE text. [40] They argue that verses 83a, 83b, and 92 are alternative means of bringing the Iliad passage into alignment with that of the Odyssey. (By this line of reasoning, both 83a–b and 92 are interpolations and therefore not “Homeric.”) They therefore maintain that the text from which Aeschines made his citation did not contain 92.

Allen saw long ago that this supposition is by no means demonstrable: “Aeschines’ quotation stops at 91; it is therefore impossible to say with certainty that he omitted 92, the sense of which he had already given.” [41] And yet most modern scholars do not even admit this possibility, which would weaken the case for interpolation. Giorgio Pasquali argues circularly that the presence of 83a and 83b in Aeschines guarantees that his text did not read 92. [42] Stephanie West declares that 92 is “replaced” by 83a and 83b. [43] Both Janko and Richardson in their respective discussions in the Cambridge commentary state simply—without even hinting that we have no way of being absolutely certain that this is so—that the text of Aeschines omits 92. [44]

In the face of such a unified front, I propose to reformulate, adding some of my own examples, the general argument of Aldo Di Luzio (1969), who has shown persuasively that 83a and 83b are not likely to be interpolations based on the Odyssey 24 passage. Di Luzio argues instead that 83a and 83b were originally in the text along with 92. I omit for now but will return later to the question of whether to include 81a or the internal variations in 77 and 82. I also omit for now but will return later to the even larger question of what “interpolation” even means for an orally composed poem.

Di Luzio’s presentation of the text reads as follows: In this version of the text, there is ring composition as in nearly every speech in Homer. This ring composition is reinforced by the repetition of the reference to the golden amphora in 83b. The ring composition is further enhanced by the parallelism between ὁμοίη in 83a and ὁμή in 91 as well as the ὀστέα of 83 and 91. The verses 83–83a then correspond to verse 91, while 83b corresponds to 92. [45]

Di Luzio’s arguments demonstrate that the passage as printed above is just as “Homeric” as the medievally transmitted Iliad, and possibly more so. He writes: “con 83ab, il passo 83–92 manifesterebbe una struttura isomorfica con quella di altri discorsi del testo epico in cui le frasi alla fine del discorso sono spesso la ripresa di frasi occorrenti all’ inizio di esso” (Di Luzio 1969:83). Not only is the overall structure Homeric, but the individual constructions within this structure correspond to Homeric usage as well. Together 83a and 83b form the familiar pattern of a negative protasis followed by a reinforcing positive apodosis that expresses the same idea. Compare these lines from the Odyssey: The three-verse structure here is nearly identical to that of 83–83b. We can observe the same phenomenon in a more compressed form in the phrase “ἔοικέ τοι, οὔ τοι ἀεικές” (Iliad 9.70). Moreover, within this structure Di Luzio distinguishes still further examples of Homeric usage (Di Luzio 1969:84–86). The first half of 83b (χρυσέῳ ἐν ἀμφιφορεῖ) consists of a complementary locative phrase which specifies the preceding clause. The second half of the line is a parenthetical relative clause of a kind that is used frequently in Homeric poetry to designate particular objects. A parallel is Iliad 4.215–216: λῦσε δέ οἱ ζωστῆρα παναίολον ἠδ’ ὑπένερθε / ζῶμά τε καὶ μίτρην, τὴν χαλκῆες κάμον ἄνδρες. There are numerous other examples.

Di Luzio’s compelling analysis illustrates precisely why variants presented by fourth-century quotations and the Ptolemaic papyri cannot be dismissed as inept or unworthy. Thus there is room to disagree with Stephanie West’s analysis of this passage. She writes: “The objections to this version are obvious: 83a seems to be based on Σ 329, but barely makes sense; its insertion makes the construction of Patroclus’ speech very confused” (West 1967:172). And yet, as we have seen, there is nothing inherently objectionable within the lines themselves. They simply do not survive in the medieval transmission of the text.

Nor does 92 present problems of usage. If we analyze it independently of 83b, we see that it contains an appositional phrase consisting of a synonym that specifies a preceding noun, the σορός of 91 (Di Luzio 1969:84). [47] To this we can compare the following examples: In the first example the name Altes is repeated in the same case in order to introduce a specifying relative clause. In the second example there is a similar construction, but πάντα of line 120 is further specified by the apposition of κτῆσιν at the beginning of line 121. I adduce the last example because of its striking similarity to 23.92, the very verse in question. Like 83a and 83b, it seems that on its own 92 presents no difficulties.

