The Posidippus Papyrus Bookroll and Reader

william johnson

The Posidippus papyrus is of intense interest in a number of specific respects: for the sudden access to the poetic character of an author whose name was well known to us, but whose work was almost entirely lost; for the poetry itself, which ranges in quality but has many interesting pieces; for illuminating the exact content of an early epigram collection; and, lastly, but perhaps most importantly, for what it may tell us about the way that a poetry book was used and put together at this, the time of the beginnings of such poetic collections.

I wish here to pursue that final set of questions: how an ancient poetry book was put together, and how it was used. I will concentrate on what the Milan papyrus tells us as an artefact. That is, what conclusions can we draw from the ways in which this bookroll was constructed and copied, and what we can say about its history as an object prior to its discard and reuse as cartonnage in the early second century BC? I organize my discussion around two questions: (i) What seemed usual or unusual about this bookroll when the Ptolemaic reader picked it up in his or her hand? (2) What can the papyrus tell us about its use by readers over time, and was there anything exceptional about that use?

The huge majority of extant papyri are from the Roman era. The corollary is that there are relatively few literary papyri from the Ptolemaic period, and, as it happens, most of these are in poor, fragmentary condition.1 Even expert papyrologists are, consequently, far less acquainted with what is typical of a Ptolemaic bookroll such as the Milan papyrus, and it will be useful to summarize what about the artefactual details of the papyrus is routine, and what is not.

In the most general terms the look and feel of this bookroll are unexceptional. By 'look and feel', I refer in particular to the way in

1 The Leuven Database of Ancient Books ( lists about 300 bookroll fragments surviving in the period from iv bc to the crossover to i bc (i.e. including papyri dated ii/i be); 2,100 bookrolls from ad ii—iii, including crossovers (i.e. papyri dated i/ii ad, iii/iv ad).

which the rather wide columns are tightly spaced so as to flow almost one into the next—that is, in technical terms, the intercolumn is very narrow, as narrow in fact as logistics allow (PL i). These columns are flanked at top and bottom by, again, quite narrow white space; the upper margin in particular is very narrow. The overall effect is of an extended and somewhat oversized rectangular block of nearly continuous writing that dominates the run of the bookroll. That look, which would be unusual in a well-written Roman-era bookroll, is in fact easy to parallel among Ptolemaic verse texts. I think it fair to say, in fact, that this is one characteristic look for verse texts of the third and second centuries BC. The rather typical example in PI. 2 (P. Sorb. inv. 2272 + 72, a well-known late third-century-BC cartonnage from el-Ghorab containing Menander's Sikyonios) is slightly shorter in the column (with a height of about 13—14 cm rather than 16 cm), but otherwise noticeably similar in look, particularly in the tendency for longer verse lines almost to intersect with the next column. The height of the column of the Milan papyrus (c. 16 cm), and the height of the roll itself (c. 20 cm), are reasonably typical for a Ptolemaic verse text of this type. The column width, though somewhat narrow in comparison with other Ptolemaic verse texts (on which more below), is not far out of the expected range.2

The fact that the overall look and feel are typical of the era does not, however, mean that in matters of detail the bookroll is entirely run-of-the-mill. The section headings are unusual in several respects, most remarkably in their mere presence.3 In formal terms, the most striking feature is simply the script itself, which is unusually tiny, especially for a Ptolemaic book hand. PI. 2, in which the Menander text and the Posidippus text are set side by side at the same scale, conveys an idea of how tiny the script is. The small script accounts directly for the somewhat more narrow than usual column width, and the large number of lines contained within what is, in physical dimensions, a normal height of column. For this column height, one expects roughly twenty-five lines; the Milan papyrus, by con-

2 A. Blanchard, 'Les papyrus littéraires grecs extraits de cartonnages'; W. A. Johnson, Bookrolls and Scribes in Oxyrhynchus (Toronto, 2004), §3.3 (column height), §3.6 (roll height), §3.2.4 (column widths).

