SOURCE: Santirocco, Matthew S. “Horace's Odes and the Ancient Poetry Book.” In Unity and Design in Horace's “Odes,” pp. 3-13. Chapel Hill: University of North Carolina, 1986.

ONE

One of the most important achievements of recent Horatian criticism has been the rediscovery of structure—not the mechanical dissection of poems into component parts, but an awareness of how form is inseparable from content and how unity proceeds from design.¹ Although the individual ode has by now received sufficient critical attention, the structure of the first lyric collection, the three books of Odes published together in 23 b.c., remains problematic. This is not to say that it has been neglected. The initial observations of von Christ and Kiessling a century ago inspired continental scholars to search for the one principle according to which all the odes were arranged.² Suggestions were usually numerological and spatial: applying the criteria of metrical, thematic, and verbal reminiscence, they described triads, pentads, enneads, and decads, with the poems in each group and the groups themselves disposed in an abstract pattern such as concentric framing or chiasmus.³ By the turn of the century an investigator could justly complain that not a single possibility had been left untried.⁴

Detailed refutation is unnecessary as such grand schemes have ceased to convince most people, if they ever did so. They were possible only because their adherents isolated meter from content and were very selective in their handling of evidence, ignoring more obvious, not to say ostentatious, signs of order. Methodology aside, very elaborate patterns also raise the question of probability. One is reminded of the reviewer who discovered that the sections of Duckworth's book on the Aeneid were inadvertently related to each other in the same Golden Ratio that Duckworth detected in the Aeneid.⁵ Finally, there is a further and more fundamental problem with any approach that sacrifices texture for architecture and that reads poetry only in terms of visual effects and not also in terms of music. As Charles Segal reminds us: “The danger is that such schemes, in their abstractional purity, lead us farther and farther away from the primary experience of the work as poetry. … These analytic patterns and numerologies, if developed beyond a certain point, lack any connection with either our fictions or our lives.”⁶

It is not surprising that a reaction has set in. Though interest in structural study is running high, most current applications of the approach to Horace are less ambitious,⁷ offering only an overview of the problem or focusing on isolated phenomena.⁸ One scholar, for instance, speaks of “any stray principle” and another of “a little window dressing.”⁹ Recently there has been a further retreat from the large-scale schematizations of the past. In their commentary on the Odes Nisbet and Hubbard characterize studies of order as “such trivialities,”¹⁰ and Gordon Williams, who omits the entire subject from his Tradition and Originality, mentions it elsewhere only to dismiss it: “While some plausible reasons for particular collocations may be guessed at, it is a waste of time to speculate on a matter of which the poet himself probably had no clear idea and which, in any case, has minimal literary relevance.”¹¹

Although understandable as a reaction to past excesses, this degree of skepticism is itself excessive for several reasons. First, several considerations suggest that Horace had at least some concern for design. These include the well-documented interest in poetry books among Hellenistic and Roman writers (discussed in sections two to four below); the aesthetic implications of the book roll, the physical format of which necessitates sequential reading; the existence of certain undisputed signs of arrangement in Odes 1-3, such as the frame of C. 1.1 and 3.30 or the grouping of Parade Odes (C. 1.1-9) and Roman Odes (C. 3.1-6); and, finally, the relatively restricted thematic repertoire that facilitates the discovery of connections among odes by poet and reader alike. It is not a priori unlikely, then, that the Odes should reveal signs of larger order.¹² In addition, the extremely skeptical position is based on two unwarranted preconceptions: first, that the whole inquiry is by nature too subjective, and second, that external arrangement has no real bearing on literary appreciation. But controls on subjectivity can be applied to this as to all other literary study (see section five below), and the critical relevance of the subject, if not yet apparent, is nonetheless precisely what remains to be tested. Ultimately, though, the skeptical position is unsatisfactory for the same reason that earlier imaginative schemes failed: both misrepresent the special nature of the collection, its heterogeneity. That this is unique to the Odes becomes apparent from consideration of the parallels and precedents available to Horace.

TWO

Most discussions of the order of the Odes cite as parallels the earlier Augustan poetry books, the works of the young Vergil, Tibullus, and Propertius,¹³ and even of the young Horace himself.¹⁴ For all these publications it is possible to demonstrate some sort of elaborate design even though scholars disagree on what the precise design is, or what bearing it has on interpretation.¹⁵ And yet, to consider the Odes in light of these earlier poetry books is to set up false and misleading expectations. Although they are certainly precedents, these earlier collections, including Horace's own first book of Sermones, are not truly parallel to the Odes, for they enjoy what the Odes lack, namely a built-in coherence, both formal and thematic, apart from any coherence that conscious artistic arrangement might later impose. The Bucolics are all hexameters, so too are the Sermones, and elegies are composed in the couplet that gives its name to the genre. Similarly, a general affinity of subject matter unifies each collection: the Bucolics are largely pastoral, the Sermones satiric (in Horace's sense of the word, not ours), and the elegies erotic.

The first three books of Odes, on the other hand, are a very heterogeneous collection. Formally, they comprise eighty-eight poems in twelve different meters, an achievement that Horace repeatedly vaunted. In tone and topic, too, they range equally widely. Thus, whereas a unified plan may lie behind the other Augustan poetry books, the Odes are less likely to be structured in so coherent a fashion. For them, true parallels are to be found not in the homogeneous collections of the other Augustans but rather in the heterogeneous collections of earlier writers. The common misinterpretation that the ordered poetry book originated in the Augustan Age¹⁶ and the fragmentary nature of literary survivals from other periods have obscured the relevance of Hellenistic and Roman republican practice.