The problem must originate in the existence of both 83a–b and 92 together at some pre-Alexandrian stage of the text. As Di Luzio points out, this sort of doubling, though unquestionably Homeric, was just the sort of thing that troubled Alexandrian critics. It is only to be expected that one or the other would have been omitted by Aristarchus’ time in many texts. And indeed it seems that 83a and 83b had almost completely dropped out of the tradition, in deference to 92, by Aristarchus’ day but that remnants of an alternative excision of 92 still remained. The T scholiast states: ἐν πάσαις οὐκ ἦν ὁ στίχος. καὶ Ἀρίσταρχος ἐκ τῆς Νεκυίας αὐτὸν ἐσπάσθαι φησίν (“The verse is not in all. And Aristarchus says that it was taken from the Nekuia”). Di Luzio suggests that Aristarchus conjectured that verse 92 was an interpolation based on the Odyssey only after he felt compelled to explain why it was not present in all the texts available to him (Di Luzio 1969:83). This scenario is all the more likely if some of Aristarchus’ texts no longer contained 83a and 83b.

Nor is it surprising that verses 83a and 83b lost out so decisively to verse 92 when we consider that Iliad 22.121, discussed above, has likewise disappeared from a large majority of manuscripts (including papyrus 27). [49] Di Luzio’s examples exhaustively demonstrate that this sort of construction is Homeric, yet its repetitive quality troubled editors enough to jeopardize the place of these verses in the tradition. Verses 83a and 83b had double reason to be suspect in the eyes of the Alexandrians: they contained the repetitive negative and positive formation like that of 22.121 on the one hand, and on the other they nearly duplicated 92.

My formulation of the history of the passage is the opposite of that of Haslam, West, Janko, and Richardson. I posit the loss of 83a–b or 92 in all or most texts between the fourth and second centuries BCE rather than their insertion sometime prior to Aeschines’ quotation of them. It is easier to understand the loss of such lines that we know to have been troublesome for scholars of the Hellenistic era than to postulate an interpolation in a text as early as that of Aeschines. The attribution of interpolation to such an early text indeed stretches the limits of the term, for there could have been scant means and little motive to interpolate in a time when literacy was limited and performance was still alive and well as the primary means of access to the Homeric epics for most people. T. W. Allen hit the nail on the head in 1924 when he wrote, “The origin of additions and omissions cannot be referred to anything except to the recitation of rhapsodes, and therefore the phenomenon must have existed and presumably been most frequent during the period when a reading public hardly existed” (267). If we understand “performance” in place of Allen’s “recitation” (which implies memorization) we have here an early formulation of what Parry and Lord would later demonstrate by way of fieldwork. What Allen could not have realized is that these additions and omissions are not interpolations but genuine performance-derived multiforms no less “Homeric” than those that survive in the medieval transmission.

Not only is there no reason to theorize that 83a–b/92 is an interpolation, such a theory does nothing to explain other multiforms in the passage that are equally difficult to explain by conventional text-critical methods. The first words of the first line of the passage are not the same as we find in medieval manuscripts: οὐ γὰρ ἔτι versus οὐ μὲν γὰρ (ζωοί γε φίλων ἀπάνευθεν ἑταίρων). The phrase οὐ γὰρ ἔτι is certainly Homeric; it occurs four times in our Iliad and Odyssey. The phrase οὐ μὲν γὰρ, on the other hand, occurs forty-three times. Van der Valk objects to Aeschines’ reading, stating simply: “It is obvious that the solemn formula of the Homeric mss. οὐ μὲν γὰρ represents the original text” (1964:327n230). It is not obvious at all: the number of occurrences suggests that οὐ γὰρ ἔτι is the more marked form and arguably the more solemn. But by far the most remarkable evidence in support of Aeschines’ reading is the fact that the A scholia attest that this was the reading in some of the “city editions” (ἔν τισι τῶν πολιτικῶν). An old attestation with ancient authority, it deserves at least as much consideration as the text transmitted into medieval times.

