Michael Psellus and the Date of the Palatine Anthology Cameron, Alan Greek, Roman and Byzantine Studies; Winter 1970; 11, 4; ProQuest pg. 339 Michael Psellus and the Date of the Palatine Anthology Alan Cameron THERE has until recently been general agreement that the unique manuscript of what is commonly known as the Palatine An­ thology-Palatinus graecus 23-dates from the tenth century. It would hardly be possible to assign an exact date to the codex as a whole, since it falls into two distinct parts, one older than the other, both of which were completed and revised by at least two other more or less contemporary parties. But P. \Valtz, the Bude editor, placed the completion of the work ca. 980, while taking the earliest elements back as early as ca. 930. Now the Palatine Anthology is largely based on an earlier anthol­ ogy (now lost), compiled by one Constantine Cephalas. Accepting the common and natural identification, Cephal as was active in Constanti­ nople (as protopapas) in 917. His anthology presumably dates from the end of the ninth or beginning of the tenth century. Thus the earliest part of the Palatinus might be no more than a few decades later than the anthology on which it was based. In the course of a challenging reassessment of the manuscript tradi­ tion of the Greek Anthology as a whole,! R. Aubreton has recently at­ tempted to cast doubt on this generally accepted picture. While not denying that there is a Cephalan kernel to the Palatine Anthology, Aubreton argues that there are substantial later accretions round this kernel-accretions which suggest to him a longer interval between the two works. Accordingly he has proposed a much later date for the Palatinus, between 1050 and 1070. Quite how far the Palatinus does reflect the anthology of Cephal as is admittedly very uncertain.2 And if we could be sure that there was an interval of 150 rather than 50 years between them, some traditional 1 "La tradition manuscrite des epigrammes de l'Anthologie grecque," REA 70 (1968) 32- 82. 2 See the recent discussions of A. S. F. Gow and D. L. Page, The Greek Allthology: HellCllistic Epigrams I (1965) xvii f; F. Lenzinger, Zur griechischCll Allthologie [Diss. Zurich 1965] 2f; and my own paper "The Garlal1ds of Meleager and Philip," GRBS 9 (1968) 323f, by which, on the 339 340 PSELLUS AND THE DATE OF THE P ALA TINE ANTHOLOGY assumptions might well need to be reconsidered. But can we be sure? Aubreton's case as so far presented rests almost entirely on his own assumptions about the relationship between the two anthologies, for the palaeographical study of the Palatinus by G. Rochefort to which he refers for support has not yet been published. I am not myself com­ petent to judge the palaeographical issue, but one eminent authority I have consulted, Mr N. G. Wilson, tells me that he remains firm in his adherence to the communis opinio. In a new paper Aubreton has tried to support his position by another line of argument.3 AP XIV 5, 35 and 58, anonymous in the Palatinus, are also transmitted in a collection of riddles ascribed to Psellus in Parisinus graecus 968, first published by Boissonade in his Anecdota graeca III (Paris 1831) 430f, now more accessible in the Didot Anthologia graeca III, ed. Cougny (Paris 1890) VII 34-45. This collection is dedicated to Psellus' feeble pupil, the Emperor Michael Ducas, who reigned from 1071 to 1078. If these poems really are by Psellus, then the traditional date of the Palatinus would certainly have to be abandoned. Written as they there are in the body of the text by two of the main scribes of the manuscript (B and B2), they cannot be explained away as subsequent additions or interpolations. But are they by Psellus? Aubreton claims that it is only because of their blind acceptance of the traditional date that previous scholars have failed to take the ascription to Psellus seriously. This mayor may not be so, but a number of substantial objections to the ascription can be formulated nonetheless. The collection in Par.gr. 968 comprises: first the dedication to Michael Ducas, then three riddles in political verse (not, as Aubreton states, in prose), next the three elegiac epigrams AP XIV 5, 35 and 58, and finally twelve more riddles in iambics (or rather, 'Byzantine twelve-syllables'). We may begin with some objections to Aubreton's case as he for­ mulates it himself. Instead of arguing, as one might have expected, that the epigrams must have been written between 1071 and 1078 whole, I still stand, though I might now express myself more cautiously on some points in the light of Aubreton's discussion. When writing my paper I had nor yet seen E. Hirsch's somewhat more adventurous study "Zum Kranz des Philippos," WissZeitUnivHalle 15 (1966) 401f. 8 "Michel Psellos et l'Anthologie Palatine," Ante/ 38 (1969) 459[. ALAN CAMERON 341 (when Psellus probably died) and that the Palatinus cannot therefore be earlier than 1071, on the contrary Aubreton firmly maintains that it is earlier than 1071. Believing (as he does) on palaeographical grounds that the scribes Band B2 wrote "vers 1060," Aubreton strangely suggests that at this date only the elegiac riddles had been written. This is why Band B2 did not include the iambic pieces, which did not see the light till the edition dedicated to Michael Ducas after 1071. It is difficult to take this sort of reasoning very seriously. Whether Psellus wrote (which I doubt) or merely collected these riddles, in