THREE

The poetry book seems to be a Hellenistic phenomenon. Before that time (and even to some extent afterward) literature arose out of a specific social context, either ritual, agonistic, or sympotic, and was intended for performance within that social setting rather than for individual reading. It was in the fifth century that books began to be circulated, owing perhaps to the popularity of tragedy and the teaching methods of the sophists. However, we are totally ignorant about the principles of arrangement, if any, operative in these books, and, in any case, it was not until the Hellenistic Age that a truly “bookish” society emerged.¹⁷

The standard reconstruction of the origin and development of the poetry book during and after this period was offered by Kroll over fifty years ago.¹⁸ Since then, however, impressive papyrus discoveries and literary analyses have made a new account necessary. Kroll, for example, maintained that the editions of the older poets which the Hellenistic scholars produced were used by the Hellenistic poets as models for the creation of new poetry books. Though the interaction between literature and scholarship was very great at Alexandria, this genealogy is nonetheless dubious because the Hellenistic editions of the classical authors and the books in which the Hellenistic poets published their own works were structured along very different lines. Basically, they arose out of two very different impulses which, for convenience, may be termed the editorial and the aesthetic.

The former is represented by the standard Alexandrian edition of the lyric poets in which Horace surely read his models.¹⁹ For several of the authors contained therein, very general features of arrangement can be discerned. An epigram (A.P.7.17) by Cicero's freedman, Tullius Laurea, alludes to a division of Sappho's works into nine books,²⁰ and the assignable fragments suggest that the first books were metrically homogeneous, with a shift to arrangement by subject matter in the last book, the Epithalamians.²¹ Within the individual books no principle of ordering can be safely adduced despite Lobel's suggestion of alphabetization.²² In the arrangement of Pindar, Bacchylides, and Simonides, on the other hand, meter played no discernible role as poems were grouped by eidē or types into hymns, paeans, epinicians, and the like. Again, there is no evidence as to how poems were arranged within these large categories, except for alleged alphabetization in the British Museum papyrus of Bacchylides's Dithyrambs²³ and for the fact that Pindar's epinicians are classed by festival²⁴ whereas those of Simonides were named after the type of event commemorated.²⁵

From this brief and necessarily tentative reconstruction, it is nonetheless clear that the Hellenistic scholar's interest in the arrangement of earlier poets was a corollary of his larger scholarly activity, the classification and editing of texts. An arrangement based on meter, eidē, and perhaps alphabetization is editorial rather than aesthetic; that is, it exists for convenience of classification rather than to enhance meaning or create effect.²⁶ For this reason, the Hellenistic editions of the classical poets are far less significant than the collections the Hellenistic writers made of their own works. Here, for the first time, aesthetic design rather than editorial decision plays the major role in arrangement.

Basic to this type of collection is variatio as the Latins call it, or poikilia in Greek, the exact opposite of the editors' method of grouping poems by categories.²⁷ Diversification is now the key as meter, theme, and even dialect vary, with poems disposed to make the most of this variation. And yet, variatio has been overemphasized by Kroll and others. Ultimately it is a negative approach to arrangement, the avoidance of certain obvious groupings and collocations. But there are also more positive types of design.

In this, as in many other literary innovations, Callimachus seems to have led the way. His Iambs, for example, display most of the techniques available to the Hellenistic poet. There is, first of all, the framing of the collection by a programmatic prologue (1) and epilogue (13). Within this frame there is some grouping on the basis of affinity or contrast. Thus, on metrical grounds, two groups of stichic poems (1-4, 8-13) surround a group of epodes (5-7). On thematic grounds, 7 to 11 are placed together since all offer aetiologies, whereas 11 and 12 are set side by side to juxtapose their forms, an epitaph and a genethliacon. There is even greater formalization in the pairing of alternate poems and the creation of a midpoint break in the collection (i.e. Iamb 6 balances 12 in that both are personal poems at the end of varied half dozens).²⁸ The Aetia shares the same techniques. After the polemical apology (fr. 1 Pf.),²⁹ the second fragment and the last (fr. 112 Pf.) frame the work with references to the Muses and their initiation of Hesiod on Mount Helicon (cf. Theog. 22 ff.). There is also a break in the middle as the fiction of a conversation with the Muses, which provides the narrative structure of Books 1 and 2, seems to have been dropped in Books 3 and 4.³⁰ Finally, a new technique is now added to the poet's repertoire, a sort of cross-reference or connecting of works within a larger oeuvre. Thus, when Callimachus in the last fragment of the Aetia bids farewell to the Muses and announces that he will pass on to those goddesses' more prosaic pasture, Μουsέων πεzὸν … νομόν (fr. 112.9 Pf.), he seems to be looking ahead to the Iambs.³¹

Not all the Hellenistic poets display Callimachus's interest in design. For example, Lawall's suggestion that the first seven Idylls of Theocritus and the eight Mimiamboi of Herodas were arranged as poetry books³² not only lacks any testimonial evidence but also seems contradicted by the manuscripts, the dissension among the three Theocritean families regarding the sequence of Idylls,³³ and the fragments of a ninth mime contained in the British Museum papyrus of Herodas.³⁴ One last category of Hellenistic poetry, however, is relevant to arrangement: the epigram. A revival of interest in this old genre led to the creation of new epigrams, either fictitious dedications and epitaphs or else new noninscriptional forms of erotic and sympotic content.³⁵