Line 81a, another plus verse, holds up to similar scrutiny. This line is perfectly formulaic. The phrase μάρνασθαι δηΐοισιν occurs in verse-initial position four times in our Iliad (9.317, 11.190, 11.205, 17.148). The participle μαρνάμενον is likewise attested in various cases in verse-initial position throughout the Iliad: μαρνάμενον 3.307, 6.204; μαρνάμενοι 6.257, 6.328, 14.25; μαρναμένων 12.429, 13.579, 16.775; μαρναμένοισι 13.96. The phrase Ἑλένης ἕνεκ’ ἠυκόμοιο is also found at Iliad 9.339, and there are six instances of the phrase Ἑλένης πόσις ἠυκόμοιο, all in verse-final position. [50] Others have viewed the formulaic nature of verse 81a as proof of interpolation. [51] But this view is fundamentally at odds with the findings of Lord that in the system within which the Iliad and the Odyssey were composed, every line is formulaic. [52] Stephanie West (1967:13) is sympathetic to the idea of interpolation, but acknowledges: West here distinguishes between two types of interpolation: those composed by the “interpolator,” and those taken out of their original context and inserted improperly by the “interpolator” into another passage. West’s own analysis suggests that the plus verses are as “Homeric” as medievally transmitted verses, but she nevertheless treats them as “interpolations” that have no place in our texts of the Iliad and the Odyssey.

Line 23.82 of Aeschines’ quotation presents a comparable difficulty for those who insist on the aberrance of the fourth-century quotations of Homer. The Venetus A at this line reads ἄλλο δέ τοι ἐρέω καὶ ἐφήσομαι αἴ κε πίθηαι. As with 77, the variation is internal to the line and therefore is not a question of plus or minus verses (or the numerus versuum, as it is often referred to by scholars). Line 82 is the only instance of this verse in Homer, although the construction is a familiar one. (Cf. Iliad 1.207: ἦλθον ἐγὼ παύσουσα τὸ σὸν μένος, αἴ κε πίθηαι.) The phraseology of Aeschines’ reading, on the other hand (ἄλλο δέ τοι ἐρέω σὺ δ’ ἐνὶ φρεσὶ βάλλεο σῇσιν), is attested eleven times in the medieval texts of the Iliad. I do not suggest that Aeschines’ reading is superior or more “Homeric”: I merely propose that it is no less so.

On this point the positions taken by editors break down. The work of Di Luzio, Labarbe, Haslam, and others has challenged the negative assumptions scholars make about the validity of ancient multiforms. But where the intent of Di Luzio, for example, is to find the ipsissima verba of the one true Homer, the editor of a printed edition must necessarily privilege one reading over another in every case. Gennaro D’Ippolito (1977), to cite another example, prides himself on being the only editor to include plus verses in his text of Odyssey 5, [53] but his acceptance of the authenticity of plus verses does not solve the question of what to do with variation such as what we find in Iliad 23.82, or the situation in which a papyrus or quotation presents an entire formulaic line in place of another. [54] Within the model of the single monumental composer that D’Ippolito proposes, they cannot both be accepted. Janko and others, on the other hand, accept the idea of an Aristarchean numerus versuum, but do not admit that the textual variants that Aristarchus records can be genuine performance variants and worthy of inclusion in the text. [55] Like D’Ippolito, they have no solution for multiple readings, mutually exclusive in their view, that occur in the same line or formular unit—other than the elevation of one reading at the expense of others which may be equally genuine. [56]

The substitution of one formula for another is part of the poetics of the oral-formulaic system. Parry (1932:15 [= 1971:336]) demonstrated the significance of this for the text of Homer. He writes: Variation is a clear sign of the oral poetics of recomposition in performance and no one formula is more or less Homeric than another. Here we come to the crux of the problem. What do textual critics of Homer mean when they say that a verse has been interpolated? [57] For Parry the term “interpolation” would be applicable neither to plus verses nor to variations such as that found in line 82 of Aeschines’ quotation.