3 The subject headings are 'more specific than the very general divisions familiar from AP and first attested for Agathias' Cycle': Parsons, 'Callimachus and the Hellenistic Epigram', 121—2. The only close parallel is P. Strassb. WG 2340 (iii Be) = Pack2 1749, which contains a heading, ttoXçiâiko., above very fragmentary elegiacs (cited in Bastianini, 'II rotolo degli epi-grammi di Posidippo', 115); cf. K. Gutz wilier, Introduction n. 15. Of different type are explanatory titles for individual epigrams, found in P. Oxy. LIV 3725, LXVI 4501—2 (i ad, ascribed by Parsons to the same hand), and SH 985 (iii Be); and similarly for titles transmitted in the manuscripts of Theocritus.

trast, has forty lines in the column. The overall effect, as we see, are columns considerably less squat in appearance than in the usual Ptolemaic bookroll, an effect which is entirely the consequence of the size of script.

The copy was made by someone trained in the craft, not necessarily a professional scribe but at least someone who knew how to produce a book-like product. The script, though not entirely regular and with a tendency to lapse into a faster mode of writing, is assured and easy, the lines reasonably even and regularly spaced. The column height remains stable within a variation of very few millimetres over the run of sixteen columns. The width from column to column is likewise consistent within tight limits, obviously the product of routine measurement. The measurements from column to column are within 2 mm of 9.5 cm with the exceptions of cols. II, IX, and X; in each of these columns the first line itself exceeds 9.5 cm, necessitating a slightly wider column of 10.1—10.2 cm (PI. 3). This is what I mean by intercolumns that are as narrow as logistically possible. The tendency towards strict parameters for the physical measurement of column height and the width from column to column is characteristic in all periods for scribal production of bookrolls (as opposed to less formal productions).4

Still, this was not a very fine bookroll production. The hand is workaday. The tiny script gives the column a cramped look, as though the scribe were trying to accommodate as many lines as possible within the space. The margins, especially the top margin, are about as small as margins can be, even in the context of a Ptolemaic roll (PI. 1). Moreover, the papyrus itself is of middling, not high quality, which is unusual for a bookroll. It is rare to have a bookroll in a well-trained hand that is not written on what by the Elder Pliny's time would be called an optima charta. The width of the sheets in this roll (19.4—19.7 cm) qualifies the papyrus for two grades below the optima (what Pliny later called the 'Fannian'). In my own survey of papyrus sheet sizes in literary bookrolls, only two (one Ptolemaic) use papyrus of this low a grade;5 Alain Blanchard has reported a couple of additional Ptolemaic examples in this lower category;6 but clearly the mediocre quality is unusual for a bookroll. Because of physical damage over time, it is very difficult to judge the surface quality of ancient papyrus, especially ancient papyrus from cartonnage. Even

5 Ibid. §3.1.1. On the question of the size of papyrus sheets as indicator of papyrus quality, see Plin. NH 13.74—8 and Johnson, 'Pliny the Elder and Standardized Roll Heights'.

6 Blanchard, Papyrus littéraires grecs, 21.

so, the glue joins (KoXXrjceic) seem noticeably ragged in places (PL 4a). Col. XIV preserves an interesting example of a koXXtjclc where the top layer is bent over, creating an awkward surface; and yet the papyrus was written upon without smoothing or repair (PI. 4b). This sort of irregularity is to be expected to some extent in hand manufacture, but there seems good cause to conclude that the papyrus surface itself was not meticulously prepared, and not of the quality usual for a bookroll production. The workaday hand and cramped appearance seem to accord with this observation.

When we try to imagine what impression the manuscript gave to a Ptolemaic owner as he or she picked it up, it would be, then, that the bookroll was not the work of an amateur, but rather utilitarian: certainly far from a showpiece designed to impress elite friends with the quality of the book-as-object. Ancient books were normally custom manufacture, meaning that the product resulted from a negotiation between buyer and scribe as to the quality of paper used, the beauty of the writing employed (that is, how rapid or slow the writing, which in turn determined the rate of payment per line).7 This bookroll summons to our imagination a roll of an unimpressive kind, created with a view perhaps to content rather than to display. The very fact that the bookroll seems to have been created with utilitarian purpose makes striking the (otherwise banal) observation that it appears to have been not at all heavily used. Or at least we do not find evidence of the sort of lectional marks so common on bookrolls used in the usual utilitarian reading contexts: that is, in the schoolroom, or in reading circles where the text was used as entertainment, or in circles where 'scholars'—serious readers—seem to have been studying a text. We find no marks indicating word grouping or division, no apostrophes to mark elision, no dots to mark breath pauses, no signs of Siopdujcic or of variants, or of correction to even the most obvious scribal errors, aside from a few corrections made by the original scribe in the process of the original copying and a very small number of later corrections and additions, mostly in reaction to damage to the roll. Indeed one of the most conspicuous features of this roll is the number of uncorrected errors, which range from the most trivial (and thus perhaps unimportant since easy enough for a reader or lector to correct on the fly) to errors serious enough to intrude materially upon the sense of certain epigrams.8

7 Edictum Diocletiani depretiis rerum venalium, col. vii, 11. 39-41; cf. E. G. Turner and P. J. Parsons, Greek Manuscripts of the Ancient World (London, 1987), 1-4.

