Chapter 5

Multiform Epic and Aristarchus’ Quest for the Real Homer.

{|107} Multiformity, as conveyed by poludeukḗs ‘patterning in many different ways’, the variant epithet describing the sound of the nightingale in Odyssey xix (521), is a key concept in understanding poetry as performance in ancient Greece. This has been the general argument so far, which will now be applied specifically, in the second part of this work, to the heritage of Homeric poetry as performance. The task at hand is to work out a historical perspective of Homeric poetry as it changes over time. [1]
In confronting the dynamics of change in any songmaking tradition, it is useful to apply criteria of fluidity and rigidity. We will find, it must be emphasized at the outset, that some traditions of songmaking are at any given stage relatively more fluid, while others are more rigid; we may even allow for varying degrees of fluidity and rigidity within any one given tradition. [2] In Athenian tragedy, for example, we may expect more phraseological fluidity in some of its aspects, such as the iambic trimeters declaimed by actors, than in others, such as the songs sung and danced by the chorus. [3] {107|108}
For this important distinction of fluidity and rigidity, there are striking illustrations to be found in other cultures, as in the case of Hungarian dance traditions: it has been observed that “at a certain threshold, when the collective knowledge of the tradition within the community has reached the point where it will no longer support improvisation, the dance form may become solidified into a set sequence of figures and thus preserved—in a rigid, ‘canonical’ version—for a considerable time after the disappearance of improvisatory dancing.” [4] This formulation is ideal for my purposes, provided we understand “improvisatory” in a strictly ethnographic sense as reworked on the spot, in performance. [5] In another description of other dance traditions, one anthropologist observes that various structures of performance, as they become progressively more rigid, can suffer “abrupt confrontation and loss.” [6]
With reference to this observation, I had described in my earlier work even the pan-Hellenic period of Homeric transmission as relatively rigid in comparison to still earlier periods. [7] And I substituted a more esthetic metaphor for what I have just described as rigidity, resorting to the image of crystallization. [8] This image served to convey the essential idea of an overall “evolutionary model” of Homeric text-fixation, which I envisage not as a single event but as a long-term process, a general progression from more fluid to more rigid phases. [9]
The term “crystallization,” used as a metaphor to describe the evolution of a Homeric text without the aid of writing, can be applied also to the evolution of an individual singer’s repertoire. In ballad studies, for example, it has been observed that a given song tends to be more fluid when it is being learned by one singer {108|109} from another and progressively more rigid when it becomes part of the repertoire of the individual singer. [10]
The term has been applied in other fields as well. To mention a particularly striking example: Peter Marler, a biologist who pioneered in a 1981 article the systematic description of stages in the process of a bird’s learning its distinctive birdsong, was the first to use the term “crystallization” to designate the definitive stage of birdsong. [11] Also in the forefront of research in this area is Heather Williams, whose work centers on the zebra finch, a type of semidomesticated cagebird. [12] “Most important for Williams’ research, the male bird learns its song in a 60-day period from the 30th to 90th day after hatching, and thereafter, absent some intervention from the experimenter, it sticks with the same song for life.” [13] To quote Williams’ own words on the matter: “They stop learning and fix their song when they come into sexual maturity.” [14] We may note especially her use of the expression fix their song.
With these considerations of fluidity and rigidity serving as a backdrop, let us proceed to analyze the multiformity of Homeric transmission, with the immediate goal of refining the “evolutionary model” of Homeric text fixation by answering various questions of periodization that are raised by this model. The ultimate goal is to lay down the groundwork for a multitext edition of Homer. [15]
Let us begin with an outline, to be defended in the extensive discussion that follows it, of what I see as five distinct consecutive periods of Homeric transmission, “Five Ages of Homer,” as it were, with each period showing progressively less fluidity and more rigidity: [16] {109|110}
  1. a relatively most fluid period, with no written texts, extending from the early second millennium into the middle of the eighth century in the first millennium;
  2. a more formative or “pan-Hellenic” period, still with no written texts, from the middle of the eighth century to the middle of the sixth;
  3. a definitive period, centralized in Athens, with potential texts in the sense oftranscripts, [17] at any or several points from the middle of the sixth century to the later part of the fourth; this period starts with the reform of Homeric performance traditions in Athens during the régime of the Peisistratidai;
  4. a standardizing period, with texts in the sense of transcripts or even scripts, [18] from the later part of the fourth century to the middle of the second; this period starts with the reform of Homeric performance traditions in Athens during the régime of Demetrius of Phaleron, which lasted from 317 to 307 BCE;
  5. a relatively most rigid period, with texts as scripture, [19] from the middle of the second century onward; this period starts with the completion of Aristarchus’ editorial work on the Homeric texts, not long after 150 BCE or so, which is a date that also marks the general disappearance of the so-called “eccentric” papyri, to be defined later on in the discussion.
There is progressively less variation in each of the five successive periods, though the third, fourth, and fifth yield us progressively more information about variation in all five periods. In previous work, I treated extensively the periods here described as the first and the second and the third, considering in detail both the phenomenon of pan-Hellenism and the reforms that seem to have taken place later on in Athens during the sixth century BCE, the era of tyranny when the Peisistratidai ruled Athens. [20] Here I will limit myself to the third, fourth, and fifth periods of Homeric {110|111} transmission, working my way backward in time and devoting less attention than in my previous work to the pivotal role of Athens in the third period. [21] But I must still note in passing, at precisely this point, the importance I attach to (1) the Athenian reforms of Homeric performance associated with the Peisistratidai in the sixth century BCE, centering on the performance traditions at the Feast of the Panathenaia, [22] and (2) the Athenian reforms associated with Pericles in the fifth century. [23] Both sets of reforms fall within what I call the third period of Homeric transmission. [24]
In the present {111|112} chapter, then, and also in two more that follow, I will concentrate on periods 4 and 5. Following a reverse chronological order, I will examine in depth the relevance of Aristarchus to the fifth period and of Demetrius of Phaleron to the fourth.
It is important to keep in mind from the start that the point of reference in setting up a scheme of five periods of Homeric transmission is the dimension of performance, not of text . Accordingly, we need special working definitions for the otherwise purely textual terms transcript, script, and scripture, as assigned to the third, fourth, and fifth periods respectively. By transcript I mean the broadest possible category of written text: a transcript can be a record of performance, even an aid for performance, but not the equivalent of performance. [25] We must distinguish a transcript from an inscription, which can traditionally refer to itself in the archaic period as just that, an equivalent of performance. [26] As for script, I mean a narrower category, where the written text is a prerequisite for performance. [27] By scripture I mean the narrowest category of them all, where the written text need not even presuppose performance. In order to alert the reader that this term will be used metaphorically rather than literally—for reasons that become clear in Chapter 7—“scripture” will regularly be placed within quotation marks. [28]
It is also important to keep in mind the traditional wording that was used to express the idea of performance during the five periods. The word rhapsōidoí ‘rhapsodes’ designates performers of Homer in the reports about reforms of epic performance under the Peisistratidai (“Plato” Hipparchus 228b, Diogenes Laertius {112|113} 1.57), [29] and the same word is regularly used in the same sense throughout the stretch of time that has been divided here into periods 3, 4, and 5, with the most prominent examples to be found in Plato’s Ion. [30] To repeat my diachronic formulation concerning the concept of rhapsōidoí: [31] “it is simplistic and even misleading to contrast, as many have done, the ‘creative’ aoidós [‘singer’] with the ‘reduplicating’ rhapsōidós.” [32] Suffice it to add here that the currency of the term rhapsōidós can be reconstructed to extend to an era even before that of periods 3, 4, and 5 all taken together, that is, to an era corresponding to what I call here period 2. [33] Moreover, there is a parallel to be drawn between the rhapsodic transmission of Homer and that of Hesiod. [34]
In brief, then, this scheme of five periods in Homeric transmission brings into play primarily the dimension of performance, in particular the traditions of the rhapsōidoí, and, secondarily, the dimension of text as a derivative of performance, where each successive period reflects a progressively narrower concept of textuality, from transcript to script to “scripture.” It should be stressed again that the ultimate purpose in drawing up this scheme is to lay the groundwork for an eventual multitext edition of Homer, one that would be expected not only to report variant readings but also to relate them wherever possible to different periods in the history of textual transmission, such as the five {113|114} categories proposed here. An example of this kind of approach is my earlier work on variants in the textual transmission of Theognis, especially on the phenomenon of what may be called “Solonian” and “non-Solonian” doublets, which seem linked with different periods in not only the textual history but also the political history of Megara and its daughter cities. [35]
It is essential to stress from the start, moreover, that a multitext edition of Homer is clearly not what Aristarchus, who became ultimately the most influential textual critic of Homer in the ancient world, had in mind. The era of Aristarchus corresponds to what has just been described as period 5 in the history of Homeric transmission, when the text of Homer was becoming equivalent to “scripture.” By the time we reach the end of the last chapter, I hope that I will have justified my use of this term, as also the overall scheme of the third, fourth, and fifth periods in the history of Homeric transmission. Let us turn, then, to Aristarchus, as we proceed to review the historical circumstances, working backward in time.
Aristarchus of Samothrace became head of the Library of Alexandria sometime after 180 BCE, the estimated date of the death of a distinguished predecessor, Aristophanes of Byzantium; we do not know how soon it was after 180 that the accession of Aristarchus took place, but in any case he held on to the position of head of the Library through the reign of Ptolemy VI Philometor, whose death in 145 set off a chain reaction of events climaxing in the violent removal of Philometor’s son Ptolemy VII by Philometor’s brother Ptolemy VIII; that same year (145/4) Aristarchus departed from Alexandria and from Egypt altogether, along with many of his pupils and other scholars. [36] “From this secessio doctorum,” it has been said, “the first crisis ensued in the history of scholarship.” [37] Looking further backward, let us focus on another key figure, Aristophanes of Byzantium, who had been head of the Library probably starting with the death of Eratosthenes, perhaps sometime between 196 and 193 BCE, until his death, around 180 BCE. [38] These two figures, {114|115} Aristophanes and especially Aristarchus, are essential for my formulation of period 5.
There is a mass of references to Aristarchus in the Homeric scholia, especially in those of the tenth-century Venetus A manuscript of the Iliad. [39] From these references, derived in large part from the reports of Didymus (second half of the first century BCE and beginning of the first CE), we may infer that Aristarchus was the author of a hupómnēma ‘commentary’ on Aristophanes’ ékdosis ‘edition’ of Homer, also known as diórthōsis, and that he later went on to produce his own edition, also writing a commentary to accompany it, which in turn was followed by a revised edition made by members of his school. [40] These ekdóseis or diorthṓseis ‘editions’ and hupomnḗmata ‘commentaries’, it has been argued, were all still available to Didymus. [41] We know many details about Aristarchus’ editorial methodology from the Homeric scholia, especially from the scholia of Venetus A, and we even know from the reports of Didymus (also of Aristonicus, Nicanor, and Herodian), as recorded or paraphrased primarily in the A scholia, about the texts used by Aristarchus and about his editorial judgments concerning their relative worth. [42]
We must immediately confront the problems raised by the concept of edition here. The terms ékdosis and diórthōsis, commonly used in the scholia and elsewhere, may be interpreted to mean ‘edition’ only within limits, even in the case of critics like Aristarchus. [43] To the extent that he produced his own texts of {115|116} Homer, which were to be used as a point of reference in the Library of Alexandria, the interpretation of ékdosis or diórthōsis as ‘edition’ seems to fit. [44] Still, the usage of these same terms in the context of references to the work of other critics may fall short of what we would mean by an edition. [45]
With these qualifications in mind, let us consider the editorial criteria of Aristarchus, as transmitted mainly through the writings of Didymus. It appears that Aristarchus deemed as khariésterai ‘more elegant’ and khariéstatai ‘most elegant’ the texts of Homer that were “edited” by previous scholars, as also the undated texts known as the politikaí or ‘city editions’ stemming from Chios, Argos, Cyprus, Sinope, Massalia, and so on. [46] Conversely, he deemed as eikaîa or eikaiótera ‘random’ and phaûla or phaulótera {116|117} ‘inferior’ the texts of Homer that were not so “edited.” [47] Included under this heading of “worse” were the text or texts called koinḗ in the singular and koinaí in the plural, which Richard Janko and others interpret as the ‘common’ or ‘popular’ texts, as if they were merely a default category; the same goes for the term dēmṓdeis ‘popular’. [48] This version of Homer is the so-called “Vulgate.”
In what follows, the argument is that the koinḗ, which I prefer to call not the “Vulgate” but simply the Koine, can be traced back to a distinct category, even if it did indeed in the course of time become merged into a default category along with any other “unedited” and therefore “inferior” manuscripts. [49] In positing an earlier distinct status for the Koine, I will be in partial disagreement with Janko, but for now it is more important to stress my agreement with his argument that Aristarchus was unjustified if he deemed inferior those variants that happen to be recorded in the koinaí. [50] I also agree with Janko’s argument that the readings of the koinaí often “preserve oddities” that the other manuscript traditions level out—oddities “which are now explained from comparative philology or oral composition.” [51] But again I disagree, at least in part, with his inference that these “oddities” prove that “the ‘common’ texts are usually superior.” [52] The empirical methods of comparative philology and the study of oral {117|118} tradition can be used to defend a variant reading as traditional, not as superior. On the basis of comparative studies of textual variation in manuscript traditions that are based on oral traditions, these same empirical methods can be used to defend variant readings that happen to be attested only in manuscripts judged inferior by editors ancient or modern. [53]
The problem is that any given variant reading attested in texts that Aristarchus or his followers deemed “superior” and editors like Janko now deem “inferior” may be suspected of being an ancient editor’s conjecture rather than a genuine variant derived from oral traditions. The word conjecture is used here in an extreme sense, to indicate a hypothetical situation where an editor rejects all variant readings that he finds in the manuscripts and substitutes a reading of his own invention—his own rewriting—into the master copy.
If indeed it were simply a matter of conjectures, Janko would be justified in treating the Homer texts “edited” by ancient scholars as less valuable for his purposes than the “unedited” manuscripts, which I prefer to describe as the less edited manuscripts—less edited, that is, from the standpoint of the Alexandrian critics. But I will now argue that even the more “corrected” texts of Homer, including whatever traces there may be of the editions of Alexandrian Homer critics earlier than Aristarchus, can provide genuine variants stemming from oral traditions. The argument will extend to a still earlier Homer critic, Aristotle. [54]
The problem can be restated this way: if we read a report about an ancient critic who makes a diórthōsis or ‘correction’ in a manuscript of Homer, especially in contexts where the report itself questions the judgment of the given critic, are we to assume that this diórthōsis can only be a conjecture? [55] For an example of such a report, let us consider an anecdote in Plutarch’s Alcibiades (7), the {118|119} setting of which is to be dated 435 BCE, that tells of an Athenian teacher who claims to own a copy of Homer that he himself had corrected; the verb here is diorthoûn ‘correct’ (ἑτέρου δὲ φήσαντος ἔχειν Ὅμηρον ὑφ᾿ ἑαυτοῦ διωρθωμένον). [56] Alcibiades follows up with a play on words: he remarks mockingly that the teacher is ‘correcting’ Homer instead of students. We may note in this context another anecdote, this one reported by Diogenes Laertius (9.113) about Timon of Phleious (ca. 320–230 BCE), who advised Aratus to read tà arkhaîa antígrapha ‘the ancient copies’ of Homer, not those that were ‘already corrected’, ḗdē diōrthōména (ἤδη διωρθωμένα); again the verb is diorthoûn ‘correct’. Rudolf Pfeiffer has suggested that Timon was here “alluding no doubt to the editorial work of Zenodotus.” [57] Rudolf Blum goes further, associating such an example of diórthōsis with the school of Aristotle himself. [58]
Let us start with Zenodotus and then work our way back to Aristotle. The attitude of Timon as it is reported here is indeed clearly antithetical to that of Zenodotus—and even more so to that of Aristarchus. For Aristarchus, copies that were ‘already corrected’ would have been ‘superior’, as we have seen from the testimony of the Homer scholia. We may note again that Pfeiffer immediately associates the very word diorthoûn ‘correct’ with the “editorial work” of Zenodotus. [59] He has good reasons, as we will now see from a brief review of relevant facts suggesting that Zenodotus was a methodological forerunner of Aristarchus.