the absence of any indication to the contrary we are surely bound to assume that he published them all together-that is to say in the only edition we know of, that dedicated to Michael Ducas. How do we account for the fact that these three elegiac riddles are left anonymous in AP? Aubreton casually remarks that B was appar­ ently unaware of their Psellan authorship. Yet if they were hot from the pen of the greatest living man of letters, how can B not have known who wrote them? The scribes and redactors of AP took more pains over their ascriptions than we find in any other Byzantine anthol­ ogy. And we are dealing here with two scribes: B copied XIV 5 and 35, B2,58. It would be surprising ifB, B2and their reviser Jhad all three overlooked the claims of Psellus-especially after taking the trouble to include his newly written work in a collection of ancient riddles in the first place. More generally, it would be most unusual if all eighteen of the riddles in Psellus' collection had been original products ofPsellus' own ingenuity. Riddle books, in ancient and Byzantine times as today, tend to be for the most part collections of anonymous puzzles handed down from generation to generation. It is the exception rather than the rule for a riddle to be associated with a named author. Of all the riddles and puzzles assembled in AP XIV, only one has a really plau­ sible looking ascription: 63, to Mesomedes. I would not myself place overmuch confidence in the attribution of 101 to the sage Cleobulus or his daughter. Since Tannery, it has been generally recognised that the last section of Book XIV, 116-146, was taken, complete with their elaborate and learned scholia, from a collection of problems by one Metrodorus, of uncertain date but hardly later than the fourth cen­ tury.4 Against 116 B2 has written MYJrpo8wpov i7rLypeXfLfLcxrcx apdJfLYJTtKeX, 4 P. Tannery, ed. Diophanti Alexandrini Opera omnia II (ST, Leipzig 1895) xi f, and pp.43f 342 PSELLUS AND THE DATE OF THE PALATINE ANTHOLOGY yet neither 116 itself nor its successors is provided with a regular ascription to Metrodorus. We find instead the formula &,uo.5 The col­ lection was Metrodorus', but he was not credited with the author­ ship of its individual components in the ordinary way. Let us take a much later example, the probably thirteenth-century collection attributed to Eustathius Macrembolites, equipped with metrical solutions by Maximus (or Manuel) Holobolus.6 The lemmata would seem to suggest that only the first eleven were believed to be by Macrembolites himself, and they are the only ones Holobolus pro­ vided with his own solutions. Of the rest, some are attributed to two shadowy twelfth-century figures called Aulikalamus and Prosuch,7 while thirteen others are left anonymous. Of these thirteen, no fewer than seven were taken from earlier collections, those of Christopher of My til ene, Psellus, and another mysterious figure (of quite uncertain date), Basil Big-nose (Megalomytes).8 Nor can we be sure how many even of the first eleven were really written by Macrembolites, since no. 8 turns out to be none other than AP VII 311, attributed to Agathias by Triclinius on Sophocles, Electra 150, but falsely,9 since there exists a very close Latin translation dating from the late fourth century.10 Editors of the Anthology have all missed its appearance in Harleianus 5624, attributed to the Emperor Julian, perhaps correctly.ll Basil Big-nose's book would seem to belong somewhere in the period between Psellus and Macrembolites. It contains three riddles for the Palatine scholia. There is a useful summary in F. Buffiere's new Bude ed., Anthologie grecque XII (Paris 1970) pp.34f, in most other respects a careless and undistinguished piece of work (at p.37 J. A. Fabricius appears as an ancient author!). 5 On which see Gow, The Greek Anthology: Sources and Ascriptions (London 1958), 29 and K. Preisendanz, Gnomon 34 (1962) 657. Quite unjustifiably and misleadingly Buffiere heads 116 MTJTPOSWpov and its successors TOV av.rov. 6 M. Treu, Eustathii Macrembolitae quae ferunwr aenigmata (Program des Konig!. Friedrichs­ Gymnasiums zu Breslau, 1893). 7 On whom see Treu's discussion, op.cit. (supra n.6) 31f. 8 I have not been able to trace a Single word of scholarship devoted to Basil, but it seems clear to me that his book is later than Psellus' and draws on it, not vice versa. Concerning his curious sobriquet, Robert Browning draws my attention to a ·Small·nose' in Theophanes Continuatus p.175.9 Bonn. 9 Though accepted, for example, by I. Hilberg, BZ 3 (1894) 174. 10 Ps-Ausonius, Ep. 26 (p.260 Schenkl). Since the discovery of the Epigrammata Bobiensia there is no need to doubt that this poem, if not actually by Ausonius, is at least genuine fourth-century work. 11 With an interesting new ascription, "To a silkworm" (see J. Bidez/F. Cumont, edd. Juliani imperatoris epistulae, leges, poematia,Jragmenta varia (Paris/Oxford 1922) p.22I. ALAN CAMERON 343 by Christopher of Mytilene and several taken or adapted from Psel­ Ius' collection. As before, there is no indication that any of the items in question were in fact by Christopher or Psellus, though on the other hand (and again as usual) none of them is expressly ascribed to Basil himse1f.12 This is not plagiarism in the ordinary way. It is simply that riddles were felt to be common property.13 We come at last to Psellus' collection. We may reasonably assume that the first three riddles in political