8 Many small corrections are made in scribendo by the scribe, but there is no evidence of systematic diopdojcic; the very few corrections by a reader are limited to cols. IV and V, except

The roll did, however, grow old, whether in the first owner's hands or in another's; and it is in its latter years that the life story of this bookroll becomes particularly intriguing. At some point in its life, the roll was damaged. We can see moisture damage and overtracing quite clearly at a couple of places (PI. 5, highlighted area).9 The editors imagine a spill of water, but moisture damage from long storage without adequate exercise and airing of the roll is at least as likely. Whatever the cause, it is clear that the beginning of the roll deteriorated to the point that the front edge had to be refashioned. I will linger for a moment on this front, repaired edge, since it has, I think, much to tell us not only about the later history of the roll's life, but also about its use.

The editors conclude from three related pieces of evidence that the protokollon was a later addition, not part of the original manufacture—that is, that the current protokollon is the result of a repair.10 First, the left sheet of the manufactured roll is only 11.3 cm wide in a roll in which the other sheets are consistently 19.5 cm (PI. i).11 Secondly, no blank space between the protokollon and the left edge of the manufactured roll is found; the writing begins directly to the right of the protokollon, contrary to custom. It looks, in short, as if the blank space that normally precedes the first column of writing (the agraphon) was cut away. Thirdly, the protokollon is affixed in an unusual fashion. In Greek bookrolls, joins were made so that the left sheet overlaps the right sheet. The rule seems to be as true of the protokollon as of any other sheet in a bookroll. In this case, however, the protokollon is attached so that the sheet at the right overlaps the protokollon. This arrangement is necessary, since the protokollon otherwise would have obscured the first letters of the column (the overlap is usually about 2 cm wide). The editors conclude that this protokollon cannot belong to the original manufacture. Their conclusion is not in every respect unassailable, but the physical details do seem to fit best a scenario in which the torn original roll was cut at the very left of 'col. I', and a new protokollon affixed to repair the front edge of the bookroll.12

a single correction, by a different reader, in col. XI: BG, p. 15. On instances of reaction to damage, see just below.

9 So col. XI, 11. 12—15; the correction at col. V, 1. 1 has a similar appearance. See BG, p. 13.

10 Ibid.

11 Variation, even cheating, is normal for the specified size of a sheet (/cdAAi^a) in manufactured rolls, but one does not expect or find such wide variation in the very first icoWyj/m: Johnson, Bookrolls, §3.1.1.

12 Damage at the front of the roll, which received the most wear and tear, was common; cf. e.g. P. Oxy. 223 for damage and repair similar to the Milan papyrus.

The apparent fact that the protokollon is a repair job may fit in with another detail, one not remarked by the editors. The first line of the first column appears to be in a different ink. The difference in ink is particularly apparent in the infrared image (PI. 6, right), but also apparent in the colour image (left). I say 'appears to be' and 'apparent' both because a variation in ink flow was natural to ancient writing and because of the problems of variation of appearance in ancient papyri, especially papyri surviving from cartonnage. Still, the ink does appear noticeably distinct, and though I am not so rash as to affirm a difference of hand on the basis of so few letters, it is perhaps worth remark that the letters seem in part differently formed (PI. 7a).11 What is tantalizing about the apparent difference in ink is the placement of the line itself, which appears to be not in fact at the expected top of the column, but above the top of the column. I want, in short, to raise the possibility that the first line is a later addition. As observed earlier, the papyrus has a narrow, but quite consistent top margin (of 1.2—1.4 cm) that, despite repeated breaks, survives all along the top of the sixteen surviving columns of the roll (PI. 1). The same line of margin appears to be in place above the first column, but such that 1. 2, and not 1. 3, is located at the expected extent (1.4 cm from the top edge: PI. 8). I am acutely aware of the dangers of such an argument—my own book provides the best available documentation for the capricious breakage of margins in ancient papyri14—but the coincidence of apparent difference in ink, script, and extent of margin for the top line of col. I is a conjunction at least worth bringing to notice.15 Though by no means certain, the real possibility exists that the first line of poetry as we have it was inserted later to fill out that first poem, presumably as part of the repair that occasioned the cutting and refashioning of the front edge of the roll.