Though there are controversies surrounding the precise date, Zenodotus of Ephesus was put in charge of the Library of Alexandria about a hundred years before Aristarchus became head of the Library. [60] Zenodotus is described in the Suda as the first {119|120} diorthōtḗs ‘corrector’—let us at least for the moment continue to render this word as ‘editor’—of the Homeric poems. [61] In the Prolegomena to Tzetzes’ excerpts from Scholia on Aristophanes and Dionysius Thrax, there is a catalogue of the earliest Alexandrian critics who diōrthṓsanto ‘edited’ various ancient scrolls, and it is said that ‘Zenodotus at first and later Aristarchus’ should be given credit for ‘editing’ what are vaguely called the ‘poetic’ scrolls (τὰς δὲ ποιητικὰς Ζηνόδοτος πρῶτον καὶ ὕστερον Ἀρίσταρχος διωρθώσαντο). [62] The editorial work of Zenodotus included, besides Homer, Hesiod and Pindar. [63] Pfeiffer summarizes his own interpretation of the data: Zenodotus “was indeed the first diorthōtḗs of the Homeric and other poems, revising and emending the text, and [the word diorthoûn] was the proper technical term.” [64] The reference to Aristarchus, he goes on to say, “proves conclusively that this is what the Prolegomena mean.” [65]
Such a parallel assessment of Zenodotus and Aristarchus brings us back to the problem of determining to what extent we may interpret a diórthōsis of Homer by any critic as an ‘edition’. Even in the case of Aristarchus, it should be stressed again, there is only a limited conceptual equivalence of diorthoûn ‘correct’ with contemporary notions of edit a text. [66] There are also other related problems. Is the critic who is responsible for a given diórthōsis reliable? Or at least are his methods of diórthōsis to be trusted? Further, are the theoretical underpinnings of his methods sound? {120|121} Even if the answers are positive in the case of, say, Aristarchus, can we extend such a positive assessment further back to Zenodotus? The last question can be taken even further back, to Aristotle.
It has been shown that Aristotle was very much engaged in research on problems of Homeric textual transmission. [67] Moreover, there is considerable evidence linking him with the very concept of diórthōsis. Aristotle consistently uses the verb diorthoûn ‘correct’ and its abstract derivative diórthōsis in the sense of provide the right interpretation of a difficult text, or provide the right solution to a question (erṓtēma); as in Sophistici Elenchi (chapters 18 and 19). [68] In the same work, we even find the expression τὸν Ὅμηρον ἔνιοι διορθοῦνται ‘some people correct [diorthoûn] Homer’ (Sophistici Elenchi 166b3), with reference to an exegetical problem in Iliad XXIII (328): some critics interpret the ΟΥ in the received text as οὗ instead of οὐ (cf. also Aristotle Poetics 1461a21). [69] It is evident that such “corrections” were envisioned as markings in the margins or diacritics in the text. [70] For the sophistic tradition in general and for Aristotle in particular, the format of diórthōsis was distinctly a matter of marginalia. [71] A case in point is Aristotle’s mention of the parásēmon ‘marginal mark’ in Sophistici Elenchi (177b), in a context of discussing a variant accentuation (and breathing). [72] We may note as well the wording used to describe a copy of Homer owned by Alexander the Great: φέρεται γοῦν τις διόρθωσις τῆς Ὁμήρου ποιήσεως, ἡ ἐκ τοῦ νάρθηκος λεγομένη, τοῦ Ἀλεξάνδρου μετὰ τῶν περὶ Καλλισθένη καὶ Ἀνάξαρχον ἐπελθόντος καὶ σημειωσαμένου τινά ‘it is reported that there was a diórthōsis of the poetry of Homer, called “the one from the nárthēx,” when Alexander with Callisthenes and Anaxarchus went over it and {121|122} made some marks [sēmeîa] on it’ (Strabo 13.1.27 C594). [73] Elsewhere, we hear that this nárthēx edition of Homer resulted from a diórthōsis of Aristotle himself, Ἀριστοτέλους διορθώσαντος (Plutarch Life of Alexander 8.2). [74]
The editorial format or diórthōsis of Zenodotus and of those like Aristarchus who came after him was likewise a matter of marginalia. There is in fact a noticeable continuity from Aristotle to Aristarchus not only in the format of diórthōsis but also in the discourse associated with it, as evidenced most clearly in the case of Aristarchus himself, whose criteria at times bear a striking resemblance to those represented by the school of Aristotle. Let us take for example Aristarchus’ criterion of khariésterai ‘more elegant’ or khariéstatai ‘most elegant’, which as we have seen was applied with reference to the “edited” texts of Homer. I have found the same critical term in an earlier fourth-century context. The speaker is Isocrates, and the work in question is the last oration that he composed, the Panathenaicus or ‘Panathenaic’ speech, issued in 339 BCE (Oration 12), when the author was ninety-seven years old. [75] As the author says, an illness had prevented him from finishing the work earlier (Panathenaicus 267–270), and it seems that he had originally intended to issue the work on the occasion of the Great Panathenaia of 342 BCE (Panathenaicus 7). [76] We are about to see Isocrates referring negatively to some so-called ‘sophists in the Lyceum’. [77] These sophists are said to practice the art of the rhapsodes (1) as they perform the poems of Homer, Hesiod, or others and (2) as they are mnēmoneúontes {122|123} ‘mentioning’, in a supposedly derivative way, the khariéstata or ‘most elegant things’ about these poems:
Ἕως μὲν οὖν τοὺς λόγους ἡμῶν ἐλυμαίνοντο, παραναγιγνώσκοντες ὡς δυνατὸν κάκιστα τοῖς αὑτῶν καὶ διαιροῦντες οὐκ ὀρθῶς καὶ κατακνίζοντες καὶ πάντα τρόπον διαφθείροντες, οὐδὲν ἐφρόντιζον τῶν ἀπαγγελλομένων, ἀλλὰ ῥᾳθύμως εἶχον· μικρὸν δὲ πρὸ τῶν Παναθηναίων τῶν μεγάλων ἠχθέσθην δι᾿ αὐτούς. ἀπαντήσαντες γάρ τινές μοι τῶν ἐπιτηδείων ἔλεγον ὡς ἐν τῷ Λυκείῳ συγκαθεζόμενοι τρεῖς ἢ τέτταρες τῶν ἀγελαίων σοφιστῶν καὶ πάντα φασκόντων εἰδέναι καὶ ταχέως πανταχοῦ γιγνομένων διαλέγοιντο περί τε τῶν ἄλλων ποιητῶν καὶ τῆς Ἡσιόδου καὶ τῆς Ὁμήρου ποιήσεως, οὐδὲν μὲν παρ᾿ αὑτῶν λέγοντες, τὰ δ᾿ ἐκείνων ῥαψῳδοῦντες καὶ τῶν πρότερον ἄλλοις τισὶν εἰρημένων τὰ χαριέστατα μνημονεύοντες· ἀποδεξαμένων δὲ τῶν περιεστώτων τὴν διατριβὴν αὐτῶν ἕνα τὸν τολμηρότατον ἐπιχειρῆσαί με διαβάλλειν, λέγονθ᾿ ὡς ...
Isocrates Panathenaicus (Oration 12) 17–19
Anyway, so long as they [= Isocrates’ detractors] were abusing my discourses [logoi] only to the extent that they were reading them side-by-side [paranagignōskein] with their own, doing so in the worst possible way by making divisions of wording incorrectly and by mangling and corrupting the text in every which way, I [= Isocrates] was not yet worried about the things that were being reported to me about them. But then, a short time before the Great Panathenaia, I got very annoyed at them [= Isocrates’ detractors]. For, according to what was reported to me by some friends that I happened to meet, there were these run-of-the-mill sophists, sitting together in the Lyceum, three or four of them, the kind who tell you that they know everything, the kind who quickly turn up at every occasion, and here they were discussing various poets, and especially the poetry of Hesiod and Homer, saying on their own part nothing about them but rather performing rhapsodically [rhapsōideîn] their {123|124} poems [that is, the poems of Homer, Hesiod, and other poets] and mentioning [mnēmoneúein] the most elegant things [khariéstata] taken from what has previously been said [about the poems] by others. Then, when the bystanders showed their approval of their [= the sophists’] performance [diatribḗ], the most audacious one of them [= the sophists] started trying to slander me, saying that ...
περὶ μὲν οὖν τῶν πεπαιδευμένων τυγχάνω ταῦτα γιγνώσκων. περὶ δὲ τῆς Ὁμήρου καὶ τῆς Ἡσιόδου καὶ τῆς τῶν ἄλλων ποιήσεως ἐπιθυμῶ μὲν εἰπεῖν, οἶμαι γὰρ ἂν παῦσαι τοὺς ἐν τῷ Λυκείῳ ῥαψῳδοῦντας τἀκείνων καὶ ληροῦντας περὶ αὐτῶν, αἰσθάνομαι δ᾿ ἐμαυτὸν ἔξω φερόμενον τῆς συμμετρίας τῆς συντεταγμένης τοῖς προοιμίοις.
Isocrates Panathenaicus (Oration 12) 33
Such, then, are my opinions about educated men. As for the poetry of Homer and Hesiod and the others, I [= Isocrates] do have the desire to speak about it, since I think I could silence those who rhapsodically perform [rhapsōideîn] their poetry [= the poetry of Homer and Hesiod] in the Lyceum and speak idly about them, but I sense that I am being carried along beyond the proportion set for the introductory remarks.
We may note with interest the criterion of khariéstata ‘the most elegant things’ that these men are ‘mentioning’ about poets like Homer, Hesiod, or others. This criterion, attributed to these ‘sophists in the Lyceum’ who perform just like rhapsodes and who allegedly offer no critical judgment of their own about such poets, resorting instead to ‘what has previously been said by others’, seems to me a precursor of the Aristarchean criteria which privilege those Homer editions that are supposedly khariésterai ‘more elegant’ or khariéstatai ‘most elegant’—and which prefer the variant reading that is supposedly khariestátē ‘most elegant’. [78]
We may note also the idea that these ‘sophists in the Lyceum’ are mnēmoneúontes ‘mentioning’ received knowledge about these poems, which they are able to perform just like rhapsodes. The question arises: are they not only performing but also ‘commenting’ or ‘making commentaries’ on these poems by virtue of ‘mentioning’ received knowledge about them? [79] We may compare the claim of Socrates, in Plato’s Ion, that a rhapsode is expected to be a hermēneús ‘interpreter’ of a poet like Homer, and that therefore he must surely know the poet’s intention, or diánoia (531c). In other words, the rhapsode is expected to make a commentary on the poet he performs. [80] To which Ion replies that he {124|125} can indeed ‘speakmost beautifully about Homer, more so than any of his predecessors (καὶ οἶμαι κάλλιστα ἀνθρώπων λέγειν περὶ Ὁμήρου 530c; cf. 533c-d), and that the diánoiai that he ‘speaksabout Homer are more beautiful than those spoken by any of his predecessors (ὡς οὔτε Μητρόδωρος ὁ Λαμψακηνὸς οὔτε Στησίμβροτος ὁ Θάσιος οὔτε Γλαύκων οὔτε ἄλλος οὐδεὶς τῶν πώποτε γενομένων ἔσχεν εἰπεῖν οὕτω πολλὰς καὶ καλὰς διανοίας περὶ Ὁμήρου ὅσας ἐγώ 530c-d). [81]
Although the format of diórthōsis was a matter of marginalia for both Aristotle and Aristarchus, we can expect to find important changes or improvements in the system developed by the experts at the Library of Alexandria. One such change is the way in which Aristarchus deals with accentual idiosyncrasies in Homer. For Aristotle, as we have seen, questions of Homeric accent were a matter of diórthōsis. For Aristarchus, by contrast, such questions were to be taken up not in the diórthōsis—which is by now to be interpreted in a more strict sense, closer to our own notions of ‘edition’—but in the hupomnḗmata or ‘commentaries’. Accents were not part of the text, as Aristarchus understood the concept of text. The assumption of Karl Lehrs and others [82] that the Homer edition of Aristarchus entailed the systematic placement of accent-signs over the words of Homeric verses has been challenged—successfully, I think—by Bernhard Laum. [83] As Laum has shown, information provided by Aristarchus about accentual variations in Homer was primarily recorded in his hupomnḗmata ‘commentaries’; and, however valuable it must have been, this information did not get systematically transferred into the texts of Homer as edited by the experts at the Library of Alexandria. [84] Only in the post-Aristarchean era did some of this information in his hupomnḗmata ‘commentaries’ make its {125|126} way from there into the marginalia of Homer texts. [85]
The situation was not radically different in the case of texts from outside the Library, as evidenced by the papyri from Hellenized Egypt: even these texts, some of which we may expect to have served as “scripts”—for learning situations or even for performances before audiences—are very seldom marked systematically for accent. [86] Still, though the accentual notations that we find in the texts from Hellenized Egypt are for the most part only sporadic, they provide a wealth of information about the history of ancient Greek accentuation. [87]
For the experts in the Library of Alexandria, questions of accent were primarily though not exclusively a matter of exegesis: since accent could be the only element distinguishing one word from another, it would be crucial to know the right accent in order to distinguish one meaning from another. For non-expert readers of papyri, by contrast, there were more practical considerations: questions of accent were primarily a matter of getting the pronunciation {126|127} right. [88] In the papyri, especially in those texts that served as scripts or quasi-scripts for performance [89] or just for teaching, [90] we see that accent tends to be marked mostly where it differed from everyday pronunciation: the maximum accentual difference—and therefore the maximum accentual marking—is to be found in the poetic texts and the minimum, in the prosaic. [91]
Despite the difficulties we encounter in trying to recover the information that Aristarchus and his predecessors had collected about Homeric accentual idiosyncrasies, this information is vital for purposes of arguing (1) that there was a continuum from Aristotle to Aristarchus in the procedures of editing and commenting on the Homeric text and (2) that there was a continuum in Homeric performance traditions that are indirectly reflected by these procedures.