The first book of such epigrams may be the Soros mentioned in the scholia to the Iliad (Schol. Ven. A to Il. 11.101). Because the Soros is there said to contain poems by Posidippus, and because Asclepiades, Posidippus, and Hedylus not only occur in a single couplet in Meleager's preface (AP 4.1.45-46 = 3970-71 G.-P.) but also share joint attributions in the Anthology, Reitzenstein concluded that the Soros contained works by all three poets.³⁶ The Homeric scholiast, however, does not indicate anyone other than Posidippus, and such literary collaboration is otherwise unknown from antiquity. Moreover, if the poem identified by Lloyd-Jones as the sphragis of Posidippus was originally attached to the Soros, it is too personal a signature for a joint collection.³⁷ Reitzenstein's hypothesis, then, remains unsubstantiated and the Soros may well have contained just poems by Posidippus. In any case, the title, which means a heap of grain, is very suggestive. First, it may suggest winnowing, i.e. the removal of chaff, and thus affirm Alexandrian standards of polish and labor.³⁸ And second, a heap or pile implies perhaps some degree of diversity or poikilia.³⁹ Thus, it is at least likely that this publication of epigrams shared with the other Hellenistic poetry books a commitment to Alexandrian aesthetic refinement in all areas, including its arrangement.

More relevant to Horace on account of its variety, size, and date is the Garland of Meleager. Although no longer extant, its principles of arrangement can be reconstructed from uncontaminated Meleagrian stretches in the Anthology. Alan Cameron has suggested that the Garland was divided into four books on the basis of subject matter.⁴⁰ This is an editorial mode of arrangement resulting from the special nature of the Garland as an anthology of many works on many subjects by many hands. But the framing by prologue (AP 4.1 = 3926 ff. G.-P.) and epilogue (AP 12.257 = 4722 G.-P.) and the variation worked out in topic, expression, and even author sequence (the rhythmical alternation of poems by the four major Hauptdichter)⁴¹ are aesthetic devices. Meleager signals this when, in his opening poem, he calls his work a πάγaαϱπον ἀοιδάν, “a many-blossomed song” (AP 4.1.1 = 3926 G.-P.). Gow and Page object that “ἀοιδάν is a little odd of a book the miscellaneous contents of which are emphasized by the adjective.”⁴² But the singular noun indicates that the Garland was conceived not as a mere anthology editorially arranged, but rather as an aesthetic whole, a poetry book.

FOUR

From republican Rome, as from Alexandria, much relevant literature is again either mutilated or missing. In the absence of much of Lucilius⁴³ and almost all of Laevius, Horace's unacknowledged predecessor in lyric, it is the Catullan corpus that most closely resembles Horace's Odes in being a miscellany of meters and subjects.⁴⁴ Although the verbal and thematic influence of Catullus on Horace has been amply documented,⁴⁵ no one has seriously explored the possibility that the two collections might be structured along similar lines, much as Horace in his first book of Sermones and Tibullus in his first book of elegies might have looked to Vergil's Bucolics as a model of arrangement.⁴⁶ The investigation, however, is complicated by the difficulty of determining the principles of arrangement employed by Catullus himself. Some sort of authorial organization is indicated by the mention of a libellus in the first poem and by references in Pliny (Epist. 1.16.5)⁴⁷ and Martial (4.14.4, 11.6.16) to a body of poems, the latter suggesting the title Passer. But despite the magisterial authority of Wilamowitz—“Catullus devoted the most careful thought to the arrangement of his book of poems. If there's anyone who can't see that, so much the worse for him”⁴⁸—there is no real agreement concerning the extent of the original libellus and its relation to the present corpus.

Thus, the most obvious principle of arrangement, the tripartite division into polymetra (1-60), long neoterica (61-68, with 65 marking a permanent shift to the elegiac meter), and short, non-neoteric epigrams (69-116),⁴⁹ can afford a parallel to Horace's distribution of his Odes over three books only if several recent scholars are correct in maintaining that the entire Catullan corpus as it stands was arranged by the poet.⁵⁰ But this thesis is rendered problematic by the sad state of the text and by the fact that the total number of lines, over 2,300, far exceeds the known capacity of ancient books⁵¹ and also violates Callimachean strictures against length. Thus, it is easier to follow Clausen in locating the original libellus among the polymetra,⁵² and it is here, and not in the overall disposition of the corpus, that parallels to the order of the Odes can be found.

First of all, the variatio familiar from Hellenistic poetry books is here carried to such an extreme that interlocking and interrupting poems complicate whatever cycles or patterns have been discerned.⁵³ Hellenistic framing is perhaps also operative in that the polymetra nearly open (2-3) and nearly close (58) with poems on Lesbia, initially positive but at the end bitterly invective. Then there are certain recurrent techniques of grouping such as the separation of two related poems by a third to form an A-B-A pattern,⁵⁴ or the use of transitions either on the surface or at some deeper level,⁵⁵ to move the reader from one poem to the next. Although examples of all these techniques can be found in Horace's Odes, the most striking parallelism, perhaps, is the way in which both collections open. Building on Barwick's study of the first Lesbia poems, Charles Segal has shown that Catullus 2-11 are a cycle, tracing the progression of the love affair from its beginning (2-3), through its fruition (5, 7) to its bitter end (8, 11), and integrating it with other central themes.⁵⁶ If these poems stood at the head of the original corpus, they may have influenced Horace to open his book too with a cycle, the Parade Odes, which, as we shall see, introduce most of the meters and many of the themes and personalities prominent in his poetry.