One final detail concerning the top of the first column. In the top margin, the editors read a dotted kappa and dotted alpha. These traces are the basis for their supposition that the heading to this section appeared above the first column (restored as [At0t]«:a).16 In

13 e.g. oval back-leaning omicron extending over the full bilinear height of the script; hastae of iota and mu straight and squared off at the ends, in contrast to the bent hasta with rounded, globular ends characteristic of the main scribal hand. 14 Johnson, Bookrolls, §3.5.

15 Overlaying the adjacent column ('col. II') at its least distorted point gives the viewer a strong impression that the top line of col. I is a later addition. (This is not, however, fully consistent among all the columns, since the column height varies over the course of the roll by a few millimetres.)

16 BG, p. 13 give the section heading as one of the primary reasons for believing the first column to be the original col. I. At 13 n. 23, they list examples for titles in the top margin of the first column, but these are very few and mostly not unambiguous (for details, see Bastianini's article there cited).

PL 9, I highlight the area where the editors describe these letters to be. I have not seen the papyrus in person, and I think the editors, Bastianini and Gallazzi, have been remarkably and laudably careful; but the section heading has to be judged an uncertain reading by any standard. Stray ink is an endemic problem in papyri recovered from cartonnage.

I have dwelt on the first column because the details of that column dovetail in ways that are suggestive for our view of the contents and use of this bookroll. Before trying to put the pieces together, however, we need to examine a final exceptional feature, the stichometry. The stichometry was added after the text, but whether at the time of the original copying or later is uncertain. The editors resist the conclusion that the stichometric letters could be written by someone other than the original scribe, even while conceding that some of the letters are differently formed.17 In fact, it is difficult to judge (PL 7b). Now the addition of stichometry was normally a part of the activity of the bookroll's manufacture. In the Milan papyrus, there is no evidence of a systematic corrector (Siopdwrrjc); so one expects the stichometry to have been the original scribe's reckoning of the line totals, as a help towards securing the accuracy of the line counts and as a measure for payment. But the stichometry in the Milan papyrus is quite unusual, indeed unique (so far as I know), in another, obvious respect. The totals are running totals for each section of poetry, and not therefore suited for the usual purposes of controlling or measuring the scribe's overall work in copying the book.18 Moreover, if, as suggested, the first line of the first column is a later insertion and not part of the original column now labelled 'col. I', then the stichometry would be not only not original to the roll, but added after the repair. This follows from the fact that the count for 'col. I', written at the bottom, reads mu, that is, 40, but the lines only add up to forty if we count the (apparently) inserted line at the top; and similarly for the running total given at the end of the first section on gemstones.19

18 The use of running totals does occur when individual works (such as short speeches) or books of epic (Homer) are included together in a single large bookroll. But that is quite different from the case here. See BG, p. 16 and n. 28 for some Homeric examples; generally on stichometry K. Ohly, Stichometrische Untersuchungen (Leipzig, 1928); D. Obbink, Philodemus, On Piety, Part 1 (Oxford, 1996), 62-73, esP- 62 n. 1. S. Stephens and D. Obbink, 'The Manuscript', 15 suggest that the running totals may be added as a control to the creation of an epigram collection created from an incipit list.

19 One can easily imagine a scenario in which this hand was in fact the original scribe. If the owner had a relationship with a scribe trained to write a bookhand, whether a 'professional' scribe, his or her own slave, or the slave of one of his or her friends, it would be natural to reuse that same person for the process of manufacture of a new epigram roll.

Now if we put these various features together—the protokollon added later, with the left part of the manufactured roll cut away; the initial line very possibly added later; the stichometry also possibly a later addition; the problem that the section heading at the top of the first column may itself be a later addition, or even a phantasm—we can imagine a manuscript with a history rather different from that supposed by the editors.