The Homeric accentual idiosyncrasies reported by the experts at the Library of Alexandria are an ideal test-case, in that even Aristarchus, as we have seen, treated accents as if they were not at all part of the Homeric textual transmission, thus giving us reason to think that accents were instead part of the Homeric performance tradition inherited by rhapsodes. Moreover, we have noted a scholarly interest, as early as the fourth century, in the idiosyncrasies of Homeric accentuation and even in the actual performances of rhapsodes. In the first case, we have seen that Aristotle himself spoke of Homeric accentual questions in terms of diórthōsis. In the second case, we have seen Isocrates’ disparaging picture of ‘sophists’ acting like rhapsodes by performing Homeric, Hesiodic, or other such poems and by delivering learned commentaries about them. [92] {127|128}
For an example of the accentual idiosyncrasies in Homer, let us begin with the accentuation of ἀγυιῇ in Odyssey xv (441). The pattern that we might expect on the basis of Classical Greek is ἀγυίῃ, which in this case is also attested as a textual variant in the same verse. In a 1914 article, Jacob Wackernagel proposed that such sporadically attested prosodic anomalies as ἀγυιῇ reveal authentic traditional patterns. [93] Here the evidence of linguistics, as adduced by Wackernagel, is decisive. On the basis of comparative Indo-European linguistics, we can be sure that the anomalous final-syllable accentuation of ἀγυιῇ is an archaism and that the recessive accentuation of ἀγυίῃ is an innovation. [94] Wackernagel proposed further that the authenticity of prosodic anomalies like ἀγυιῇ is supported by the very fact that they caused problems for the ancient critics of Homer and even led to false analogies in the diction of later poets. [95]
The thesis, then, as formulated by Wackernagel and as more recently reformulated by myself and others, is that such transmitted accentual patterns were reported by Alexandrian critics not on the basis of grammatical conjecture but on the basis of the actual pronunciation perpetuated by rhapsodes in their performances of the Homeric poems. [96]
To this thesis I now add two further theses (the wording of both theses has been modified in this new edition of Poetry as Performance):
In considering these two theses, I start with a most striking example: it is the accentuation of δηιοτῆτι in Iliad III (20). The scholia attribute this {128|129} anomalous accentuation explicitly to the authority of Aristarchus. As Wackernagel points out, the accent of δηιοτής is anomalous when we compare the everyday Greek words κακότης, νεότης, φιλότης. [97] And yet, the accent of δηιοτής can be verified as an archaism in terms of Indo-European linguistics, on the basis of cognate formations, especially in Vedic Sanskrit. [98] What is striking in this example, as well as in many others, is that the witness for this anomalous form is specifically named in the scholiastic tradition as Aristarchus. There is an irony here, in that Aristarchus had the reputation, even in his own era and thereafter, of being the supreme Analogist, that is, of seeking to replace, in texts that he edited, anomalous forms with analogous forms. [99]
At this point, I draw attention to a work I published in 1970, where I accepted Wackernagel’s thesis of a rhapsodic performance tradition as the ultimate witness for the archaizing prosodic anomalies preserved in the transmitted Homeric text, and where I adduced the work of Karl Lehrs, who, even before Wackernagel, had mentioned the rhapsodes as a possible source for the preservation of prosodic anomalies in the Homeric text. [100] As Lehrs argues, the very fact that a later commentator like Herodian, who flourished around the second half of the second century CE, was quite knowledgeable about the prosodic patterns reported by the earlier {129|130} commentator Aristarchus but seemed to have no idea about how or where Aristarchus got his information suggests that the earlier critic relied on evidence that goes beyond the level of pure text. [101] That evidence, Lehrs inferred, may be the testimony of rhapsodes. A similar inference was made by Martin West in his 1970 Oxford Classical Dictionary article on rhapsodes, where he cites the arguments of Wackernagel. [102]
This inference, I argue, needs to be adjusted. What Aristarchus and his successors knew about prosodic anomalies was based on indirect rather than direct experiences with the rhapsodic tradition. And, as I also argue, Aristarchus had access to earlier relevant information stemming from the era of Aristotle.
If later Alexandrian critics like Aristarchus did not have direct access to rhapsodes, whether by choice or otherwise, then we have a reason to account for the almost complete absence of references in the Homer scholia to rhapsodes. [103] The one exception of which I know is a reference in the scholia bT for Iliad XXI (26) mentioning one “Hermodoros the rhapsode” (Ἑρμόδωρος ὁ ῥαψῳδὸς χεῖρας ἐναίρων ἤκουε “χειροκοπῶν,” κατεχρήσατο δέ). [104] Erbse remarks about Hermodoros: “vir aliunde ignotus.” [105] Hermodoros may have been a contemporary of Aristarchus, but it may be more likely that he is from an earlier era. [106] In any case, to judge from the bit of exegesis attributed to Hermodoros in the scholia, Aristarchus would surely have held him in low regard. [107]
A question remains: how exactly did the performance {130|131} traditions of rhapsodes preserve the archaic and ultimately anomalous accent patterns? The answer, I suggest, has to do with the inherited melodic contours of the Homeric hexameter, however reduced the component of melody may have become in hexameter as opposed to the lyric meters, with their overt melodies. [108] These reduced melodic contours, as perpetuated in the performance traditions of rhapsodes, would have aided in preserving archaisms in the pitch accentuation—archaisms that were otherwise leveled out in everyday Greek. [109] My reformulation here, which derives from my 1970 work, is built not only on the work of Wackernagel, which goes back to 1893, [110] but also on a 1951 work by Meinrad Scheller, whose own reformulation had originally led me to appreciate Wackernagel’s insights. [111] Scheller adduces a rule in traditional Greek music, to the effect that unaccented syllables did not have higher pitch than {131|132} the acute-accented syllable: with such preexisting rules, argues Scheller, embedded patterns of archaic accentuation could be preserved within a traditional melodic frame. [112] Such a frame is what I have just called the melodic contour. [113]
By now we have a variety of reasons to justify the idea that the Homer scholarship of the Alexandrian critics, especially when it comes to information about performance, was a continuation of traditions set by the school of Aristotle. A basic question remains, however: where do we find a historical point of contact between the Homeric research of Aristotle and that of the Alexandrian critics? I will argue in the course of the next two chapters that the missing link, as it were, was Demetrius of Phaleron.
Before we consider this link, however, we must follow through in confronting the more basic question, which is, how reliable is the editorial judgment of Alexandrian critics? Their reliability, as we have seen, must be tested with special reference to cases {132|133} where they report variant readings. On the matter of variant readings in accentuation, I have already concluded that the testimony of Aristarchus is indeed reliable. But we have yet to examine variants in actual wording. To test the authenticity of such variants as reported by the Alexandrian critics, we may use the criteria of comparative philology and formulaic analysis, just as Janko has done in testing the authenticity of readings taken from the Homer texts called koinaí by Aristarchus. The difference is that I will use these criteria to test only authenticity, not correctness. To repeat my previous point, the empirical methods of comparative philology and the study of oral tradition can be used only to defend a variant reading as traditional, not to establish it as the superior reading—let alone the correct reading. [114]
Let us start with the earliest of the three major Alexandrian Homer critics, Zenodotus. For purposes of the present argument, a telling example of a variant that is backed up by the authority of Zenodotus and that turns out to be justified through the application of comparative philology and through the study of the attested formulaic system of Homeric diction is the phrase ἔλπομαι εὐχόμενος ‘I hope, praying...’ in Iliad VIII (526), as opposed to εὔχομαι ἐλπόμενος ‘I pray, hoping...’, the reading that is found in the majority of manuscripts and that is supported by the authority of Aristarchus himself. [115] In his 1976 monograph on the formulaic behavior of the Homeric verb εὔχομαι, Leonard Muellner shows convincingly that in fact both manuscript variants, ἔλπομαι εὐχόμενος as well as εὔχομαι ἐλπόμενος, can be generated syntactically from parallel formulaic patterns attested elsewhere in the Homeric text even as we have it. [116] In this case, we happen to find more internal evidence as precedent for the reading given by Zenodotus, but there are clear indications that {133|134} the “Vulgate” reading—or, as I prefer to call it, the Koine reading—is “genuine epic diction” as well. [117]
To take the argumentation further, I insist that neither variant in this example has a claim to be the original reading or, to put it positively, that both variants are traditional multiforms. In a multitext format of editing Homer, we would have to take both forms into account, and then we could still pursue the question whether one variant was more suitable than another at a given time and place. An ideal example is Zenodotus’ reading οἰωνοῖσί τε δαῖτα in Iliad I (5), reported in Athenaeus (1.12f), as opposed to the reading attested in all the manuscripts, οἰωνοῖσί τε πᾶσι. [118] There is evidence that Zenodotus’ reading follows a version that was current in the Athenian performance traditions of the fifth century, the era of the three canonical tragedians: witness the expression in Aeschylus’ Suppliants (800–801): κυσὶν δ᾿ ἔπειθ᾿ ἕλωρα κἀπιχωρίοις | ὄρνισι δεῖπνον. [119] Yet there is evidence that the manuscript reading οἰωνοῖσί τε πᾶσι, apparently defended by Aristarchus against Zenodotus, was also authentic: we have not only the external testimony of the manuscripts but {134|135} also the internal testimony of formulaic expressions, found throughout the Homeric poems, with the same idea of “hyperbolic allness” that we find in the idea that the corpses of heroes were prey to “all” birds. [120]
Let us pursue further the central question, whether any given variant reported on the authority of, say, Zenodotus, or even of Aristarchus, can be treated as just that, a variant, or whether it is merely a conjecture. The credibility of all Alexandrian editors, Aristarchus included, as witnesses to genuine readings was already seriously questioned in the eighteenth century by Friedrich August Wolf. [121] In recent times, the work of Marchinus van der Valk is most prominently cited for its sustained polemics against the credibility of all major Alexandrian scholars. [122]
There have been many variations in the history of such polemics. Earlier scholars could be selective in their approaches, concentrating their attacks on the reliability of some Alexandrian scholars while defending that of others. Thus for example Ulrich von Wilamowitz-Moellendorff and Giorgio Pasquali tended to favor Zenodotus at the expense of Aristarchus, while Karl Lehrs and Arthur Ludwich championed Aristarchus, often at the expense of Zenodotus. [123] T. W. Allen relied heavily on {135|136} Aristarchus, [124] and the trustworthiness of Aristarchus was also a cornerstone in the overall account of Rudolf Pfeiffer’s History of Classical Scholarship. [125]
More recently it has been the negative assessment of van der Valk that seems to dominate a number of influential works on Homer. With reference to the Alexandrian critics and to the so-called politikaí or ‘city scrolls’ of Homer, valued as independent textual sources by both Zenodotus and Aristarchus, Geoffrey Kirk in the introduction to his Iliad commentary has this to say about the polemics of van der Valk:
Moreover the city and individual texts, when their readings are taken as a whole, seem to be very erratic and to possess no special ancient authority; indeed the ‘common’ or ‘worse’ ones often appear, by modern criteria, more reliable than the ‘ancient’ or ‘more refined’ ones! Obviously this is a large and difficult topic; most scholars from Nauck and Wilamowitz on have held that Aristarchus sometimes made conjectures and on other occasions relied on earlier texts. That seems like a reasonable view on a priori grounds, but on the whole I side with van der Valk, who in Researches [on the text and Scholia of the Iliad] II, 86 records his opinion reached after astute if sometimes arcane studies, that ‘Aristarchus’ readings are nearly always subjective and personal conjectures’, and that the cited texts, whatever their description, are comparatively recent products of Hellenistic and especially Alexandrian criticism. That applies a fortiori to Zenodotus also, whose distinctly shorter text, in particular, is clearly the result of his applying stringent and sometimes {136|137} foolish standards of τὸ πρέπον, ‘what is appropriate’ in Homer, rather than being due to any authoritative special sources which modern criticism can discern. [126]
A similar though far more moderate position is taken by Richard Janko in the introduction to his commentary on Scrolls 13–16 of the Iliad, part of the overall Iliad commentary that has been put together under the general editorship of Kirk. [127] Although I have benefited a great deal from Janko’s discussion, and although I agree with much of what he has to say, I object when he writes: “I agree with van der Valk and Kirk (vol. I, 43) that most readings where the Alexandrians lack support in the papyri and other codices are conjectures.” [128] Janko speaks of van der Valk’s “radical re-evaluation” of Alexandrian scholarship, “which Allen had prized too highly.” [129]
I disagree, arguing that van der Valk’s efforts to discredit in general the reliability of the Alexandrian scholars and in particular the value of the variant readings that they report must be systematically juxtaposed with the efforts of earlier scholars like Arthur Ludwich, and even earlier ones like Karl Lehrs, whose work persuades me that variant readings attributed by later ancient sources to editors like Aristarchus were just that, variants attested in the extant manuscripts or manuscript traditions available to these editors, and that these variants did not as a rule stem from conjectures supposedly made by these editors or by their predecessors. [130] As for Zenodotus, a recent study by Klaus Nickau concludes after a thorough analysis that, even if this critic may have made conjectures, it is impossible in any given instance to {137|138} prove it. [131] Moreover, using the textual evidence of the actual surviving poems of Apollonius of Rhodes, Callimachus, and other Hellenistic authors, Antonios Rengakos has argued convincingly that these poets “cited” Homer—other Classicists would rather say “alluded” to Homer—on the basis of various privately-owned Homeric texts that either stem from the Homer text of Zenodotus, a contemporary of Apollonius and Callimachus, or are at least closely related to it; he argues, further, that the variants contained in these pre-Aristarchean texts are exactly that, variants, not conjectures. [132]
Further support for the thesis that the Alexandrian editors of Homer preserved genuine variants comes from the work of M. J. Apthorp, who argues that the edition of Aristarchus, taken as a whole, contained practically the sum total of genuine Homeric verses. [133] According to Apthorp, and here he follows the position of George M. Bolling, the numerus versuum of Aristarchus’ Homer edition is a functional norm, reflecting the conventions of an earlier era in Homer transmission. [134] Janko in fact implicitly {138|139} agrees with Apthorp’s inference that the plus-verses, that is, those verses that were evidently not included in Aristarchus’ edition, are “interpolations,” to be excised from modern editions of Homer. [135]
Despite my disagreement, for reasons to be elaborated presently, with Apthorp’s argument that only the verses contained in Aristarchus’ edition are “genuine,” I accept his judgment that Aristarchus’ numerus versuum is indeed genuine to the extent that it cannot be simply the cumulative result of conjectural selections. [136] I even accept his idea that the plus-verses are “interpolations” in the medieval manuscript tradition—though they would be so only retrospectively, from the hindsight of an Aristarchean editorial tradition. [137] But I find it difficult if not impossible to reconcile Janko’s {139|140} acceptance of Apthorp’s privileging of Aristarchus’ edition when it comes to variations in the number of verses—Bolling’s numerus versuum—with his simultaneous acceptance of van der Valk’s discrediting of this same edition when it comes to variations in the actual wording of verses. [138]
I propose that variations, both in wording and in numerus versuum, be treated as parallel phenomena as we reconsider the editorial task that confronted the Alexandrian Homer critics. With the establishment of any final text of Homer by Aristarchus, however faithfully this critic may have collected all the facts that he knew from all the available manuscript evidence, we would have to expect that he was left with a mass of variants, on the level of wording within lines, that would have to be omitted in the text established by him. If such a text becomes a definitive edition, from then on the re-entry of any of these variants may indeed be considered an interpolation retrospectively, from the standpoint of the hypothetical edition. Let us apply the same reasoning to the problem of numerus versuum. In this connection, we cannot lose sight of the mechanism of fluctuating expansion and compression in oral poetics, a phenomenon discussed at length elsewhere. [139] This phenomenon produces fluctuation between more and fewer lines, and the very fact of this fluctuation must be, at least in some cases, a matter of variants. Granted, if a shorter version is accepted into a canonical edition established by Aristarchus, then a longer version could re-enter the tradition represented by that edition only as an interpolation—even if this longer version is diachronically a variant of the given shorter version.