FIVE

To sum up thus far: the origin and development of the pre-Augustan poetry book can only be reconstructed. Not only are we missing potentially relevant works such as the lyrics of Callimachus and Laevius, but also of those works that do survive, some, like the Iambs of Callimachus, are fragmentary, and others, such as the poems of Theocritus, Herodas, and to a lesser degree Catullus, do not reveal with any certainty the extent to which they were arranged by their authors. Nonetheless, a survey of the extant literature indicates that certain common assumptions about the poetry book are inaccurate. It did not, for example, arise in Augustan Rome but in Alexandria. Similarly, to locate its source in the scholarly edition ignores the very great distinction between an editorial response to a classic text and the more aesthetic designs a poet can work out in the publication of his own verse. Of these, variatio, though important, has been overemphasized. There were other, more positive, principles of arrangement, and, though the pre-Augustan poetry book was not coherently structured around any one of them, all were operative.

It has been necessary to dwell at some length on this background, for it has important critical implications for the study of Horace's Odes. First, the extensive prior history of the poetry book renders it very likely that Horace too had some concern for the larger unity and design of his lyric collection. Second, this concern appears to be yet another of his unacknowledged debts to Alexandria,⁵⁷ for a true parallel to Odes 1-3 is to be found not in the other Augustan poetry books which are relatively homogeneous, but in the Hellenistic collections which were formally and thematically varied. Finally, appreciation of this special characteristic of the Odes enables us to avoid the excesses of much previous scholarship, which has either ignored their heterogeneity by searching for a single principle of arrangement or made too much of their heterogeneity by denying any extensive ordering. As befits Horace, the champion of aurea mediocritas, the truth lies somewhere in between.

What we expect, and what we shall actually find, is that Horace shares the Hellenistic poet's familiarity with many methods of arrangement and that he relies on no one of them exclusively. Thus, the reader must be constantly alert to a wide variety of possibilities—to relationships based on contrast as well as similarity, to dynamic movements as well as static patterns, to groupings of contiguous poems and linkages among poems widely separated in the collection. Most importantly, the reader must be attuned to the various and ever-shifting criteria on which these larger designs are founded—not just meter, theme, and addressee, but internal structure, imagery, and that vague but crucial quality, tone.

Although it must be admitted that such an investigation might seem rather subjective, there are controls that can be applied to this as to all other literary study. Obviously, when a number of criteria converge—as, for example, in the Roman Odes—we can be sure of the poet's intent. We must not insist on such certainties, however, for many of the most interesting and suggestive relationships among poems exist at only one of these many levels. But that does not mean that they are fortuitous. Faced with the formidable task of arranging so many and varied poems, Horace would have had to exploit connections, random or planned, at whatever level they appeared.⁵⁸ Fortunately, a second control is provided by sequential reading. As a methodology, the consideration of the poems in their published order not only reproduces their effect on the ancient audience, for whom the physical format of the papyrus roll necessitated such a sequential reading, but it also respects the intention of their author, who is reasonably assumed (in the absence of evidence to the contrary) to have put them in this order in the first place. Finally, and most importantly, we can control subjectivity by pursuing arrangement not as an end in itself but as part of the larger interpretative enterprise. In other words, a perception of design, in order to be plausible and meaningful, must accord with an interpretation of the individual odes that is self-consistent. Recent studies of other lyric traditions—of the modern poetic sequence by M. L. Rosenthal and Sally Gall, or of Keats's odes by Helen Vendler—suggest the fruitfulness of such an approach, which places the individual poem within the larger context provided by the other poems.⁵⁹ Something similar is needed for Horace's Odes—not another study of arrangement but a new critical reading of the poetry which recognizes that the ensemble is itself one important context for understanding and appreciation.

Notes

1. This chapter derives from my article in Arethusa 13 (1980). I have taken this opportunity not only to update it but also to discuss methodology more fully and to revise (esp. in the notes) some of my earlier views.

2. Von Christ, “Uber die Verskunsi,” pp. 1-44; Kiessling, “Horatius I,” pp. 48-75.

3. A roughly chronological survey of such theories with some refutation was provided by Raiz, “Die Frage nach der Anordnung,” pp. 43-56; see also Simon, Zur Anordnung der Oden; Häusser, Review of Raiz and Simon, pp. 362-66. When meter was abandoned as the sole criterion for order in favor of verbal or thematic links, the results were equally fanciful; see Simon, ibid.; Verrall, Studies Literary and Historical, pp. 90-120 and passim; Belling, Studien über die Liederbücher des Horaz. Many of these studies, however, at least had the advantage of noting verbal echoes, for which see now Huber, Wortwiederholungen in den Oden des Horaz.

4. See Draheim, “Die Anordnung der Gedichte,” col. 1268.

5. Duckworth, Structural Patterns and Proportions; Dalzell, Review of Duckworth, pp. 314-316. For general remarks on numerology see Wilkinson, Georgics of Virgil, pp. 316-22.

6. Segal, “Ancient Texts,” p. 13. The entire article is interesting, and on the larger questions it raises see also Wilkinson, “Ancient Literature,” pp. 13-26.