The original bookroll may, I suggest, have extended not just to the right of the fragmentary bookroll that we possess but also to the left. That is, there is no compelling reason to think that 'col. I' was the original first column of the roll, and thus no reason to think that the original bookroll even began with the section on gemstones.20 The owner seems to have negotiated with the scribe for an economical copy, but the reader or readers of the bookroll neither corrected the text, even in the most obvious or egregious instances, nor made any of the usual lectional markings, like punctuation or word groupings, in this sometimes difficult poetry. Over the course of its life, the roll suffered moisture damage. Some blotted lines were crudely overtraced to correct for that; and the damage to the front of the roll (which may also have been water damage, but of course we do not know) occasioned the cutting-off of some papyrus at the front, apparently including at least one and perhaps many columns of writing.21 Someone took the trouble to count the number of lines in each section of the poetry and add running totals; perhaps after the repair.22 Someone, possibly the same someone, set a mark—tou(to)—alongside a very few of the epigrams, presumably for the purpose of either reading or copying out certain selected epigrams. We do not know when this person sat down to make these notes of selection. But it is certainly striking that the unusual counting activity occurs in the same roll as the equally unusual activity (among papyrus witnesses, that is) of selecting out certain epigrams.

Indeed what seems most significant about the users' activities, or at least the traces we have of those activities, is the absence of marks for

20 The line added to the top of our 'col. I' would have been taken from the bottom of the previous column before the cut and repair was made. (Ancient manuscripts do not accommodate to the modern habit of ending a column with the pentameter.)

21 H. C. Gotoff points out to me that at least two columns would be probable at the left. The c.8 cm cut away from the left of the first /coAAi^a is too narrow to have held a complete column (of 9.5 cm) by itself. We need therefore to assume, probably, at least one additional /cdAAi^a (of 19.5 cm), thus 27.5 cm (19.5 + 8) of available space, enough to accommodate two columns of 9.5 cm with 8.5 cm of space for the agraphon at the left. Whether one /cdAAi^a or several preceded what is now 'col. I' is of course anyone's guess.

22 If the original scribe, then working with the owner a second time. See above, n. 19.

reading in combination with these exceptional indications of counting and selecting. In the study of ancient bookrolls we commonly see, especially in non-calligraphic productions, those marks of punctuation, correction, and even marginalia that reveal the use of a book by a lector or reader, often with implications of the book's use by readers, including groups of readers, for reading entertainment, or for more serious study.23 The very fact that in this particular bookroll the lack of lectional marks combines with the presence of marks of counting and selecting seems to speak to how the bookroll was used. Readers' interaction with this epigram book seems to have differed in a fundamental way from their use of, and activities surrounding, other types of poetry books. This artefact shows, rather, marks suggesting that the selection and editing activity was an essential aspect of the readers' approach to the text.24

Other early examples of epigram are by now well documented and reasonably well known.25 Most are too fragmentary to draw firm conclusions about the nature of the collection. But the few examples we have display a notable interest in balance, arrangement, and indeed in the numbers. The still-unpublished incipit list in Vienna (P. Vindob. G 40611), dating to the third century BC, contains a list of the opening lines of epigrams, together with line counts, divided by books.26 Kathryn Gutzwiller has pointed out that the first book documented by the Vienna incipit list, with 83 epigrams and 344 lines, averages just a hair over four lines per epigram, while the second book (33 epigrams, 264 lines) averages eight. Her conclusion is that this difference in average line length 'suggests that the length of individual poems was one criterion for organization'.27 Berol. 9812, also a third-century-BC papyrus, gives fragments of three four-line poems, all dedicatory.28 P. Köln V 204, a second-century-BC papyrus, contains six epigrams by Mnasalces, all four-line poems, and all inscriptional

23 K. McNamee. 'Greek Literary Papyri Revised by Two or More Hands'; Johnson, 'Toward a Sociology of Reading in Classical Antiquity'; id., 'Scholars' Texts and Reading Communities in Hellenic Egypt'.

24 See Gutzwiller, Poetic Garlands, 14 and passim.

25 Good, accessible account ibid., 20—36; cf. Cameron, Greek Anthology, 1—18, esp. 7 ff.; Parsons, 'Callimachus and Hellenistic Epigram', esp. 115—22.

26 See Parsons, 'Callimachus and Hellenistic Epigram', 118—20; Gutzwiller, Poetic Garlands, 24 and n. 34; Cameron, Greek Anthology, 9—10. Cf. also P. Oxy. 3724, a similar incipit list from the ist c. ad with markings that look like someone working on the arrangement and what was to be included. These two incipit lists can be, and have been, interpreted as catalogues of an existing poetry book rather than as witnesses to the work surrounding the creation of a new poetry book.