As for Kirk’s acceptance of the position taken by van der Valk, it is far more extreme than Janko’s: for Kirk to go even further than August Nauck—not to mention Wilamowitz—in insisting, as we have seen in the passage quoted above, that the Alexandrian editors invented and imposed their own readings is to ignore the arguments that Arthur Ludwich had worked out specifically to counter the arguments of Nauck and Wilamowitz concerning the alleged “conjectures” of the Alexandrian critics. {140|141}
On the other hand, I disagree with some aspects of the further inferences drawn by Ludwich and others from the evidence that they collected—evidence meant to show that the Homer edition of Aristarchus was a reliable collection of genuine readings. Even if it is justified to infer that Aristarchus’ edition accurately reflected an official Athenian version of Homer, and we will presently review some reasons for arguing such a possibility, we cannot infer further that Aristarchus’ edition therefore reflects the “original” Homer. We have already seen one decisive argument against this further inference, which is, that there are variant readings stemming from the editions known as koinaí—and rejected by Aristarchus—that are genuine readings as well.
There are also at least two other arguments against the inference that the Aristarchus edition of Homer represents the only genuine Homer. The first comes from the evidence of the so-called “eccentric” papyri. [140] As Stephanie West describes them, these papyri are “characterized by a high proportion of variants and additions,” and they can be dated mostly before around 150 BCE; they generally “die out” after this terminus, while “later papyri offer a text which differs little from that of the medieval manuscripts.” [141] West allows for the possibility—and this is as far as she is willing to concede—that these “eccentric” papyri from the Ptolemaic period contain variants derived from the performance traditions of rhapsodes. [142] The case could be made more forcefully. [143]
West and others explain the disappearance of “eccentric” versions as due directly or indirectly to the influence of Aristarchus’ new edition of Homer, which apparently was finished also around 150 BCE. [144] This theory seems to me insufficient. To start with the obvious, I agree with Allen that such a strategy of explanation, which requires {141|142} that an Alexandrian critic’s edition caused the obsolescence of the “eccentric” versions, “is an excellent example of the argument post hoc ergo propter hoc.” [145] West herself raises a problem with this theory: that even the papyri dated after 150 BCE “offer too wide a range of variants to allow the hypothesis that they might all be copies of a single edition.” [146] Any explanation of the obsolescence must take into account a factor mentioned by Allen, who notes that these “eccentric” versions “had depended on the rhapsode,” but now they “withered of themselves” as the rhapsodic art withered. [147] In this connection, we may note with interest the argument, advanced by Aldo di Luzio, that the linguistic peculiarities of the “eccentric” Homer papyri reflect a phase of rhapsodic transmission emanating from Athens. [148]
The subject of rhapsodes leads us to yet another argument against the inference that the Aristarchus edition represents the only genuine Homer. This time, the evidence comes from the works of Plato, whose citations of verses from Homer have been systematically studied by Jules Labarbe. [149] In the era of Plato, as we can deduce from these citations, there seems to have been no single Athenian Homer text of the sort posited by Ludwich and others in their arguing for some kind of archetype as a source for Aristarchus’ Homer edition. Labarbe finds demonstrably genuine variants that are distinct from those adopted by Aristarchus, [150] and some variants are in fact attested only in Plato and nowhere in the medieval manuscript tradition of Homer. [151] There is even an instance, with reference to a verse in Iliad XXIII (335), where Plato and Xenophon each report variant readings, each differing from the other, which do not survive into the medieval manuscript tradition. [152] {142|143}
Labarbe allows for the possibility that Plato used an Athenian text of Homer for his citations. [153] Still, there is no way to equate such a text with some kind of Athenian “archetype” as reconstructed by Ludwich and others. [154] More important, the Homeric variations attested in Plato, whether or not they were mediated by way of a text, must derive ultimately from the oral tradition: the criteria that Labarbe applies to test whether or not any Homeric variant cited in Plato is “genuine” are soundly based on Milman Parry’s methodology of formulaic analysis, and to this extent Labarbe is justified in claiming that at least some of the Homeric variations attested from the citations of Plato must stem from the performance traditions of rhapsodes. [155]
And yet, even though I resist the idea that a Homer text owned by Plato can be derived from an archetypal Athenian text of Homer, there is one aspect of Plato’s “text” that does indeed suggest the existence of at least a conceptual Athenian archetype. Unlike the “eccentric” papyri that are dated from around 300 to 150 BCE, Plato’s Homer is not characterized by plus-verses. [156] Labarbe attributes this aspect of stability in Plato’s Homer text to the hypothetical existence of a “control text,” which he equates tentatively with the so-called “Peisistratean Recension,” the popular influence of which would have regulated the numerus versuum even of commercial copies, such as the one that Plato presumably possessed. [157] In previous work, I have already given reasons to doubt the construct of such an early “control text,” arguing instead that a crystallizing Athenian performance tradition of Homer at the Panathenaia could be sufficient in and of itself to account for the quasi-textualization of Homeric poetry in the era of the Peisistratidai. [158] Patterns of stabilization in length of performance need not presuppose the agency of a written text. [159] {143|144}
More important for now, the stabilization of the numerus versuum in papyri after 150 BCE can hardly be due to the influence of a new Aristarchean edition—if indeed it is true that the Homer citations of Plato already reveal a similar pattern of stabilization, one that we find in place as early as the fourth century. [160] In the discussion that follows, extending into the next two chapters, I prefer to argue that any stabilization of the Homeric numerus versuum in the fourth century is due to the regulation, by the Athenian State, of rhapsodic performance traditions, and that the “eccentric” papyri dating from around 300 to 150 BCE, with their plus-verses, reveal a later and relatively more fluid phase of rhapsodic tradition when such regulation by the State was no longer in effect. [161]
That the “eccentric” papyri become obsolete in a still later phase, after around 150 BCE, raises the possibility that we are witnessing the beginnings of some new kind of interference by the State, in this case the state of Ptolemaic Egypt. [162] Such a possibility, however, is not necessarily incompatible with Allen’s theory that the performance tradition of the rhapsodes had by this point “withered.” Granted, Allen’s metaphor surely overstates the case in that we cannot say that the performance traditions of Homer simply died out around 150 BCE. As we will see, {144|145} they persisted for several centuries beyond that point, even in areas like Hellenized Egypt. But at least the higher levels of flexibility in the rhapsodic tradition as reflected in the “eccentric” papyri do indeed seem to have ceased after 150 BCE or so, and this cessation coincides with the emergence of the Aristarchean text of Homer as the only version of Homer.
The ultimate outcome, with the atrophy of the “eccentric” papyri after 150 BCE or so, is that both the later performance traditions and the later commercial “books” or scrolls of Homer revert to reflecting more closely an earlier and more canonical Athenian rhapsodic tradition that pre-dates the era of the “eccentric” papyri. There will be more to say about such a rhapsodic tradition in the next two chapters. Further, these patterns of reversion in the popular Homer do indeed correspond to the canonical new edition that had just been prepared by Aristarchus—an edition that reflects most closely a canonical Athenian tradition. Thus it may be more apt to find a metaphor other than withering away, such as sclerosis, in describing the fate of the performance tradition of Homer after 150 BCE. We may prefer, however, a metaphor that leaves room for the esthetic possibilities of the envisaged process, and thus I revert to the image with which we started this chapter, that of crystallization.
Let us return to the important implications of what Labarbe had discovered in his study of Plato’s Homer. We have seen that at least some of the variations that are attested in this instance of Homeric transmission reveal clearly the performance traditions of rhapsodes. “If Labarbe and other modern critics are right,” as van der Valk comments, then “the Homeric text was originally transmitted orally.” [163] “It is obvious,” he continues, “that the acceptance of this theory has far-reaching consequences.” [164] In fact, “the whole basis of our Homeric text becomes uncertain.” [165] To put it another way: if Labarbe is right, we can no longer determine the archetype.
Van der Valk’s reaction is to posit a purely textual rather than oral transmission, going back all the way to a time when “Homer {145|146} put down his poems in writing.” [166] Whereas I prefer Labarbe’s findings about a rhapsodic phase of transmission and resist van der Valk’s radical alternative, Labarbe makes further inferences about his findings that I cannot share. According to Labarbe, we simply do not know exactly what the real Homer said, but we do know that the rhapsodes could change it in the context of a continuing oral tradition. [167] So this critic too, like those who believe in a purely textual transmission, posits an archetype, albeit an unwritten one, setting up yet again the choice between right variants that supposedly come from this archetype and wrong ones that come from elsewhere. According to Labarbe’s model, there are inferior variants stemming from the rhapsodes and superior ones, from Homer, so that only the second category is allowed to be “authentic.” [168] And yet, the criteria for establishing what is superior or inferior, right or wrong, seem to me subjective.
What is needed is a set of objective editorial criteria that take into account the phenomenon of variation in reperformance. This phenomenon is reflected not only in the textual variants that we find in Plato’s Homer but also in those reported by Alexandrian critics like Aristarchus. Again I find that most arguments about whether a given variant is spurious or genuine, inferior or superior, are unfounded. We are entitled to like or dislike any given variants that the various Alexandrian critics had chosen from time to time, and we may even classify these critics according to their methods or prejudices in choosing one kind of variation over another, but what we cannot do is simply assume that they have made a conjecture just because their choice of a reading does not suit our own sense of editorial verisimilitude. [169] {146|147} A rigorous case-by-case review of instances where van der Valk has argued that ancient critics substituted conjectures for genuine readings can lead to conclusions quite different from his. The case-by-case review of Ludwich 1884 / 1895 remains in my opinion a most valuable aid.
There has been one such review by Vittorio Citti, in which many of the variant readings stemming from the politikaí or ‘city editions’ of Homer have been defended, in my opinion successfully, from van der Valk’s arguments against their authenticity. [170] Though I disagree wherever Citti concludes that a given reading of the politikaí is “superior” to that of the surviving manuscript traditions, [171] or for that matter wherever he says that a reading is “inferior,” [172] we may note with interest that he treats some variants, like the reported reading χέει ἄσπετον of the Massaliotike at Iliad XII 281 (scholia AT) and the reading χέει ἔμπεδον of the surviving manuscripts (and of Aristarchus: scholia AT), as “both ancient” (ambedue antiche). [173] The central question is not even whether both such readings are ancient but more simply whether both are authentic. {147|148}
In general, a most convincing proof of a variant’s authenticity is its relative archaism. A particularly striking example is the reported reading δούρασιν ἄμφω of the Massaliotike at Iliad XXI 162 (scholia AT), with an archaic indeclinable ἄμφω (also attested in Homeric Hymn to Demeter 15), as opposed to the reading δούρασιν ἀμφίς of the surviving manuscripts. Still, δούρασιν ἀμφίς may be just as archaic in terms of a formulaic system that could generate both forms. [174] What turns out to be an even more striking example of proven archaism appears in a variant: at Iliad XXI 351, where the surviving manuscripts read ἠδὲ κύπειρον, the politikaí read ἠδὲ κύπαιρον (scholia AT: αἱ ἐκ τῶν πόλεων ἠδὲ κύπαιρον εἶχον). Citti thinks that kúpairos is a “Dorism” as distinct from Ionic kúpeiros, [175] while van der Valk says that kúpairos “is, in my opinion, a corruption or an instance of local orthography.” [176] I propose instead that kúpairos is an archaism, which could have entered the oral poetic tradition at a relatively early period, perhaps even as early as the second millennium BCE: in the Linear B documents, for example, we find the form ku-pa-ro 2 = kuparyos, ancestor of kúpairos. [177] It is precisely this same form kúpairos that we find attested in the “city editions.” Even in this case, however, I would argue that kúpeiros too is an authentic variant.
It is instructive in the context of this discussion to assess van der Valk’s opinion about a pair of variant readings that we have already considered, the case of the phrasing ἔλπομαι εὐχόμενος ‘I hope, praying...’ at Iliad VIII 526, which is the reported reading of Zenodotus and is found in a minority of manuscripts, as opposed to εὔχομαι ἐλπόμενος ‘I pray, hoping...’, which is the reported reading of Aristarchus and is found in a majority of manuscripts. In this case, van der Valk concludes from the sense of the Homeric passage in question that “Zenodotus’ reading seems to be the better one.” [178] So Zenodotus is here exceptionally being rescued from the charge of conjecture. But now the blame is shifted to Aristarchus instead: according to van der Valk, if Zenodotus {148|149} did not make a conjecture in this case, then surely Aristarchus must have. [179] Whoever has the “better” reading, if we follow this line of thought, must be using the real text. Whoever has the “worse” reading, to continue in this line, must be making a conjecture. The problem with such an approach, as Muellner’s findings reveal, is that the question of a “better” or “worse” reading is moot here, once we re-examine the question from the standpoint of formula analysis: both readings can in fact be shown to be genuine.
Thus I reaffirm my position that we need to take all authenticated variants into account in establishing a multitext format for the editing of Homer. Only within such a multitext editorial framework can we turn to questions of whether one variant was more suitable than another at a given time and place. But it is important also to reaffirm that Aristarchus and his predecessors, even though they collected a wide range of variants, had in mind an editorial goal very different from the one I am advocating. They treated the textual traditions of Homer as primary evidence and the performance traditions, which as we will see were still alive in their time, as mostly irrelevant to their primary goal, which was the recovery of an original Homer. [180]
In this respect, Aristotle may have had a different outlook, if I am not mistaken in detecting in his work traces of a sustained interest in the performative aspects of Homer. A clear example is his critique, in Poetics 1462a, of the techniques of a rhapsode called Sosistratos, otherwise unknown to us, as an actor. [181] It may also be pertinent to cite the anecdote that has Plato giving Aristotle the sobriquet anagnṓstēs ‘the one who reads out loud’ (Vita Marciana, Aristotle Fragments 428.2 ed. Rose). [182] Rudolf Blum remarks about this sobriquet: “in order to understand the joke one must remember that in Antiquity people read aloud, but that well-to-do gentlemen had slaves read aloud to them.” [183] We may note that anagnṓstēs can designate, more specifically, a slave who is trained {149|150} to read out loud to copyists in the process of book-production. [184] As we will see in the next chapter, however, to say this much about the meaning of anagnṓstēs may not be enough, especially in view of a custom current in an earlier historical period, conveyed by the verb paranagignṓskein, of reading out loud to performers. [185] For now, however, it is enough to say that the sobriquet anagnṓstēs shows that Aristotle was interested in how the text of Homer should sound, as it were.
 By contrast, as we move forward in time to the era of Aristarchus and his followers, we find that they were interested almost exclusively in the textual rather than the performative dimensions of Homeric transmission. [186] To that extent they were not all that different from many contemporary investigators of the Homeric text, who assume that their task is a quest to recover the original composition despite the historical reality of multiformity in the text—a reality that reflects multiformity in performance.