7. A remarkable exception is Dettmer, Horace: A Study in Structure. In a complicated exposition illustrated over 150 charts and diagrams, Dettmer attempts to show all 88 odes are arranged by theme in a ring composition which is occasionally interrupted at parallel intervals by smaller groups that may not be themselves symmetrically arranged. Utilizing a complex terminology that distinguishes between “outer-ring” and “inner-ring” poems, and between “structural” and “non-structural” ones, Dettmer searches for verbal, numerological, and metrical “clues” that the poet furnished to point up his complicated “thematic concatenation.” Such ring composition, finally, is alleged to be the “predominant organizational pattern” not just of Odes1-3 but of all the Augustan poetry books as well as of the Catullan corpus. A strange throwback, this approach suffers in general from the same deficiencies as the earlier scholarship discussed above, namely problems of methodology, probability, and conception. For a more specific critique of this, see p. 210, n. 2, below; see also the review of Dettmer by Syndikus, pp.11-15. Because our ways of reading are so fundamentally opposed, I have for the most part chosen to engage in a dialogue with Horace rather than with Dettmer, whose argument the reader may evaluate independently. I have, however, cited Dettmer on points of agreement or information.

8. Apart from Dettmer, Horace: A Study in Structure, the only other comprehensive and systematic study of the collection was published sixty years ago by Port who, more judiciously, espoused no monolithic theory (“Die Anordnung,” pp. 280-308, 427-68). For a briefer account of the Augustan poetry book that omits Horace and stresses ring composition, see Michelfeit, “Das augusteische Gedichtbuch,” pp..347-70. Other, briefer overviews include Stemplinger, “Horatius,” col. 2372; Schanz and Hosius, Geschichte der römischen Literatur 2:126-27; Wickham, Works of Horace 1:25-30; Perret, Horace, pp. 80-84; M. Schmidt, “Die Anordnung der Oden,” pp. 207-16; Giardina, “Sulla struttura delle odi,” pp. 44-55; Silk, “On Theme and Design,” pp. 47-54; Collinge, Structure of Horace's Odes, pp. 36-55 (with the review by Ludwig, pp. 171-77). I regret that it has not been possible for me to obtain the thesis of Fontaine, Enchaînement et groupement, which is mentioned by Collinge, ibid., p. 36 n. 2. More common than overviews, finally, are useful observations on specific groupings and collocations, especially in the introductions to individual poems by Kiessling and Heinze, Oden und Epoden; in random remarks in Wili's Horaz und die augusteische Kultur; and in the periodical literature. All of these will be cited where appropriate in the notes below.

9. Wilkinson, Horace and His Lyric Poetry, p. 15; Dawson, “The Iambi of Callimachus,” p. 140.

10. Nisbet and Hubbard, Horace: Odes, Book I, p. xxiii; see also ibid., p. xxiv: “It is only too easy to imagine some subtle principle either of similarity or difference in every juxtaposition, not to mention more complicated sequences and cycles. Most of these suggestions seem completely fanciful, and equally ingenious reasons could be adduced to justify any arrangement.” The same authors’ later commentary on Book 2 is somewhat more moderate: “The book contains 20 odes … and as usual is arranged with a measure of design. … The organization is most apparent in the first half, particularly in the formal aspect of metre; otherwise Horace may have noticed some superficial resemblances, which may or may not now be divined, but even then he was likely to have been left with a few poems that did not fit any scheme precisely. After all he was not composing a cycle of odes, but in ancient terminology ‘arranging a garland’” (Horace: Odes, Book II, pp. 5-6).

11. G. Williams, Third Book of Horace's Odes, p. 23; see also his Horace, p. 35: “The arrangement of the odes has attracted increasing attention—to little purpose.” For the inadequacies of the treatment of arrangement generally in Williams's Tradition and Originality, see the review by Otis, pp. 316-30.

12. It should perhaps be noted at this point that there is no good reason to doubt that the order of the Odes as we have it is the poet's own. Whereas the position of the lyrics with relation to Horace's other works varies in the manuscripts, there is unanimity regarding the placement of individual lyrics, with only occasional dissension about poem divisions. Testimonia concur in this numeration. Finally, Horace may himself allude to a presentation copy of this work in Epistle 1.13, where the description of the contents as carmina (17) and the references to plural volumes (volumina, 2; libellis, 4; chartae, 6) would suit Odes 1-3 very well.

13. A convenient survey of theories on the Bucolics may be found in Rudd, Lines of Enquiry, pp. 119-44; to this should now be added Van Sickle, Design of Virgil's Bucolics; Dettmer, Horace: A Study in Structure, pp. 2-14. For Tibullus, see the listing in Harrauer, Bibliography to Tibullus, pp. 45-46; to this should be added Powell, “The Ordering of Tibullus Book I,” pp. 107-112; Leach; “Poetics and Poetic Design,” pp. 79-96; Dettmer, ibid., pp. 14-22; Mutschler, Die poetische Kunst Tibulls, pp. 157-200, 279-302. For Propertius, studies of arrangement are complicated by the textual transmission—see Ullman, “The Book Division of Propertius,” pp. 45-52; and, more recently, the novel suggestion of Butrica, “The Propertian Corpus,” p. 6. The extensive literature is listed in Harrauer, Bibliography to Propertius, pp. 101-2; to this should be added Barsby, “Composition and Publication,” pp. 128-37; King, “Propertius' Programmatic Poetry,” pp. 108-24; Putnam, “Propertius' Third Book,” pp. 97-113; Dettmer, ibid., pp. 22-32; Hutchinson, “Propertius and the Unity of the Book,” pp. 99-106; Stahl, Propertius, passim.