This passage challenges the way we deal with multiforms, particularly those that are demonstrably formulaic and in strict accordance with Homeric usage. The variations take the form of both internal variation within lines and fluctuation in the numerus versuum resulting from expansion and compression at the performance level. [58] Many of the features that are Homeric in 83–92 are present in a more compressed form even if 83a, 83b, and 92 are eliminated from the text. Let us consider: The variation found within verse 84, caused by the fluctuation between the shorter and longer version, has been eliminated. There is still ring composition and verbal repetition of ὀστέα in 83 and 91. Verses 83–84 alone are a somewhat elliptical version of the construction featuring a negative protasis / positive reinforcing apodosis—if we understand an ellipse of τιθήμεναι in 84. The presence of 83a–b and 92 expands and reinforces these features. The textual critic who wishes to print an edition of the Iliad is forced to choose between these two quite different texts, either of which is defensible in terms of Homeric usage.

For the Alexandrian editors, that choice was nearly always in favor of the shorter of two variants. As Di Luzio points out, many characteristics of oral poetry such as the adding of description, specification, recapitulation, and repetition, all connected paratactically or in apposition, were antithetical to the aesthetics of the day in much the same way that they are now. [59] Verses that contained these features met with a very difficult reception. The plus verses which are so weakly attested and other verses which have survived medieval transmission but which we know to have been athetized or omitted by Alexandrian editors are proof of this. Their possession of these same features surely accounts for the disappearance of plus verses from most texts that survived to Aristarchus’ day.

***

How does the presence or absence of the golden amphora in this passage affect our larger Iliad? If we take a step beyond the history of transmission and interpret the significance of Aeschines’ quoted multiforms from an artistic or “literary” perspective, we can see that in fact the issue is neither minor nor banal. Homeric multiforms can and do present vastly different narrative consequences. These narrative consequences are not accidental, nor are they ad hoc inventions; rather, multiformity can signal alternative performance traditions that are not incorporated into our Iliad and Odyssey. In this case, the presence or absence of the golden amphora has not only a literary but also a cultural and religious consequence: it is a case of signaling or not signaling the reassembly of Achilles’ bones into an immortalized hero. [60]

Throughout the Iliad there is reference to a larger event that will take place outside of the confines of the epic, and that is the death of Achilles. We are constantly reminded of his short life and impending doom, close on the heels of Hektor’s. [61] The presence here of the golden amphora, which Achilles’ mother gave him in anticipation of his approaching death, evokes in Patroklos’ reminder not only that impending death but its aftermath. [62] Our awareness of the presence or absence of the golden amphora in the narrative of Achilles’ death indicates whether he will achieve immortality through cult or pass into obscurity in the underworld. [63] Although the Iliad centers on Achilles’ mortality, here we have a glimpse of the immortality that is so important in the tradition outside of the Iliad. [64] Andrew Stewart (1983) has argued that the golden amphora is the focal point for the two conceptions of Achilles on the François Vase as well: he understands the compositional unity of the François Vase to be centered on the golden amphora depicted on it. The wedding of Peleus and Thetis is set amidst narratives that explore the tensions between mortality and immortality, peerless excellence and savage wrath, and mighty prowess and terrible hubris in the figure of Achilles. Stewart notes: “Appropriately, all these themes intersect in the motif of Dionysos’ amphora and its twin promises of death and immortality” (1983:66). The golden amphora points to a critical dichotomy in how the Achilles story ends.

***

Fourth-century quotations of Homer are some of our oldest witnesses and preserve information about the various epic traditions that we have otherwise lost. In one of those traditions, Achilles was not left to dwell in Hades, lamenting the choice he had made to re-enter the fight and now preferring a life of servitude to death with κλέος (as in Odyssey 11.487–491). Instead his bones were assembled and placed in a golden amphora along with those of Patroklos, thereby securing his immortality in cult. A multitextual approach can embrace both possibilities, as well as demonstrate the reality of oral performance and transmission, which requires that no oral text will ever be composed, performed, and recorded the same way twice.

Many of the issues raised by the two quotations examined here will need to be confronted again in connection with the Homeric papyri, which are discussed in the next chapter and which are similarly multiform. I would like to conclude the present chapter by making two final points about how we treat the multiforms preserved in the quotations.

First, the question of just what text a court reporter would have read out to the jury in the fourth century BCE is a good one. We have no direct evidence that an official state text existed for Homer or any other poetry at this time. [65] Elsewhere I have suggested that the later Lycurgan law that called for official state copies of the tragedies of Aeschylus, Sophocles, and Euripides to be placed in the Metroon had little to do with the establishment of the text and more to do with symbolically elevating the poetry to the status of law (Dué 2000). We simply cannot assume—nor should we expect—that at this transmissional stage any one text of Homer was canonical.