The intellectual framework, then, of this quest for the real Homer was pioneered by the likes of Aristarchus. It seems clear that they believed in a real Homer, an original Homer. And to believe this much is of course nothing new. It was not only the editors of Homeric texts who posited an original: so too did myth, and myth posits an original author as well, called Homer. [187] The further back we go in time, the greater the repertoire of this Homer, including in the earlier times all the so-called Cycle, all the Theban epics, and so on; as we have seen, the very notion of “Cycle” had once served as a metaphor for all of Homer’s poetry. [188]
The further we go forward in time, by contrast, the less there is that Homer did himself. Not only is his repertoire becoming restricted to the Iliad and Odyssey: there are many parts even of {150|151} these epics that now become suspect: for example, Homer surely could not have composed the Shield of Achilles in Iliad XVIII (483–608), in the opinion of Zenodotus (scholia A for Iliad XVIII 483a). [189] And the original Homer of this more critical and suspicious age becomes all the more specific and even brittle in identity, reflecting ever more the critics’ understanding of his archetypal creation, his text. For Aristarchus, it appears that Homer was an Athenian who lived around 1000 BCE, in the time of Athenian migrations (Proclus F a 58–62 Severyns; cf. Life of Homer p. 244.13, p. 247.8 Allen; cf. scholia A for Iliad XIII 197); [190] moreover, the scholiastic tradition stemming ultimately from Aristarchus implies that Homer wrote his poems (scholia A for Iliad XVII 719) and that Hesiod actually had a chance to read them (scholia A for Iliad XII 22a). [191]
Even though Aristarchus, following the thought-patterns of myth, posited a Homeric original, he nevertheless accepted and in fact respected the reality of textual variants. He respected variants because, in terms of his own working theory, it seems that any one of them could have been the very one that Homer wrote (and Hesiod read). That is why he makes the effort of knowing the many different readings of so many manuscripts. He is in fact far more cautious in methodology than some contemporary investigators of Homer who may be more quick to say which is the right reading and which are the wrong ones. Aristarchus may strike us as naïve in reconstructing an Athenian Homer who “wrote” around 1000 BCE, but that kind of construct enables him to be more rigorous in making choices among variants. [192]
What, then, would Aristarchus have lost, and what would we stand to lose, if it really is true that the variants of Homeric textual tradition reflect for the most part the multiforms of a {151|152} performance tradition? If you accept the reality of multiforms, you forfeit the elusive certainty of finding the original composition of Homer but you gain, and I think this is an important gain, another certainty, an unexpected one but one that may turn out to be much more valuable: you recover a significant portion of the Homeric repertoire. In addition, you recover a sense of the diachrony. From the sketch of Homeric periodization that I have just offered, one can develop a sense of different Homers for different times, such as a relatively “proper Homer” for the late fourth century and thereafter, periods 4 and 5, as opposed to a “primitive Homer” in, say, periods 1 and 2, the era before the reforms of the Peisistratidai. As for period 3, we will see in the next chapter that the most appropriate description may be the “common” Homer—or let us say the Homer of the Koine.
Let us return to the fact that some Homer experts who accept Lord’s formulation of “oral” poetry seem ready at times to discount the value of Aristarchus’ editorial repertoire of variants, which go far beyond the Koine or “Vulgate” texts of Homer. There is an irony here. It would be more understandable for proponents of a “writing Homer” to reject variant X or Y, accepting Aristarchus’ implicit premise that only one variant can be right and that Homer could not have written X or Y for such-and-such reasons. It is unnecessary, however, for proponents of an “oral Homer” to insist on one and only one right version, unless they are also willing to believe that the oral tradition ground to a dead halt sometime around the second half of the eighth century BCE, after the text was supposedly dictated. [193] In earlier work, presenting arguments that challenge the idea of an early dictation, I substituted an “evolutionary model” to account for the process of Homeric text-fixation. [194] Here I have refined that model with a scheme of five consecutive periods of Homeric tradition culminating with the text of Aristarchus. Still, we are left with the clear impression that multiformity, however reduced, remains a persistent feature even in the terminal phases of this tradition. {152|153}


[ back ] 1. Cf. GM 29: “the language of a body of oral poetry like the Iliad and Odyssey does not and cannot belong to any one time, any one place: in a word, it defies synchronic analysis.” Cf. di Luzio 1969:11, where he refers to the “diachronic” nature of oral epic composition; cf. D’Ippolito 1984. In my own work, the terms synchronic / diachronic refer not to the internal standpoint of any given structure but only to the external standpoint of one who analyzes that structure (cf. PH 4). Thus it is preferable to say that the analysis of oral composition requires the simultaneous application of synchronic and diachronic perspectives, and that the absence of either of these perspectives can lead to a warped analysis. Aldo di Luzio offers a comparable formulation that appears toward the end of his work, p. 138. {He cites Pagliaro on Dante; in an earlier context, p. 13, he cites Pagliaro Saggi pp. xii ff. and Nuovi saggi pp. 381–408.He cites Pagliaro on Dante; in an earlier context, p. 13, he cites Pagliaro Saggi pp. xii ff. and Nuovi saggi pp. 381-408.}
[ back ] 2. Cf. Lord 1995 ch. 2.
[ back ] 3. Cf. PH 45–46.
[ back ] 4. Kraft 1989:278n18, who continues: “I have the impression that this has been the fate of many Hungarian dance traditions in the villages of Hungary proper.” For more on the rigid/fluid distinction, see PH 60–61.
[ back ] 5. Miller 1982b:15 and pp. 5–15 in general, with a catalogue of instances in contemporary scholarship where the term “improvise” has been oversimplified with reference to oral tradition.
[ back ] 6. Royce 1977:104.
[ back ] 7. PH 60–61.
[ back ] 8. PH 53.
[ back ] 9. See also HQ ch. 3; and Seaford 1994:144-154. Cook 1995:4 extends my evolutionary model: “the crystallization of the Odyssean tradition into a written text, the growth of Athenian civic ritual, and the process of state formation in Attica were simultaneous and mutually reinforcing developments.”
[ back ] 10. Andersen 1991, especially p. 26 (where he uses the term “crystallization”), with further bibliography. At p. 37 he says about one of his informants: “Stanley Robertson is capable of imitating the singing styles of his aunt Jeannie and his cousin Lizzie to perfection, but he will never sing like them in public.”
[ back ] 11. Marler 1986. I owe this information to Professor Heather Williams (per litteras 23 September 1993).
[ back ] 12. The results of Williams’ work are cited from the lucid account of M. R. Montgomery, Boston Globe 26 August 1993 pp. 61 and 64.
[ back ] 13. Montgomery 1993.
[ back ] 14. Montgomery 1993, quoting Williams.
[ back ] 15. On the concept of a multitext edition of Homer, see Bird 1994.
[ back ] 16. The category of “period,” used here in setting up tentative boundaries of periodization, is meant to be more precise than the category of “phase” and the subcategory of “stage” as I use those words in HQ 109–10.
[ back ] 17. The word is used in a narrow sense, to be defined below.
[ back ] 18. The word is used in a narrow sense, to be defined below.
[ back ] 19. The word is used in a narrow sense, to be defined below.
[ back ] 20. Summary in ch. 3 above.
[ back ] 21. For more about the Athenian impact on the Homeric tradition in what I call here the third period, see HQ 42–63. For a discussion of the evidence of vase paintings as a criterion for determining the gradual fixation of Homeric traditions around the middle of the sixth century, especially in Athens, see Lowenstam 1993, especially p. 216.
[ back ] 22. HQ 42–43.
[ back ] 23. HQ 75–76n37. For a reference to reforms, instituted by Pericles, of performance traditions at the Panathenaia, see Plutarch Pericles 13.11: φιλοτιμούμενος δ᾿ ὁ Περικλῆς τότε πρῶτον ἐψηφίσατο μουσικῆς ἀγῶνα τοῖς Παναθηναίοις ἄγεσθαι, καὶ διέταξεν αὐτὸς ἀθλοθέτης αἱρεθείς, καθότι χρὴ τοὺς ἀγωνιζομένους αὐλεῖν ᾄδεινκιθαρίζειν. ἐθεῶντο δὲ καὶ τότε καὶ τὸν ἄλλον χρόνον ἐν ᾨδείῳ τοὺς μουσικοὺς ἀγῶνας ‘It was then for the first time that Pericles, ambitious as he was, got a decree passed that there should be a competition [agṓn] in mousikḗ at the Panathenaia, and he set up the rules, having been elected as an athlothétēs [= organizer of the athloi ‘contests’] for those who were competing [agōnízesthai]—rules for them to follow about the aulós - playing and the singing and the kithárā-playing. At that point in time and in other periods of time as well, it was in the Odeum that people used to be spectators [theâsthai] of competitions [agônes] in mousikḗ.’
[ back ] 24. This note elaborates on the previous note. Davison 1968:63 argues that Plutarch’s reference to Pericles’ reform should not be taken to mean that the institutions he mentions actually began with the reform. Herington 1985:86 infers that the contests of kitharōidoí ‘lyre-singers’, aulōidoí ‘pipe-singers’, kitharistaí ‘lyre-players’, aulētaí ‘pipe-players’- as indicated by αὐλεῖνᾄδεινκιθαρίζειν, may be understood here as being in addition to contests of rhapsōidoí. I agree, on the basis of the comparative evidence provided by parallel wording in other passages such as Isocrates Panegyricus 159, where Homeric performances are described as taking place ἐν τοῖς μουσικοῖς ἄθλοις. We may compare also the parallel wording in IG XII ix 189, an inscription from Eretria in Euboea (ca. 340 BCE). Line 5: τιθεῖν τὴμ πόλιν ἀγῶνα μουσικῆς. Lines 10-15: τὴν δὲ μουσικὴν τιθεῖν ῥαψωιδοῖς, | αὐλωιδοῖς, κιθαρισταῖς, κιθαρωιδοῖς, παρωιδοῖς, | τοὺς δὲ τὴν μουσικὴν ἀγωνιζομένους πάντα[ς] | ἀγωνίζεσθαι προσόδιον τεῖ θυσίει ἐν τεῖ αὐλεῖ ἔ|[χο]ντας τὴν σκευὴν ἥμπερ ἐν τοῖ ἀγῶνι ἔχουρ[ι]. For more on rhapsōidoí in agonistic contexts where they are mentioned as parallel to kitharōidoí and aulōidoí, see PH 29, 54, 104 (with reference to the Eretrian inscription, IG XII ix 189). In light of the fact that Plutarch describes Pericles as taking on the role of athlothétēs [= organizer of the athloi ‘contests’] (καὶ διέταξεν αὐτὸς ἀθλοθέτης αἱρεθείς), we may compare Aristotle Constitution of the Athenians 60.1, where the athlothétai are described as arranging the Panathenaic procession and the agṓn of mousikḗ: διοικοῦσι τήν τε πομπὴν τῶν Παναθηναίων καὶ τὸν ἀγῶνα τῆς μουσικῆς. Elsewhere (HPC), I argue the following point: when Pericles is represented as saying in Thucydides 2.41.4 that ‘we’ Athenians do not need Homer as an epainétēs, an official giver of praise, it is implicit that ‘we’ Athenians already own Homer and therefore do not need to hear ‘our’ ownership made explicit by way of excessive references to Athens in the narrative of Homer. Relevant is the interpretation offered by Lardinois 1995:161 for the Homeric verse that we know as Odyssey xvi 161, οὐ γάρ πως πάντεσσι θεοὶ φαίνονται ἐναργεῖς ‘for it is not to everyone that the gods appear as manifest’. According to Lardinois, this verse may be interpreted as an oblique Homeric reference, in the context of Homeric performance at the Panathenaia, to the notional presence of Athena at the Panathenaia. According to this interpretation, the verse implies that those attending the Panathenaia are a privileged audience. Also relevant is the fact that in Odyssey iii 240 Athena makes herself enargēs ‘manifest’ to Nestor.
[ back ] 25. HQ 66–69.
[ back ] 26. HQ 66–69.
[ back ] 27. HQ 68.
[ back ] 28. In any case, my literal understanding of “scripture” is not casual: I take seriously the efforts of Smith 1993 to achieve greater semantic precision in using this word. At p. 209, he notes: “even when scripture is seen as, is understood to be, explicitly written, it is an error to suppose that this means written rather than oral.” Cf. Graham 1987:7.
[ back ] 29. Summary in ch. 3 above.
[ back ] 30. PH 22.
[ back ] 31. Cf. PH 21–28.
[ back ] 32. GM 42, based on a formulation made in N 1982. This formulation is corroborated by the article of Ford 1988.
[ back ] 33. GM 40–47. Citti 1966:8 uses the criteria of stadio di trasmissione libera and stadio di trasmissione rigida to distinguish what he describes as the era of the aoidós from that of the rhapsōidós. I agree with the wording of his criteria but not with its application to the concepts of aoidós and rhapsōidós, in that I am arguing for a broader range of applications in the case of the latter word. Also, in the extensive report on rhapsodes in the scholia to Pindar Nemean 2.1e, we may note the usage of apangéllein in referring to the performance of rhapsodes in the circle of Kynaithos (οἱ περὶ Κύναιθον ... τὴν Ὁμήρου ποίησιν ... ἐμνημόνευον καὶ ἀπήγγελλον) in light of the usage of this same word in Herodotus 7.142.1, as discussed in PH 168. On the relationship of master and disciple in the traditions of the rhapsodes (as indicated by the expression οἱ περὶ Κύναιθον), see Ritoók 1970:23–24.
[ back ] 34. PH 29n66 on the sunthútai Mousôn Hēsiodeíōn ‘fellow-sacrificers to the Hesiodic Muses’ at Thespiae, IG VII 1785. Cf. the reference to paideutaí ‘students’ belonging to a gymnasium called the Mimnermeîon (after the poet Mimnermus) at Smyrna, Ionia/Smyrna doc. 661.9 [CIG 3376]; there was a gymnasium called the Homēreîon (after Homer) at Chios, Ionia/Chios doc. 268b4 [CIG 2221]; there is also a Homēreîon at Smyrna, Ionia/Smyrna doc. 703.1, and at Delos, Inscriptions de Délos b.147.
[ back ] 35. N 1985:46–51.
[ back ] 36. Pfeiffer 1968:210–212.
[ back ] 37. Pfeiffer 1968:212.
[ back ] 38. Pfeiffer 1968:172.
[ back ] 39. Edited by Erbse 1969–1988. On the D scholia, not covered in Erbse’s edition, see Montanari 1979:3–25 (cf. Henrichs 1971, especially pp. 100–105).
[ back ] 40. The historical sequence formulated here, with an Aristophanes edition followed by an Aristarchus commentary followed by an Aristarchus edition followed by a second Aristarchus commentary followed by a second edition by the school of Aristarchus, is essentially the construct of Pfeiffer 1968:217; followed by Janko 1992:26. On the second edition, supposedly produced by Aristarchus’ students, see Apthorp 1980:132. On the notion of a hupómnēma ‘commentary’, see Lührs 1992:10, who visualizes it as a combination of what we would call an apparatus criticus and a commentarius criticus. On the early sources of the D scholia, especially as mediated in the format of the hupómnēma, see Montanari 1979:14–15.