14. On the Iambi see Carrubba, Epodes of Horace. On the Sermones see Ludwig, “Komposition der beiden Satirenbücher,” pp. 304-25; Jensen, “Secret Art,” pp. 208-15; Rambaux, “Composition d'ensemble du livre I,” pp. 179-204; Dettmer, Horace: A Study of Structure, pp. 32-35; and the series of lengthy articles by van Rooy under the general title “Arrangement and Structure of Satires,” listed in the bibliography.

15. Although we are concerned with the precedents available to Horace, it should be noted that interest in arrangement persists in later books: see, for example, Ludwig, “Anordnung des vierten Horazischen Odenbuches,” pp. 1-10; Putnam, Artifices of Eternity; McGann, Studies in Horace's First Book of Epistles, pp. 33-87; Buchheit, Studien zur Corpus Priapeorum, pp. 43-53; Buchheit, Studien zur Corpus Priapeorum, pp. 43-53; Loercher, Aufbau der drei Bücher von Ovids Amores; Evans, Publica Carmina, passim; Jacobson, Ovid's Heroides, passim; Newmyer, Silvae of Statius, pp. 45-58, 122-30. Not unrelated to the poetry book is the disposition of episodes in the Kollektivgedicht: see, for example, Ludwig, Strucktur und Einheit der Metamorphosen Ovids; Otis, Ovid as an Epic Poet, pp. 45-90; Galinsky, Ovid's Metamorphoses, pp. 79-109.

16. E.g. Otis, Review of G. Williams's Tradition and Originality, p. 326: “So far as we can tell, this sort of collection or poetical context was Roman and not Greek. There is, for example, no indication that Hellenistic poetry books were constructed in this way”; Otis, “Propertius' Single Book,” P. 38 n. 8; Ross, Backgrounds to Augustan Poetry, p. 49, who seems to suggest that Gallus came up with the idea.

17. For a general account of the transition in Greek culture from the oral to the written word, see Havelock, Literate Revolution in Greece; Pfeiffer, History of Classical Scholarship, pp. 16 ff.; Davison, From Archilochus to Pindar, pp. 86-128. It should be noted that, even after the rise of the book, reading was largely done aloud: see Balogh, “Voces Paginarum,” pp. 84-109, 202-40, with supplements by Hendrickson, “Ancient Reading,” pp. 182-96, and correctives by Knox, “Silent Reading,” pp. 421-35.

18. Kroll, Studien zum Verständnis, pp. 225-46.

19. For a description of this edition see Wilamowitz-Moellendorff, Textgeschichte der griechischen Lyriker. That Horace used it is likely because his placement of C. 1.2, which ends with Mercury as political savior, seems to recall the Alcaeus edition in which the second position was also occupied by Hermes: see below, pp. 21-22, with p. 187, n. 27. It should be noted, however, that not every poet's text might have achieved such fixity in antiquity. For instance, the common assumption that the Ambrosian list of Pindar's seventeen books reproduces the order of a standard Alexandrian edition is called into question by the existence of other dissenting lists: see Race, “POxy 2438 and the Order of Pindar's Works,” p. 106.

20. On the remote possibility that the division of Sappho's work is older than the Alexandrian edition because the sticho-metrical subscript of POxy 1231 is in Attic notation rather than the Milesian in use in Egypt at the time, see Lobel, Σαπφοῦς Mἔλη, pp. xii-xvii.

21. See Page, Sappho and Alcaeus, p. 126, for the principles behind the disposition of the Epithalamians.

22. See Lobel, Σαπφοῦς Mἔλη, pp. xv-xvi; see also Daly, Contributions to a History of Alphabetization, p. 23 n. 3.

23. See Daly, ibid., p. 23, with literature there cited.

24. This does not mean that other factors could not also have been operative. Olympian 1, for example, may owe its prominent placement to its popularity in antiquity (Lucian, Gallus 7), the centrality of its myth to the Olympic Games, and the importance of its occasion. But an editor's occasional appreciation of such niceties is not the same as a poet's truly aesthetic design.

25. There is no evidence to suggest that the distinction between choral lyric and monody (Plato, Laws 6.764d-e) or between monostrophic and triadic systems (Schol. Epimetr. Pind. III, p. 310.27 Dr.) played any part in arrangement: see Harvey, “Classification of Greek Lyric Poetry,” pp. 157-75.

26. This conclusion seems valid not only for the lyric poets whom we have been considering, but also for the other early writers from whose works one occasionally receives the impression of organization. For instance, the elegiacs attributed to Theognis are clearly by several hands, and their collection and arrangement was the work of a later editor: see M. L. West, Studies in Greek Elegy and Iambos, pp. 40-64. For Mimnermus's Nanno also there is no evidence of aesthetic design because the fragments are very varied in content, and the idea that the poet's love for the girl provided a thematic unity may well be a Hellenistic invention on analogy with Antimachus's Lyde: see West, ibid., pp. 74-76. Finally, though the Attic scolia (Ath. 15.693-5) may have constituted a book from which uninventive guests could select a poem to recite at the symposium (see Wilamowitz-MoellendorffAristoteles und Athen, 2:316-22), the variety of social and political constituencies represented by the poems preludes any real aesthetic unity, and such grouping as has been observed seems editorial: see Bowra, Greek Lyric Poetry, pp. 375-76; Cuartero, “Estudios sobre el escolio ático,” pp. 5-38.