Nevertheless, on principle I do not believe, as has been asserted by Van der Valk, that Aeschines could have or would have deliberately altered the text either on the occasion of delivering the speech or in its subsequent publication. [66] Wherever ancient citations of Homer differ from the medieval transmission, I submit that there are generally reasons other than deliberate alteration by the quoter. Aeschines argues with the defense about the interpretation of the cited text, not about which text is the correct one. Altering Homer to suit an argument and paraphrasing (as sometimes occurs in Plato) are different matters entirely. [67] Quoting Homer is not the same as quoting the Bible, as Andrew Ford (1999) has pointed out, but Homeric poetry carried an undeniable authority that makes the notion of interpolation—even overlooking the difficulty of identifying a canonical text to be interpolated—extremely problematic. [68]

The preceding analysis shows that the multiformity presented by Aeschines’ quotation of Iliad 23 is of a formulaic nature that is in no way incompatible with what we know of the system as a whole. Rather than try to find ways to dismiss, denigrate, or otherwise obfuscate such multiformity, I submit that it is better to treat each of the surviving quotations as valid, even if fragmentary, instantiations of the Iliad or the Odyssey in their own right. Allowing each instantiation to enter our awareness and our appreciation of the Iliad and the Odyssey as orally created poems is problematic only for those who insist that there is one text alone that can be authoritative, that composed by the postulated “Homer” himself. Albert Lord explains why this is not the right approach for oral poetry (1960:100–101): By imposing an authorial model on what has survived of Homeric poetry from antiquity, modern editors necessarily exclude an abundance of alternative instantiations of the Iliad and the Odyssey, and so a wealth of data about the system in which those two poems were composed.

Just as problematic is the fact that this same authorial approach is the one taken by the vast majority of editors of al l Greek and Roman authors, not just Homer. At the outset of this chapter we saw that Plato, like Propertius, read τι at Iliad 23.103. But until relatively recently few modern scholars could know it. When I wrote my 2001 article on this multiform, I noted that although the A and F manuscripts of Plato on which modern texts are based read τι, all modern editors of Plato that I could locate at that time printed τις. These editions make this slight but significant change in order to bring Plato’s text into alignment with the medievally transmitted Iliad, whose reading of τις, I stress, is by no means certain. In addition, only one of the editors of the more than ten editions of the Republic that I consulted indicates in the apparatus criticus that they have done so. This is a dangerous practice, and it is based on two flawed assumptions: that fourth-century quotations are not reliable witnesses to the text of Homer and that somehow the same editorial principles that apply to the rest of the text of Plato do not apply to quotations. Fortunately, the practice appears to be changing. S. R. Slings’s 2003 edition of the Republic prints τι, while noting in the apparatus that most Homer manuscripts have τις.

It is not a defensible editorial practice to alter Plato’s text here or elsewhere to bring it into alignment with our concept of a Homeric “vulgate.” No such vulgate existed during the fourth century BCE at least. [69] This can be proven from the numerous fourth-century quotations of Homer in Plato and orators such as Aeschines that do not accord with our medievally transmitted manuscripts. Moreover, the surviving papyri, the next-oldest witnesses we possess (chapter 3), as well as the ancient scholia (chapter 4), confirm that the kind of variation presented by fourth-century quotations remained a feature of Homeric texts for several centuries after Plato. When combined, the picture they present of the earliest written forms of the Iliad and the Odyssey is one of fluctuation in the number of verses and considerable variation within lines. There is no justification for insisting that Plato’s text correspond exactly with the medieval manuscripts, whose relative uniformity, as we have seen already in chapter 1, cannot be entirely explained, nor can it be traced back to a single historical source. At any rate, certainly the apparatus criticus is the appropriate place to note or comment on discrepancies such as this one. Ancient quotations, papyri, and the scholia are our only sources for variant readings that were at one time a part of the living oral tradition but dropped out. When editors not only change the reading of Plato but also fail to indicate that they have done so in the apparatus, they hide from students of Homer a window into the vibrant oral tradition that is now mostly lost to us. [70]