[ back ] 41. Pfeiffer 1968:217–218.
[ back ] 42. Janko 1992:26. Also Ritoók 1987:15, who summarizes the Alexandrian editorial criteria of preferring some readings over others as follows: relative age of manuscript, majority of manuscripts showing a given reading, quality of given manuscript, internal evidence.
[ back ] 43. Allen 1924:307 says that the editions of the Alexandrians “were not editions in the modern sense, that is so many hundred copies (ἴσα) produced by scribes from a single original. The words ékdosis and diórthōsis were often verbals and meant ‘proposal for edition’, and ‘revision’.” [ back ] From the usage of the Homeric scholia, I infer that ékdosis means the production of a new copy containing readings based on the procedure of diórthōsis. This procedure entails, if we follow Pfeiffer p. 94, the collating of manuscripts and the emending of texts—which would amount to a “recension.” We should expect, of course, the quality of the procedure to vary from editor to editor (cf. Cameron 1990:117 on the editorial criteria of Eutocius, early sixth century CE). I prefer the formulation of Blum 1991:65n10, who says that ékdosis and diórthōsis are practically the same: “there was no ékdosis without diórthōsis, there were only different degrees of diórthōsis.” Allen goes on to say (p. 308): “If, as has been remarked, Aristarchus’ ‘edition’ had been multiplied and put upon the market, his successors Ammonius and Dionysius Thrax could never have quarrelled about his readings.”
[ back ] 44. The work of Apthorp 1980, reinforcing the arguments of Pfeiffer 1968:215–217 against Erbse 1959, argues persuasively that Aristarchus did indeed produce his own texts of Homer. Cf. Lührs 1992:6-13. For reasons about to be discussed, however, I am not persuaded by Apthorp’s arguments that the texts of Aristarchus recovered practically all that was genuine in the Homeric tradition, any more than I am persuaded by the arguments of others, also about to be discussed, seeking to prove that Aristarchus’ base text was riddled with spurious conjectures. Further, we will see that there are important insights to be gained from the position taken by Erbse and defended by Nickau 1977:18–19 against Pfeiffer, especially with reference to the fluidity, however reduced, of textual transmission even after the editions of Aristarchus.
[ back ] 45. On the vagueness of the term ékdosis as used in texts other than the scholia and with reference to the work of other scholars, for example Apollonius Dyscolus, see also Nickau 1977:18–19n39, citing a concession on this point by Pfeiffer 1968:216.
[ back ] 46. Allen 1924:283–296, 297–299; cf. Janko 1992:22 and Apthorp 1980:47–48. As Apthorp (p. 102n2) points out, feminine adjectives like khariésterai presuppose nouns like ekdóseis ‘editions’ (in however limited a sense). I find one clear instance, scholia A for Iliad III 10, where the Chios and the Massalia texts are referred to as ekdóseis ‘editions’. For reasons that will become clear later, it is important to note the usage of the superlative khariéstatai ‘most elegant’, as in the scholia for Iliad II 53a (A), II 164a (A), II 192b (A), II 196c (T), III 18a (A), III 51 (A). Moreover, in the scholia for Odyssey x 70 (hypothesis format), a reading adopted by Zenodotus is followed by the following comment: καὶ ἔστι χαριεστάτη ἡ γραφή ‘and it is this way of writing it that is most elegant [khariestátē]’.
[ back ] 47. Allen 1924:277–278; cf. Janko 1992:26 and Apthorp 1980:47–48. As Apthorp (p. 102n2) points out, neuter adjectives like eikaîa presuppose nouns like antígrapha ‘copies’.
[ back ] 48. Janko 1992:22, 26. I find one case, scholia A for XVII 214, where koinaí explicitly describes ekdóseis, that is, the word for ‘edition’ (in however limited a sense) in the plural. Janko p. 26n29 argues that the references by Didymus epitomators to “all” or “most” manuscripts should be understood to mean all or most of the named editions (including that of Aristarchus himself), not the manuscripts in general. There may be, however, an analogous pattern of reference to unnamed editions, if Allen 1924:278 is justified in printing the emended reading of scholia A for Iliad VIII 349, αἱ πλείους τῶν δημωδῶν ‘the majority of the dēmṓdeis [= popular texts]’ (emended by Villoison from αἱ πλείους τὴν δημώδη), which would be the equivalent of αἱ πλείους ‘the majority’ in scholia T.
[ back ] 49. Relevant is the question formulated by Allen 1924:278: “is there ground to believe that the koinḗ originally meant ‘usual’, ‘universal’, and that the disparaging sense was secondary?” The discussion that follows will treat in detail Allen’s own answer to this question.
[ back ] 50. Janko 1992:26.
[ back ] 51. Janko 1992:26.
[ back ] 52. Janko 1992:26.
[ back ] 53. Cf. e.g. Davidson 1994:54–72.
[ back ] 54. On references to a diórthōsis of the Iliad by Aristotle, see Blum 1991:21–22 and 69–70n45, who nevertheless sides with the view that “the Alexandrian philologists of Homer ... did not pay any attention to the Aristotelian diórthōsis of the Iliad” (p. 70). In what follows, I attempt to explain why there are gaps in references by Alexandrian critics to the editorial judgments of the Peripatetic School.
[ back ] 55. The phrasing here reflects the attested usage of diórthōsis in the sense of an ad hoc editorial judgment, as distinct from that of an overall editorial procedure. Both senses are well represented in the Homer scholia.
[ back ] 56. Blum 1991:70n46.
[ back ] 57. Pfeiffer 1968:98; relevant are the perceptive remarks of Rengakos 1993:15.
[ back ] 58. Blum 1991:22. We would expect questions of diórthōsis to be raised in the six books of Aristotle’s Homeric Questions, now lost, to which Diogenes Laertius 5.81 makes reference.
[ back ] 59. Pfeiffer 1968:98.
[ back ] 60. Blum 1991:101 considers 291 BCE as a possible date for Zenodotus’ appointment, but he judiciously weighs the alternative possibilities of later datings. On this topic, I refer to the ongoing work of J. D. Morgan, who has revived the long-neglected argument that it was Ptolemy II, not Ptolemy I, who appointed Zenodotus head of the Library of Alexandria and tutor to his children, and that Ptolemy III, not Ptolemy II, made Apollonius of Rhodes head of the library and tutor to his son Ptolemy IV, not Ptolemy III. This reassessed chronology makes it possible to date Zenodotus’ editorial activity to the reign of Ptolemy II (283-246), after the death of Demetrius of Phaleron in the late 280-s (on which subject there is more in ch. 7).
[ back ] 61. For an objective assessment of the Suda reference on Zenodotus, see Blum 1991:101. The idea that Zenodotus, in the process of editing Homer, did indeed produce his own text is argued—to my mind persuasively—by Rengakos 1993:12–14 (his discussion also provides an admirable bibliographical survey of opposing views). He also argues that Aristarchus had direct access to the Homer edition of Zenodotus, even if Didymus and Aristonicus may not have (p. 14). So too Apollonius of Rhodes and Callimachus, both contemporaries of Zenodotus, had access to such a text (p. 14). More on this point later.
[ back ] 62. Pfeiffer 1968:105–106.
[ back ] 63. Pfeiffer 1968:117–118.
[ back ] 64. Pfeiffer 1968:106.
[ back ] 65. Pfeiffer 1968:106.
[ back ] 66. Cf. Allen’s argument on diórthōsis, as quoted at n43.
[ back ] 67. Blum 1991:22.
[ back ] 68. Laum 1928:105.
[ back ] 69. Laum 1928:104–105. Aristotle Poetics 1461a21 identifies Hippias of Thasos as the initiator of the interpretation οὗ instead of οὐ. What matters here is not the philological validity of the interpretation itself, which is negligible, but the usage of the word diorthoûn with reference to diacritics. Cf. Hintenlang 1961:76n1, who also gives another example of such issues of interpretation: Aristotle Poetics 1461a22–23 (cf. Plato Republic 2.383a) on διδόμεν for δίδομεν in Iliad II 15.
[ back ] 70. Laum 1928:104–105.
[ back ] 71. Laum 1928:108.
[ back ] 72. Laum 1928:106.
[ back ] 73. See Callisthenes FGH 124 T 10; cf. Pfeiffer 1968:71.
[ back ] 74. The Vita Marciana speaks of Aristotle’s ékdosis of the Iliad, which he gave to Alexander (τὰ γεγραμμένα αὐτῷ ὁμηρικὰ ζητήματα, καὶ ἡ τῆς Ἰλιάδος ἔκδοσις ἣν ἔδωκε τῷ Ἀλεξάνδρῳ Aristotle Fragments p. 427.5 ed. Rose). In the Vita Latina, this edition is called a dictamen (Aristotle Fragments p. 443.5–6 ed. Rose: Homerica commenta scripta ab eo et Yliadis dictamen quod dedit Alexandro). For more on the nárthēx edition of Homer—and on the reliability of the reports about it—see ch. 7.
[ back ] 75. It goes without saying that, for Isocrates, the writing of a speech, expressed by way of gráphein ‘write’ (cf. Panathenaicus 1), is tantamount to the composing and even the notional ‘delivering’ of a speech. On gráphein ‘write’ as a notional speech-act, see PH 233n86.
[ back ] 76. Cf. Jebb 1893 II 113.
[ back ] 77. Although the Lyceum cannot be identified specifically with the school of Aristotle until a later period (after the philosopher’s death, when his successor Theophrastus institutionalized the school in the Lyceum), the place was known as a sort of forum for philosophers even before the era of Isocrates (cf. e.g. Plato Lysis 203a).
[ back ] 78. For these usages in the Homeric scholia, see n46 above.
[ back ] 79. On the possibility that the ‘commentaries’ of Zenodotus, accompanying his edition of Homer, had the format of oral disquisitions rather than written texts, see Rengakos 1993:14. In the case of Aristarchus, on the other hand, we can be more certain that his hupomnḗmata ‘commentaries’ were indeed written texts: see n40 above.
[ back ] 80. The verb used in Plato Ion 530a for the rhapsode’s performance is diatríbein, which matches the noun diatribḗ that we have seen used in the passage from Isocrates with reference to the performance of the ‘sophists in the Lyceum’. We note again the idea that these ‘sophists’ are mnēmoneúontes ‘mentioning’ received knowledge about the poet they perform; so also in the Ion, the rhapsode notes that his attention is always awakened when someone ‘mentions’ Homer, and the verb used is mnēsthênai (532c: ἐπειδὰν δέ τις περὶ Ὁμήρου μνησθῇ); this verb is here made parallel to dialégesthai ‘engage in discourse’ (532b: ὅταν μέν τις περὶ ἄλλου του ποιητοῦ διαλέγηται). Later on in the Ion, the same theme of the rhapsode’s awakened attention is transferred from the act of making comments on the poet (536c: περὶ μὲν Ὁμήρου ὅταν τις μνησθῇ) to the act of actually performing the poet (536b: ἐπειδὰν μέν τις ἄλλου του ποιητοῦ ᾄδῃ).
[ back ] 81. At Plato Ion 531a-b, the rhapsode’s ‘speaking about Homer’ is now expressed by way of ex(h)ēgeîsthai—a term even more appropriate to the idea of ‘commentary‘.
[ back ] 82. Lehrs 1882:248–249.
[ back ] 83. Laum 1928:60. Laum’s work remains indispensable, despite the need for some corrections (cf. the bibliography in Turner 1987:159). On the invention, by the Alexandrian critic Aristophanes of Byzantium, of the actual notation-system for ancient Greek accents, see Laum p. 62. At pp. 100–102, Laum prints the testimony from the manuscript Parisinus 2102 reporting Aristophanes’ invention, and he traces this testimony, however flawed, to Theodosius of Alexandria, who flourished around 400 CE. Pfeiffer 1968:179n1 questions the reliability of the text as printed by Laum.
[ back ] 84. Laum 1928:327.
[ back ] 85. Laum 1928:62. Of the over 150 mentions of Aristarchean diórthōsis that Laum counted in the Homer scholia, he found in those contexts only a single reference to an accentual variation noted by Aristarchus (scholia A for Iliad XIII 191, χρόος instead of χροός).
[ back ] 86. Laum 1928:327. It can be said in general for Greek literature that only in the Byzantine editions of the ninth and tenth centuries did it become a regular practice to mark the accent on each word in a given text. Laum 1928:63n2 finds a dramatic illustration of neglected accent-markings in the case of a papyrus containing Menander’s Perikeiromene, dated between the first and second centuries CE: in this text, we find that punctuations are meticulously supplied (three different categories), that miswritings are corrected, that variants are added, that elided letters are restored, and that role-assignments are straightened out—but there are no accent-marks.
[ back ] 87. In addition to the evidence of sporadic accentual notations in the papyri, there is the evidence provided by ancient experts in Greek grammar. Two key sources of information about accent are Herodian (second half of the second century CE) and Theodosius (around 400 CE); as Laum points out (pp. 29–30), the first was interested more in the theoretical aspects of accentuation and the second, in the practical. The Byzantine conventions of marking accents go back to Theodosius, whose orthographic system reveals some surprising divergences from the accentual patterns attested by Herodian and by his predecessors in the Library of Alexandria. Modern editors of ancient Greek texts anachronistically obey the Byzantine accentual orthographic system that we can trace back only as far as Theodosius, thus bypassing Herodian, not to mention the earlier testimony of papyri with marked accents. It is not the accents of individual words that turn out to be different in the earlier sources: rather, it is the accentuation of word-combinations. For example, modern editors print a polysyllabic oxytone word consistently with a grave accent when that word is followed by another word without an intervening syntactical break, and yet the evidence of the papyri and of the Homeric scholia indicates that the accent in this context could in fact be acute, not grave: see Laum pp. 152, 159, 161. I say “could,” not “should,” because Moore-Blunt 1978 has found several instances of papyri dated earlier than 400 CE where we do see the spelling of grave as well as acute in this same context (cf. Mazzucchi 1979). Laum treats the earlier pattern of acute spellings as a constant, whereas in fact it is a gradually disappearing tendency. The point remains—and Laum says this just as effectively as Moore-Blunt—that earlier patterns of ancient Greek accentuation are conditioned by the melodic contour, as it were, of the overall syntax. Relevant is the formulation of West 1992:199 concerning a general tendency in ancient Greek melodic traditions: “when the accent [is] on the final syllable of a word, and is not circumflex, and not succeeded by a grammatical pause, then the melody does not fall again until after the next accent.”
[ back ] 88. Laum 1928:63. There is a particularly striking illustration given by Moore-Blunt 1978:161–162: “papyri also demonstrate how, in questions, the final syllable of the last word (i.e. the final syllable of the sentence) could bear the high pitch, regardless of the normal [I would prefer to say not normal but lexical] accentuation of the word.” As Moore-Blunt points out (p. 162), such a spelling of an acute is especially useful where the syntax has no interrogative particle, as in the case of ακηκουκάς at Herodas 5.49.
[ back ] 89. Laum 1928:63.
[ back ] 90. Laum 1928:63, 163, 327.