27. For a general discussion of poikilia/variatio as a larger feature of all aspects of composition, see Deubner, “Ein Stilprinzip hellenistischer Dichtkunst,” pp. 361-78; Schulze, “Über das Princip der Variatio,” pp. 857-78. It should perhaps be noted here that structural studies often use familiar technical terms in new or more specific meanings. Thus, voratio is used to refer to the diversification of poems that are placed together and not just to the practice of “theme and variation.” Similarly, “ring composition,” which usually refers to the archaic Greek style whereby the beginning and the end of a poem are temporally linked, can also refer to the disposition of poems in concentric frames.

28. See Dawson, “The Iambi of Callimachus,” pp. 1-168; Clayman, Callimachus' Iambi.

29. It is commonly assumed with little real authority that this prologue headed a second, collected edition of Callimuchus's Aetia and other works: see Pfeiffer, “Ein neues Altersgedicht,” pp. 339-41, and his Callimachus, 2: xxxvi-vii. For an alternative view, however, see Eichgrün, Kallimachos und Apollonios Rhodios, pp. 64 ff., where the idea of consequential publication is rejected. The entire question is reexamined and Pfeiffer's theory is rejected by Alan Cameron in a forthcoming monograph, Callimachus and His Critics.

30. According to the new reconstruction of the Aetia by Parsons, “Callimachus: Victoria Berenices,” pp. 1-50, the last two books were also framed, by poems in honor of Berenice.

31. This does not mean that there was a collected edition in which the Iambi followed the Aetia as they do in POxy 1011 and in the Milan Diegesis: see above n. 29. On the meaning of πεzὸν … νομόν, see Van Sickle, Design of Virgil's Bucolics, pp. 104-5.

32. Lawall, Theocritus' Coan Pastorals. While Lawall's arguments do not prove the existence of poetry books by Theocritus or Herodas, the interconnections he observes among poems are very illuminating and have great interpretive value in other respects.

33. For an account of the MSS see Gow, Theorcitus, 1: xxx ff. A detailed critique of the presuppositions underlying Lawall's interpretation is provided in the review by Giangrande, pp. 170-73. For an alternative sense in which the poems constitute a libellus, see Segal, “Thematic Coherence and Levels of Style,”

    pp. 35-68, reprinted in his Poetry and Myth, pp. 176-209. On the relationship between the collection of the corpus and ancient awareness of genre, see Van Sickle, “Theocritus and the Development of the Conception of Bucolic Genre,”pp. 18-44; and, for more general definition of the genre, see Halperin, Before Pastoral.

34. See Cunningham, Herodas: Mimiamboi, p. 160; on the possibility of pairing see Cunningham, “Herodas 6 and 7,” pp. 32-35.

35. On the history of the epigram see Fraser, Ptolemaic Alexandria, 1: 553-617 and 2:791-869, with literature abundantly cited. For a list of papyrus anthologies of epigrams from the Hellenistic period see Pack, Greek and Latin Literary Texts, p. 92. To this should now be added PKöln 204 which contains the remains of six epigrams by Mnasalkes which are arranged into pairs by theme. Dating to the middle of the second century b.c., this is the earliest extant collected edition of an epigrammatist. For more detailed discussion of this and other collections, see Cameron, The Greek Anthology.

36. Reitzenstein, Epigramm und Skolion, pp. 94 ff.

37. Lloyed-Jones, “The Seal of Posidippus,” pp. 75-99.

38. See Webster, Hellenistic Poetry and Art, p. 45, who argues for this meaning by citing as a parallel the use of the word Theocritus 7.155, a literary harvest, see also Gow, Theocritus, 2:169. It should at least be noted, however, that, while the notion of winnowing can be supplied from the Theocritean context, the word sωρός usually conveys only the undifferentiated quality of a heap.

39. A third-century b.c. papyrus appears to be the beginning of a collection of Posidippus (the Soros?): see Lasserre, “Aux origines de l'Anthologie, I,” pp. 222-47. The MS bears the title Symmeikta Epigrammata, which may carry literary critical implications parallel to those detected in the title Soros; see Webster, Hellenistic Poetry and Art, p. 46.: “Symmeikta seems from its uses elsewhere to mean ‘mixed’ rather than ‘blended’; it emphasizes the variety of the contents rather than the choice of each epigram to suit its neighbor.”

40. Cameron, “Garlands of Meleager and Philip,” pp. 323-49.

41. These principles were discovered by Radinger, Meleagros von Gadara, pp. 88-107, and Weisshäupl, “Zur den Quellen der Anthologia,” pp. 184-88.

42. Gow and Page, The Greek Anthology, 2:597.

43. On metrical voratio in Lucilius see Puelma-Piwonka, Lucilius and Kallimachos, pp. 364-67.

44. The publication of prose letters would seem initially to offer another parallel. Pliny, for example, tells us (Epis. 1.1) that his letters are not chronologically arranged but occur as they came into his hands, which suggests conscious voratio and which calls to mind his insistence on variety in his lyric poetry (Epis. 4.14.3, 8.21.4). Futhermore, though elaborate schemes are not convincing, there is evidence that “[in] each book Pliny sought to include a representative collection of the major themes, but he did not bind himself to an arithmetical proportion” (Sherwin-White, Letters of Pliny, p. 50; and see the full discussion of distribution on pp. 20-56). But Pliny is much later than Horace. The earlier letters of Cicero are not relevant either, for they seem to have been published posthumously and their organization displays no real aesthetic design: see Shackleton Bailey, Cicero's Letters to Atticus, 1:59-76, and his Cicero: Epistulae ad Familiares, 1:2-24.

45. See, for example, Ferguson, “Catullus and Horace,” pp. 1-19; van der Paardt, “Catullus en Horatius,” pp. 287-96; Lee, “Catullus in the Odes of Horace,” pp. 33-48.