27 Gutzwiller, Poetic Garlands, 24.

28 In one sense or another. See discussion at Gutzwiller, Poetic Garlands, 30 for the question of the unity of type.

in type. Again, following Gutzwiller,29 the consistency of length of poem and 'ostensibly inscriptional' type creates a 'formal coherency' within which the arranger can then create ttoikiXIo, through variety of theme and through associative and dissociative juxtaposition. A formal unity, that is, is combined with clever linkage and thematic diversity. With these witnesses as background, it is worth remarking that the epigram-selecting activity we find in the Milan papyrus marks only quatrains—perhaps, then, a mark of formal unity as a criterion for the selection.30 Be that as it may, it is clear that the Milan papyrus bears witness to reader interest in counting, in the numbers, that seems more likely to be a facet of the editing process—of selecting and re-collecting poems—than of scribal control.

Selecting, collecting, organizing the 'gems' seems, then, fundamental to the idea of the epigram collection from its earliest witnesses. At a minimum the artefactual details of the Milan papyrus suggest that the next generation, as it were, of interaction with this bookroll was focused on the activity of selecting and re-collecting for the next bookroll, that is, on the editorial work of creating afresh collection. Though the basis of evidence is not wide, there does seem to be reason to suppose that the Milan papyrus falls into a type of early reader interaction with epigram collections, in which the focus is not only on the authorial production but also on the editor's arrangement, very like the attitude of a viewer presented with artistic photographs collected in a scrapbook or photo essay—in which, that is, the viewer's awareness is of both photographer and designer.

Finally, it is important to distinguish this situation from the Roman poetry collections we know, which also show a strong interest in arrangement, balance, and to limited extent also in the numbers.31 Catullus perhaps aside, the Roman poetry books that come down to us are almost all authorial collections, and the ancient reader

29 Gutzwiller, Poetic Garlands, 31.

30 The quatrain is, however, the most common size for epigram.

31 The extent of Roman interest in such matters has been muddied by understandable scholarly reaction to various efforts to box the Augustan poets within the confines of a radical numerology or schema: e.g. G. Duckworth on Vergil {Structural Patterns and Proportions in the Aeneid, Ann Arbor, 1962), H. Dettmer on Horace (Horace: A Study in Structure, Hildesheim, 1983), O. L. Richmond on 'stanza-schemes' in Propertius (Sexti Properti quae supersunt omnia (Cambridge, 1928), esp. 27 ff.), O. Skutsch on the Monobiblos ('The Structure of the Propertian Monobiblos', in which, despite some valid pairings of poems, the insistence on a consistent overall scheme is faulty). The work of, for example, M. Santirocco on Horace has helped to clarify the principle that interest in balance and arrangement does not require a strict or mechanical overall scheme; and that interest in uoiKiXia does not imply a thoroughgoing lack of deliberate, formal structural design. For theoretical discussion and references to earlier work, see Santirocco, Unity and Designin Horace's Odes (Chapel Hill, 1986), 3-13.

will have presumably supposed in the reading that the author was in control of the original sequence and linkage among poems. I raise here the possibility that the Ptolemaic reader may not have had this disposition. If that selecting activity was at all essential and common in the reader's interaction with the text (a hypothesis which, again, stands on slim if very intriguing evidence), then the reader's attitude towards that peculiar pairing of unity and diversity characteristic of epigram collections will be marked by the relationship of reader to editor, of a reader's willingness to delight in the newness brought to poetry by the crafting of fresh arrangements and newly created relationships between poems.32 It may well be, that is, that the Ptolemaic reader would no more assume the author's hand in the selection and arrangement of an epigram poetry book than he or she would assume the author's hand in the punctuation of a text of Homer or Simonides or, for that matter, Callimachus or Herodas.

32 This sort of reader interaction with the text can be seen as part and parcel of the delight in imitation of the original context, that conveyance of the occasional that seems to inform the gentlemanly hobby of epigram-collecting; just as one might store in one's head the right selection of poetry for a symposium (cf. Parsons, 'Callimachus and Hellenistic Epigram', 104), so one could put together an elegant compilation into a poetry book to demonstrate one's cultural taste and credentials.

0 0

Post a comment