[ back ] 91. Laum 1928:63 offers the dictum that accentuated texts were meant for students, not scholars. I hasten to add that the category of “student” needs to include students of performance traditions. {Laum p. 63n1 disagrees with Wackernagel, who thought that accented texts were meant for scholars.Laum p. 63n1 disagrees with Wackernagel, who thought that accented texts were meant for scholars.}
[ back ] 92. Isocrates Panathenaicus (Oration 12) 18–19 and 33, as quoted above.
[ back ] 93. Wackernagel 1914 [1953]:1175.
[ back ] 94. N 1970:121.
[ back ] 95. Wackernagel [1953]:1176.
[ back ] 96. Wackernagel 1983 [1953]:1103: “Aber die Zitate basieren doch selbst wieder auf der mündlichen Rezitation der homerischen Gedichte [reference also to Schulze 1892:213n3]; da die Rhapsodik bis an die Anfänge der Philologie hinanreicht, haben wir hier eine ununterbrochene Traditionskette. Dass beim mündlichen Vortrag neben den Versikten auch der musikalische Wortton zum Ausdruck kam, is unzweifelhaft.” Cf. also Lehrs 1882:258, quoted by N 1970:121, where the discussion is taken further.
[ back ] 97. Wackernagel 1893 [1953]:1103.
[ back ] 98. Wackernagel 1909 [1953]:1119–1120.
[ back ] 99. The observation of Didymus in the scholia A for Iliad XVI 467c, that Aristarchus would not leave something aparamúthēton, in other words, that he would not miss the opportunity of making contextual comparisons with all available internal evidence, does not mean that his priorities ranked internal logic ahead of manuscript evidence: in this regard, I find persuasive the discussion of Ludwich 1885:92, 97, 109 (yes, Aristarchus is an analogist, but not at the expense of the manuscript evidence), 114 (striking examples where Aristarchus reads an anomalous form instead of substituting an analogous form). There is an interesting critique by Janko 1990:332-335 of the analogizing tendencies in the Monro-Allen 1920 Oxford Classical Texts edition of the Iliad. With reference to the manuscript reading δ᾿ ἑκάθεν in Iliad XVIII 107, which the OCT edition replaces with δὲ ἑκάς on the analogy of the manuscript reading in Iliad V 791, Janko argues that the manuscript reading δ᾿ ἑκάθεν in Iliad XIII 107 “is superior precisely because it is different, even though it happens to include a more recent linguistic form (which the poet uses elsewhere).” I agree with the wording except for an important detail: I would substitute authentic for superior. Given that Zenodotus and Aristophanes both read δὲ ἑκάς in Iliad XIII 107, I suggest that this reading too is authentic (I will have more to say presently on Zenodotus and Aristophanes). What is lectio difficilior for one period may be lectio facilior for another (here again I disagree with Janko, pp. 332–333n21). Cf. Pasquali 1952:122 and di Luzio 1969:144–145.
[ back ] 100. N 1970:121, following Lehrs 1882:258.
[ back ] 101. Lehrs 1882:258.
[ back ] 102. West 1970; see also West 1981:114.
[ back ] 103. Ludwich 1898:163 remarks on the absence, in the Homer scholia, of any reference to variant readings that are explicitly connected with rhapsodes. The scholia mention nothing along the lines of “ἀντίγραφα τῶν ῥαψῳδῶν” or the like. Also, I find no mention of Homērídai in the Homer scholia edited by Erbse.
[ back ] 104. The uniqueness of this attestation can be verified by way of the Thesaurus Linguae Graecae.
[ back ] 105. Erbse vol. 5 p. 130.
[ back ] 106. Cf. also the extract quoted in Suda, omicron 760: δειξάτω, οὗ κεῖται Ὁμήρου ῥαψωιδιῶν στίχος. ἀλλ᾿ οὐδ᾿ ἂν Ἴωνα δοκῶ τὸν ῥαψωιδὸν ἐξευρεῖν ‘let him show where a line of Homer’s rhapsōidíai is attested; but I don’t think that even Ion the rhapsode could find it’.
[ back ] 107. We may note in this context an anecdote reported by Vitruvius about Aristophanes of Byzantium, T 17 in the Aristophanes edition of Slater 1986. The text is given in the Appendix of the printed version of Poetry as Performance. In this narrative, the idea of reading aloud is equated with the idea of oral performance, and the details about the competition at the ludi suggest that earlier versions of this narrative concerned rhapsodic traditions of performance. Particularly noteworthy is the detail about the fixed sequence of competing poets.
[ back ] 108. In PH 20–28, there is an extended discussion of the phenomenon that I call reduced melody or recitative in hexameter traditions as performed by rhapsodes. For more on the melodic contours of the hexameter, see West 1986:45, who argues that the epic singer of the eighth century “followed the contours given by the word accents”; also, that “this tradition was perpetuated by the rhapsodes, but in a gradually decaying form,” and that “the rhapsodes preserved many archaic accentual features of Homeric Greek into the Hellenistic age”; cf. also West 1981:114 and 1992:208–209. I agree with most of these formulations, though I resist the idea of a “decaying form.” On the concept of recitative, see van der Werf 1967. In the traditions of the Old French chansons, as he argues, there are cases of distinctly recitative melodies and distinctly arioso ones but there are other compositions where “we can no longer discern whether the original of a given line was a recitative on d or an arioso melody with d as a tonal center” (van der Werf 1967:234). In other words, there are instances where “we cannot conclude from the preserved music whether a manuscript gives us a simplified variant of an arioso original or an ornamented variant of a strict recitative.” It is clear that “a trouvère recitative could easily be transformed into a trouvère arioso, or an arioso transformed into a recitative.” Though it is impossible at times to determine in which direction the shift is headed, whether it is from arioso to recitative or vice versa, it is clear that these two styles were not “two rigorously separated styles for the jongleurs, notators, and scribes at the end of the thirteenth century.” We may compare the ancient Greek traditions associated with the “lyric” Stesichorus and the “epic” Homer, as discussed in PH 49–51.
[ back ] 109. On the distinction between sung hexameter and rhapsodic hexameter, see West 1986:44. At pp. 43–45, West collects valuable comparative evidence on distinctions between repeated and varied melodic contours in line-by-line epic performance (especially with reference to the French chansons de geste, p. 43n12, and Kirghiz epic, p. 44n15).
[ back ] 110. Wackernagel 1893 [1953]:1103, quoted at n96 above.
[ back ] 111. Scheller 1951. I first applied the findings of Scheller in N 1970:111–112, 120–121. Scheller p. 10 resists the negative judgment of Wilamowitz 1916:8–9 about the value of the Alexandrian tradition on accents. According to Wilamowitz, this tradition represents an arbitrary application of analogy principles, revealing an unawareness of etymology and of the fundamental principles of word-formation; he thinks that linguists should be the last people on earth to pay so much attention to these patterns. It is thanks to the research of Wackernagel, as Scheller points out, that the negative judgment of Wilamowitz can be “modified.” Besides accent, there are also variations in breathings that must have survived by way of performance traditions: see N 1972:66 on such Homeric contrasts as ἁμός vs. ἄμμι, ὑμός vs. ὔμμι (see Laum 1928:365 on the spelling ἀμμι in Papyrus A for Bacchylides 17.25).
[ back ] 112. Scheller 1951:9n3. See further Comotti 1989:91 on the Delphic Hymns, where syllables having acute or even grave accent in any given word consistently avoid any pitch that is lower than the other pitches assigned to the other syllables in the same word. For more on the relationship of pitch accent and melody in ancient Greece, especially on the more archaic pattern where the melodic patterns are conditioned by the accentual patterns, see PH 39 and n113. Cf. West 1981:115, 1986:45, 1992:199. See also Comotti p. 91 on the concept of logôdes mélos ‘speech-like melody’ in Aristoxenus Harmonics 1.18 p. 23.14 ed. Da Rios. Comotti p. 92 juxtaposes this concept with the arguments of Dionysius of Halicarnassus De compositione verborum 11.58ff, p. 40.17 ed. Usener-Radermacher, who insists that melody controls the words, not the words the melody, and who cites Euripides Orestes 140–142 (making mistakes with the accents of some words: see Comotti p. 92n6). According to Comotti, Dionysius is slanting his argument by citing Euripides, who is musically the most innovative of the tragedians. He argues that the Delphic Hymns are musically far more conservative than the lyric compositions of Euripides. I agree. West 1992:199 attempts to explain the melodies of Euripides’ Orestes in terms of constraints imposed by the principle of responsion between strophe and antistrophe.
[ back ] 113. Cf. Laum 1928:164 on a convention found in the papyri: there is a tendency to signal an acute accent belonging to only one word within a given string of words, instead of signalling all the acutes belonging to all the words (for example, P.Oxy. III 448, from Odyssey xxii 184: τηι δ᾿ετερηι σακος ευρύ γερον, which is spelled by modern editors as τῆι δ᾿ ἑτέρηι σάκος εὐρὺ γέρον). To mark the one acute is to mark the highest point of the melodic contour. (On the practice of marking polysyllabic oxytones with acute rather than grave in some clause-medial situations, see n87 above.) It may be possible to compare this kind of pattern with what we find in the Homeric scholia, which frequently refer not to individual words but to strings of words (e.g. Laum p. 378), reflecting a practical mode of commenting on texts that had once been spelled without word-divisions. In the scholia, there is a tendency to comment on only one accent belonging to only one word within a given string of words instead of commenting on all the accents belonging to all the words (e.g. Laum p. 143). More in N 2000 about the ancient practice of marking melodic contours in papyri.
[ back ] 114. See p. 117 above. Cf. di Luzio 1969:141 and D’Ippolito 1984:224–225.
[ back ] 115. Van der Valk 1964:76 reviews the facts; as the discussion that follows will show, I do not share his interpretation of the facts.
[ back ] 116. Muellner 1976:58–62; also p. 24 with n18. {With reference to Lehrs 1882.258.With reference to Lehrs 1882.258.}
[ back ] 117. Muellner 1976:58–62.
[ back ] 118. Zenodotus athetized Iliad I 4–5 according to Athenaeus 1.12f, which means that he considered what he had read in these lines to be un-Homeric. But the point is, as I understand it, that οἰωνοῖσί τε δαῖτα is in fact what he had read at Iliad I 5. On athetesis as an editorial judgment rather than an act of omission—a judgment that may or may not affect manuscript transmission—see n134 below. Kirk 1985:53 remarks on Zenodotus’ reading οἰωνοῖσί τε δαῖτα: “Aristarchus (who is evidently Athenaeus’s source, through Aristonicus, at [Athenaeus] [1.]12e–13a, cf. Erbse I, 9) tried to refute [it] on the erroneous ground that Homer never uses δαίς of animal food—as he does in fact at [XXIV] 43.” Ludwich 1885:87 points out that we cannot be sure what exactly Aristarchus said. So I would reply to Kirk: whatever the merits, or even the substance (which, to repeat, we do not know for sure), of Aristarchus’ arguing on internal grounds for οἰωνοῖσί τε πᾶσι in Iliad I 5, his primary reason for preferring this reading over οἰωνοῖσί τε δαῖτα was surely not internal evidence but the external evidence of what he considered the “superior” manuscript versions (cf. Ludwich p. 89). More on this point presently.
[ back ] 119. For the view that Aeschylus was aware of the version οἰωνοῖσί τε δαῖτα in Iliad I 5, see e.g. Pfeiffer 1968:111, with further examples from tragedy (we may note that Pfeiffer would have assumed that such an awareness was based on the existence of such a version in the manuscript tradition, not in the performance tradition of Homer); see also Pasquali 1952:236–237. See also Janko 1992:23, who accepts the idea that Zenodotus’ reading may go back to the fifth century BCE but who still prefers to think of it as “an early emendation”—meant “to remove the ‘problem’ that not all birds eat flesh” (cf. the reasoning in the scholia b for Iliad I 5; cf. also Eustathius 1.390.26). Kirk 1985:53 describes Zenodotus’ reading as “a fussy change of the vulgate.” Rengakos 1993:30n1 draws attention to Kirk’s assumption of “change” here.
[ back ] 120. The Homeric examples adduced by Ludwich 1885:89n55, where the notion of ‘all’ is logically a matter of hyperbole, are for me persuasive (especially Iliad V 52, Odyssey xviii 85). As for Zenodotus’ reading οἰωνοῖσί τε δαῖτα, we have in its defense some internal testimony as well, most notably the picture of a ravenous lion who lunges for his meal in Iliad XXIV 43 (ἵνα δαῖτα λάβῃσιν). We have already seen this passage cited by Kirk p. 53—though he was of course not defending Zenodotus but attacking Aristarchus.
[ back ] 121. Wolf 1795. An indispensable summary of Wolf’s position is Pfeiffer 1968:215–218. Apthorp 1980:xiii, in line with the arguments of Bolling 1925, uses the term “Wolfian vulgate” in a negative sense to characterize post-Wolf Homer editions that tend to discount the judgments of Alexandrian critics, especially with reference to criteria of excluding lines in the Homeric corpus. Such an edition is the Monro-Allen 1920 Oxford Classical Texts version of the Iliad. Pfeiffer pp. 214–215 outlines the efforts of Lehrs 1882 and Ludwich 1884/1885 to rehabilitate the authoritativeness of Aristarchus as editor of Homer, as also the arguments of Erbse 1959 challenging this rehabilitation. In offering his own counterarguments to Erbse’s position, Pfeiffer begins by saying (p. 215): “it looks to me as if by a sort of unconscious counter-revolution Wolf has now been put back on the throne from which Lehrs had driven him.”
[ back ] 122. Van der Valk 1949, 1963/1964.
[ back ] 123. For Zenodotus: Wilamowitz 1916 (e.g. pp. 120–121, 261n2, 262n2; cf. van der Valk 1964:10) and Pasquali 1952 (e.g. pp. 207, 235–236; cf. van der Valk p. 15). For Aristarchus: Lehrs 1882, Ludwich 1884 / 1885. Given that there is unanimous manuscript authority for the reading οἰωνοῖσί τε πᾶσι in Iliad I 5 as opposed to Zenodotus’ reading οἰωνοῖσί τε δαῖτα (as reported in Athenaeus 1.12f), Ludwich 1885:89 defends the first reading and rejects the notion that it was a conjecture of Aristarchus. As I indicated earlier (n120), I agree with Ludwich that it was not a conjecture. But I do not agree with him that the alternative reading, οἰωνοῖσί τε δαῖτα, is therefore a conjecture. We have already seen evidence that Zenodotus’ reading follows a version that was current in the Athenian performance traditions of the fifth century, the era of the three canonical tragedians. In order to maintain the idea that this reading was a conjecture or interpolation, Ludwich is forced to say that it must have been an “old” one indeed (see also the position of Janko as cited at n119 above). For a critique of the position taken by Ludwich and others, that the Aristarchean text comes closest to a Homeric “original,” see di Luzio 1969:6–9.
[ back ] 124. Allen 1924:302–327.
[ back ] 125. Pfeiffer 1968:210–219.
[ back ] 126. Kirk 1985:43. Editions of the Iliad / Odyssey by Nauck, 1877 / 1875. On the position of Wilamowitz, see n123. Kirk’s position is disputed by Rengakos 1993:22n3.
[ back ] 127. Janko 1992:22–29.