46. It is interesting that Vergil's Bucolics, Horace's Sermones 1, and Tibullus 1 all contain ten poems, and that Horace's Iambi divide after the first ten poems, which are all in the same meter. In this concern for the number ten, as in other respects, Vergil may have influenced his fellow poets: see van Rooy, “‘Imitatio’ of Vergil,” pp. 69-88; Leach, “Vergil, Horace, Tibullus: Three Collections of Ten,” pp. 79-105.

47. Pliny compares a dead friend, Pompeius, to Catullus who is said to have placed harsh poems among light ones (inseit sane … mollibus levibusque duriusculos quosdam). Whether or not this accurately describes Catullus's methods and although it does not indicated which poems are involved, it does imply some sort of organized collection.

48. Wilamowitz-Moellendorff, Sappho and Simonides, p. 292, translated by Quinn, Catullus: An Interpretation, p. 11.

49. On the distinction between neoteric and non-neoteric see Ross, Style and Tradition in Catullus.

50. See, for example, E. A. Schmidt, “Catullus Anordnung,” pp. 215-42; Wiseman, Catullus: An Interpretation; Skinner, Catullus' Passer; Most, “On the Arrangement of Catullus' Carmina Maiora,” pp. 109-25. Those who support the view that the entire corpus was arranged by the poet may find support in Macleod, “Catullus 116,” pp. 304-9, reprinted in his Collected Essays, pp. 181-86, where the last poem is treated as suitable Callimachean epilogue; see also Van Sickle, “Poetics of Opening and Closure,” pp. 65-75.

51. Of course, the papyrus roll and the literary book are not to be equated: the physical capacity of the former could exceed the length arbitrarily fixed for the latter. For full discussion see Van Sickle, “The Book Roll,” pp. 5-42, with literature there cited.

52. See Clausen, “Catulli Veronensis Liber,” pp. 37-43.

53. Ellis, Commentary on Catullus, pp. xlviii-ix, uses this very fact to cast doubt on the discovery of conscious design in Catullus. On the other hand, Quinn, Catullus: An Interpretation, p. 16, draws a very different conclusion: “Indeed, I think it likely that the illusion of a work unfinished, of a collection made up in large part of fragments of poems which are too short, too abrupt in the way they begin or end to be felt as complete poems, but which plainly belong together—snatches of conversation, as it were, in an easily imagined context—was one of the principles adopted by Catullus in putting his collection together. Rather as a modern painter or sculptor leaves his painting or bust at a stage where it is, by normal standards, unfinished (the likeness just emerging from the marble, say) because it is finished for him; or because to finish it off means robbing it of life.”

54. E.g. 5/7 (basia), 21/23 and 24/26 (Furius and Aurelius), 37/39 (Egnatius). Such separated pairs also occur outside the polymetra (e.g. the famous Lesbia poems, 70/72) where they may be taken as traces of the poet's own conscious design, now lost, or as the work of a perceptive editor.

55. An example of a surface glide occurs early in the polymetra: poem 9, on Veranius's return from Spain as a member of Piso's cohort, leads into 10 which alludes to Catullus's own experiences in the Bithynian cohort of Memmius, and thus introduces 11 which opens with a travelogue (see Kinsey, Catullus 11, pp. 537-44; Ferrero, Interpretazione di Catullo, p. 221). An example of a less superficial sequence occurs at the end of the polymetra: poem 50, an account of the otiosi Catullus and Calvus, is followed by 51, the last strophe of which condemns otium, and thus prepares for 52-54, a return to the negotium of politics (See Wiseman, Catullan Questions, p. 15). As was the case with A-B-A patterns, sequences of contiguous poems also occur after the polymetra (e.g. 88-91, on Gellius, the longest run of contiguous poems on a single individual) and may be viewed either as the work of an editor or (in some cases at least) as traces of a now lost original design by the poet. For useful lists of such clusters throughout the corpus see Ellis, Commentary on Catullus, pp. xlv-l; Kroll, C. Valerius Catullus, p. x.

56. Barwick, “Zyklen bei Martial,” pp. 284-318; Segal, “Order of Catullus, Poems 2-11,” pp. 305-21.

57. The Alexandrian influence was first explored in great detail by Pasquali, Orazio lirico, pp. 141-641.

58. The question naturally arises at this point of whether the Odes were written to occupy their positions in the collection or whether placement was a subsequent activity. For certain collections like Vergil's Bucollics and the other homogeneous books of ten, it is possible that some elements of overall design preexisted the composition of individual poems. The Odes, however, are so numerous and varied that it is probably safer to assume that, in most cases, arrangement came later. Thus, the poems of even the most cohesive cycle, the Roman Odes, give evidence of having been composed at different times. That disposition followed composition, however, does not mean that it was an afterthought, an unrelated or less important creative activity (as we shall see). Also, we need not exclude the possibility that at least a few poems might have been reworked to suit their context in the collection, and that a few might even have been written with an eye on their placement. This hypothesis certainly suits C. 1.1 and 3.30 and perhaps also pairs like C. 1.31 and 32, which read as if they were a single poem (see the discussion on pp. 69-72). Such pairs certainly exist in Ovid (e.g. Amores 1.11-12) and Propertius (e.g. 3.4-5). On the whole question see now Hutchinson, “Propertius and the Unity of the Book,” pp. 99-106.

59. Rosenthal and Gall, Modern Poetic Sequence; Vendler, Odes of John Keats.