[ back ] 128. Janko 1992:2–25. His reference is to Kirk’s commentary, 1985:43. Cf. also van Thiel 1991 and the comment of Janko 1994:291: “T.’s attitude to the Alexandrians (p. ix-xiii) derives, like mine, from the great work of van der Valk.”
[ back ] 129. Janko 1992:21n6.
[ back ] 130. Lehrs 1882; Ludwich 1884/1885, 1898. The principle is most forcefully stated, as “Lehrs’ Law,” by Ludwich 1884:86. I agree with the criticism of van der Valk’s methods by Nickau 1977.31n1. I resist in general the idea that the Alexandrian editors would have contemplated any kind of theory-driven policy for making conjectures. Their theories may well have led them to argue for some variants over others, but that is a far cry from the idea that they rewrote the text as they saw fit.
[ back ] 131. Nickau 1977:48; cf. Slater 1989:42n17. Slater goes on to say at a later point: “ultimately, as Nickau [1972 column 34] says, no amount of generalizing theory relieves the modern critic from assuming that every reading of an early grammarian could be a variant until some alternative is demonstrated.”
[ back ] 132. Rengakos 1993 (especially pp. 11, 23, 31). He insists on the concept of citation—as distinct from allusion—as a way of underlining the fact that, for the Alexandrian poets, Homer was the absolute source, not only the unsurpassable model (p. 9). Beyond the poetry of Apollonius and Callimachus, Rengakos draws special attention to the text of another Hellenistic poet, Rhianos, as witness to a set of variants derived from a Homer text that is markedly different from the Homer texts of Zenodotus, Aristophanes, and Aristarchus (p. 10). The value of the Homeric textual sources used by other poets, including “Euripides,” Antimachus, Philitas, and Aratus, is also discussed (p. 11).
[ back ] 133. Apthorp 1980. We may note in this context Apthorp’s severe criticism (p. xiv) of van der Valk’s methods.
[ back ] 134. Apthorp’s valuable work (1980) builds on the findings of Bolling 1925, which he summarizes as follows (p. xiv): “where a line is weakly attested by the medieval manuscripts and papyrus evidence was available, then that line was almost invariably absent from the papyrus or papyri.” Apthorp adds: “papyri published since 1925 have served only to confirm Bolling’s position, sometimes dramatically.” (It is important to note that Apthorp is referring here to papyri dated after, not before, 150 BCE or so.) In terms of Bolling’s position, the evidence of the papyri indicates (to follow Apthorp’s wording) that “the numerus versuum of our medieval vulgate, when purged of these weakly-attested lines, is identical with the numerus versuum of Aristarchus.” Apthorp notes the important distinction (p. xv) between athetesis, where Aristarchus marks with an obelus a verse that he deems non-authentic though he leaves it in the text, and outright omission; as he points out, “the evidence shows conclusively that [Aristarchus] omitted only lines which were absent from the vast majority of his manuscripts.” For further insights on Aristarchus’ methodology—on the levels of (1) manuscript evidence and (2) content—in deciding whether or not a Homeric line was authentic, see Lührs 1992. I find especially important the observations of Lührs p. 11 concerning instances where Aristarchus excluded verses in his Homeric text and then apparently signaled in his commentaries (that is, in his hupomnḗmata) that these plus-verses had been included in the Homer text of Zenodotus: as Lührs argues, Aristarchus did not think of such verses as conjectures made by Zenodotus but rather, more simply, as verses deemed authentic by Zenodotus and non-authentic by himself. Presumably, such verses must have been attested in some manuscripts and absent in others for them to be included and excluded in the Homer texts of Zenodotus and Aristarchus respectively (again, Lührs 1992).
[ back ] 135. Janko 1992:21n6: “interpolations” (see also Janko 1990:334). Despite this position taken by Janko, we see that a few pages later in his commentary (1992:27–29) he defends, I think successfully, the authenticity of what he calls “the most notorious case” of such plus-verses, Iliad IX 458–461. This passage, missing in the medieval manuscript tradition and omitted from a papyrus that covers this stretch of the Iliad, happens to be known only from a citation by Plutarch (Quomodo adolescens poetas audire debeat 26f), who claims that Aristarchus omitted these lines “out of fear” (the lines concern Phoenix as he contemplates killing his father).
[ back ] 136. In other words, I infer that the omission or non-omission of a verse is really a question of variation, not conjecture, in that Aristarchus will omit a verse—whatever contextual reasons he may adduce for such an omission—only when the documentary evidence allows him to do so. This extreme conservatism of Aristarchus, Apthorp notes (1980:xv), was appreciated by Ludwich and Bolling—but definitely not by van der Valk.
[ back ] 137. Apthorp (1980:xvi) goes on to argue that “the numerous lines absent from all our manuscripts which we know to have been pre-Aristarchean but absent from Aristarchus’ edition—some cited by the scholia, some present in extant Ptolemaic papyri, some included in ancient quotations or discussions of Homer—stand condemned as interpolations alongside the weakly-attested lines of the mediaeval manuscripts.” I object to this condemnation, as also to Apthorp’s followup remark (p. xxvn2): “Thus the attempt by A. di Luzio [1969] to defend numerous plus-verses of the Ptolemaic papyri on so-called linguistic and stylistic grounds must be pronounced unsuccessful.” I remain unconvinced that a verse must be an interpolation if it is both non-Aristarchean and pre-Aristarchean.
[ back ] 138. Janko 1992:21n6. Apthorp 1980:110n64 speaks of “the limited influence of Aristarchus on the text of the subsequent tradition within his numerus versuum,” with further discussion at pp. 37–38.
[ back ] 139. HQ 76–77.
[ back ] 140. The term “eccentric papyri” is claimed by Allen: see 1924:302.
[ back ] 141. [S.] West 1988:45; cf. also 1967:15.
[ back ] 142. [S.] West 1967:13; cf. Apthorp 1980:60.
[ back ] 143. Cf. D’Ippolito 1984:224–225, especially p. 224n12. In what follows, it will be clear that I generally accept the arguments of di Luzio 1969 that these “eccentric” Ptolemaic papyri of Homer, dated mostly before around 150 BCE, contain a significant number of genuine variants, reflecting an ongoing rhapsodic tradition. In his article, di Luzio offers case-by-case counterarguments about Homeric variants discounted by West and by other critics (cf. Del Corno 1960 and 1961) who have examined the “eccentric” Homer papyri. More on di Luzio in n137 above.
[ back ] 144. [S.] West 1988:45; Apthorp 1980:1–3.
[ back ] 145. Allen 1924:303.
[ back ] 146. [S.] West 1988:47.
[ back ] 147. Allen 1924:326; cf. also 320.
[ back ] 148. See di Luzio 1969:7–8.
[ back ] 149. Labarbe 1949. The attempt of Lohse 1964 to refute the findings of Labarbe is in my opinion unconvincing: the criteria that Lohse sets up (especially pp. 5–7) for determining what is or is not oral poetry seem to me too rigid to be applicable to Homeric performance traditions in the fourth century BCE For the purposes of my own argumentation, it is essential simply to stress Lohse’s concession that any Homer text used by Plato was at least to some degree different from what we have (e.g. p. 7). I concede that there may well be instances of Homeric quotations where Plato has selectively introduced his own rewordings (cf. Lohse 1967).
[ back ] 150. Labarbe 1949:419.
[ back ] 151. Labarbe 1949:415–416.
[ back ] 152. Labarbe 1949:424; also pp. 90–94, 98–99.
[ back ] 153. Labarbe 1949:423. To this extent, I can accept the formulation of Erbse 1959:301 concerning an “Athenian Recension” of Homer; cf. Lohse 1967:230n14.
[ back ] 154. Labarbe 1949:423.
[ back ] 155. Labarbe 1949:423–425. I am not convinced by the specific discussions offered by Lohse (e.g. 1965:259n21, 262n27) concerning criteria for determining what is or is not genuine formulaic variation.
[ back ] 156. Labarbe 1949:423. Cf. Lohse 1967:229. On the concept of plus-verses, see p. 139 above.
[ back ] 157. Labarbe 1949:423. {Here he cites Mazon pp. 275, 276–277, on the existence of a Panathenaic Homer. According to Mazon, there was a text from Chios that the Peisistratidai used as a control-text; the influential role of Peisistratos in Homer-transmission is emphasized; so the text is a script, as it were, dating from the 6th century.Here he cites Mazon pp. 275, 276-277, on the existence of a Panathenaic Homer. According to Mazon, there was a text from Chios that the Peisistratidai used as a control-text; the influential role of Peisistratos in Homer-transmission is emphasized; so the text is a script, as it were, dating from the 6th century.}
[ back ] 158. HQ 40–43.
[ back ] 159. HQ 76.
[ back ] 160. Apthorp 1980:3 points out that the numerus versuum of Aristarchus does not match that of his predecessor Aristophanes in six known cases, and in each case the readings in papyri dated after 150 BCE back up Aristarchus, not Aristophanes. On these grounds, he argues against supposing “that our manuscripts are descended from some other recension which just happened to largely coincide with that of Aristarchus.” I would argue rather that Aristarchus may have simply surpassed Aristophanes in his editorial efforts to recover an Athenian version, on which more later. Apthorp concedes that Aristophanes’ numerus versuum was indeed very similar to that of Aristarchus.
[ back ] 161. We may note with interest the observation of Laum 1928:33 that papyri from the fourth and third centuries BCE tend to be less prone to archaizing tendencies—from a palaeographical point of view—as opposed to papyri from the second century BCE to the second CE
[ back ] 162. J. D. Morgan suggests (per litteras 30 November 1993) that this dating for the obsolescence of the “eccentric” papyri corresponds closely to 145/4 BCE, the date for the departure of Aristarchus and the other grammarians when Ptolemy VIII came to power (see p. 114 above). Morgan suggests further: “it was the grammarians who were interested in collecting and copying such papyri, and when they left, almost nobody was left to take an interest in them.”
[ back ] 163. Van der Valk 1964:266–267.
[ back ] 164. Van der Valk 1964:267.
[ back ] 165. Van der Valk 1964:267.
[ back ] 166. Van der Valk 1964:269. See also Bolling 1925:33–34, who claims that all Homeric variants result from a written rather than oral tradition, and that this tradition has a single “fountain-head,” which is “an Athenian text not earlier than the sixth century.”
[ back ] 167. Labarbe 1949:423–425.
[ back ] 168. Labarbe 1949:425.
[ back ] 169. This point applies even to situations that we could consider minimal conjecture, such as athetesis. When an Alexandrian editor athetizes, he does not propose to change any reading in the text proper but simply questions the genuineness of a line with an obelus in the margins. There is evidence that, at least in the case of Aristarchus, the editor formed his judgment about the alleged spuriousness of a line quite methodically, on the basis of not only the internal evidence but also the external evidence of the manuscript traditions. In the case of Aristarchus’ editorial work, Apthorp 1980 has demonstrated the consistency of his method. With reference to the tag perissós ‘extraneous’ in the transmitted judgments of Alexandrian critics of the Homer text, Reeve 1972:250 observes: “but if in one single case an Alexandrian athetesis can be shown to have rested on documentary evidence, the possibility must always be reckoned with that perissós has documentary authority behind it.” He then gives three examples of Alexandrian perissós verdicts that are indeed backed up by documentary evidence: in Iliad XXIII 92, IV 88, XXI 290—in that order. In the second case, there is an outright omission of the given line, an omission attributed to Zenodotus (actually, it is line 89 that he left out of his edition, while he leaves in 88 with wording that differs from what survives in the medieval manuscript tradition), and his judgment is backed up by the evidence of Papyrus 41 (the numbering follows the Monro-Allen 1920 OCT edition of the Iliad), on which see Apthorp 1980:1–2, 81.
[ back ] 170. Citti 1966 (cf. Rengakos 1993:74n5). See also Apthorp 1980:116n112, who finds “unconvincing” van der Valk’s efforts to dismiss as conjectures the variant readings of the politikaí. In the case of the Chios edition, Apthorp (p. 76) dates it earlier than Zenodotus. Citti (p. 32) dates the politikaí back to the fourth century BCE He also points out (p. 420) an important fact about the Homer citation in Aeschines Against Timarchus 149, where the verse corresponding to Iliad XXIII 77 contains a variant οὐ γὰρ ἔτι, which is shared by some of the politikaí (scholia A: ἔν τισι τῶν πολιτικῶν) as opposed to the variant οὐ μὲν γάρ of the surviving manuscript traditions: the verses that Aeschines here is asking the grammateús ‘recorder’ to read out loud come from a Homer edition that clearly had plus-verses.
[ back ] 171. E.g. Citti 1966:10, 43; at p. 17, there is even talk about an “original” reading.
[ back ] 172. E.g. Citti 1966:14, 18, 23, 43.
[ back ] 173. Citti 1966:15.
[ back ] 174. Iliad V 723 and XIV 123 are comparable instances of verse-final ἀμφίς.
[ back ] 175. Citti 1966:27.
[ back ] 176. Van der Valk 1964:7n33.
[ back ] 177. Ventris and Chadwick 1973:557–558, with special reference to Pylos tablets Un 249 and Un 267.
[ back ] 178. Van der Valk 1964:76.
[ back ] 179. Van der Valk 1964:76.
[ back ] 180. An illuminating discussion: Ritoók 1987:17.
[ back ] 181. Just as the art of the dramatic actor, hupokritikḗ, is associated here in Aristotle Poetics 1462a with the art of the rhapsode, so also in Plato Ion 536a, Ion is both rhapsōidós ‘rhapsode’ and hupokritḗs ‘actor’.
[ back ] 182. Pfeiffer 1968:71n2 gives further information, which is of great interest.
[ back ] 183. Blum 1991:70n47.
[ back ] 184. Sealey 1990:129 and 183n17, with reference to a successful publisher in the Roman era, T. Pomponius Atticus, who is said to have employed men described as anagnostae optimi et plurimi librarii ‘the best readers [anagnôstai] and the greatest number of scribes’ (Nepos Life of Atticus 13.3). Sealey p. 129 adds that “one could achieve multiple production on a small scale by setting one slave to read a text aloud while many slaves sat around him and wrote down what they heard.”
[ back ] 185. See ch. 6.
[ back ] 186. See the discussion of Hermodoros the rhapsode at p. 130 above.
[ back ] 187. See ch. 3.
[ back ] 188. See ch. 3.
[ back ] 189. On the factor of changing esthetics in the process of Homeric transmission, see also di Luzio 1969:142.
[ back ] 190. Cf. Davison 1955:21, Pfeiffer 1968:228, Janko 1992:32 (n53), 71.
[ back ] 191. Porter 1992:83.
[ back ] 192. This is not to say, of course, that his choices are necessarily “right.” We may recall the remark of Reeve 1972:258 about the dilemma faced by modern editors who restrict their choices to the evidence of the manuscript transmission as they have reconstructed it: “an editor who prints a reading when he regards another as more probable is not doing his job, and an editor who fancies he can avoid arbitrary procedure by sticking to the transmitted text is making a judgement of probability just as arbitrary as if he were to change it.”
[ back ] 193. For the theory that the Iliad and Odyssey were indeed dictated in the eighth century, see Janko 1992:22, 26.
[ back ] 194. Summary, with bibliography, in HQ ch. 2.