verse come from Psellus' own pen. That some of the twelve iambic pieces are by him is probable; that some are not is certain. Three can definitely be attributed to his older contemporary Christopher of Mytilene: Psellus 45 (Cougny) on the seasons= Christopher 56 (Kurtz); Psellus 38 on the tent= Christo­ pher 71 ; Psellus 42 on the scales= Christopher 21. I entirely agree with Christopher's editor E. Kurtz that there can be no question about Christopher's prior claim in all three cases. Christopher's poems are not just a riddle book but a substantial and amazingly diversified col­ lection on a multitude of subjects by a talented and versatile poet. There are just six riddles in the whole book, all of which reappear time and again in the later riddle books. Another poem attributed elsewhere to Psellus (Par.gr. 1182, f.15F) turns out to be by Christo­ pher (50 Kurtz). Indeed, Christopher's poems often suffered the fate of attribution to later and more famous writers.14 But for the preser­ vation of his whole collection in a sadly fragmentary state in one fif­ teenth-century manuscript, we should have known almost nothing of his influence. Psellus would have been one of the chief beneficiaries. Psellus no doubt has other literary creditors, whom we are no longer in a position to trace. That Psellus wrote some at least of the other nine iambic riddles is likely enough. That he wrote the elegiac riddles is not at all likely. To 12 Basil's riddles were published in J. F. Boissonade, Anecdota graeca III (Paris 1831) 436f, and again in Cougny vn 47f, with a few useful notes. Cougny did not, however, know of the existence of Christopher of Mytilene (whose poems had only been published three years previously): on Basil's debt to Christopher, see E. Kurtz's edition of Christopher (Leipzig 1903) xv. 18 I know of no modern account of Byzantine riddle poetry. In the 1890s Sp. Lampros was reputed to be preparing the definitive work, but it seems never to have appeared, and his paper in De/don Z (1885/9) 154f promised little enough. There are two sentences in F. Dtilger, Die byzantinische Dichtung in der Reinsprache (Berlin 1948) 27. For a brief popular account (based mainly on Krumbacher) see Archer Taylor, The I-iterary Riddle befoTe 1600 (Berkeley 1948) 45f. 1C See Kurtz's ed. op.cit. (supra n.1Z) pp.xv f. 7-G•R•B•S• 344 PSELLUS AND THE DATE OF THE P ALA TINE ANTHOLOGY the best of my knowledge Psellus always wrote in political or iambic verse, never in elegiacs. Even if he did write elegiacs occasionally, I doubt whether he would have been capable of AP XIV 5, 35 or 58. They are undistinguished enough by ancient standards, to be sure, yet unmistakably ancient all the same. Quite apart from the problem of the quantities, the feel for the handling of caesura and hiatus that came instinctively to all educated writers up to the sixth century was somehow lost to Byzantine litterateurs of later centuries. No one could mistake the elegies of Arethas, Ignatius, Cometas or even Christopher for genuine ancient workmanship. We must, I think, conclude that Psellus took all three elegiac riddles from an earlier collection, most probably though not necessarily an exemplar or apograph of AP XIV.lS This brings us to another and yet more decisive objection to the Psellan ascription. Aubreton assumes that Band B2 were the com­ pilers rather than the copyists of AP XlV. The three rPsellan' poems are held to have been added to a collection they had largely gathered themselves. That is to say, AP XIV did not exist before AP. Here, however, as in his earlier reexamination of the manuscript tradition of the Anthology as a whole, Aubreton has forgotten the collection of epigrams in Laur. 32.16, an anthology now recognised to have been written in Planudes' own handl6-a precursor to his mag­ num opus of twenty years later. Two of the rPsellan' epigrams-AP XIV 35 and 58-appear in the selection of riddles on f.382v. Now it is true that this famous codex dates from 1280/3, some two centuries later than even Aubreton's date for AP. Yet there are reasons, not to my knowledge so far per­ ceived, for supposing that it derives from a fuller and earlier collec­ tion than AP XIV. The riddles included in this selection are as follows (all references, except the two to Cougny, to AP XIV): Cougny VII 31; 19; 26; 30; 32; 35; 41; 42; 57;17 58;17 60; 62; Cougny VII 23; 9; 12; 20; 22; 56;18 61. 15 The not inconsiderable variants effectively exclude the possibility that Psellus drew direcdy on AP. For other redactions of the riddles of AP XIV see below. 16 C. Gallivotti's discussion, in Bollettino del Comitato per la preparaziont dell' EdiZione Nazionale dei Classici Greci e Latini 7 (1959) 37f, supersedes earlier work but still leaves much to be desired. I hope to discuss other aspects of the two epigram anthologies it contains elsewhere before long. For notes 17 and 18, see page opposite. ALAN CAMERON 345 The most striking feature of this collection is surely the order in which the riddles stand compared with their order in AP XIV. Those in the first line follow the order of AP exactly, and though the seven in the second line break the original sequence, they are still in the order of AP. It looks as ifPlanudes excerpted the same original sequence twice. Having copied out his first choice of a dozen riddles (the riddle sequence ends at XIV 64 and the last poem in L is XIV 62), Planudes decided to fill up his page with a few more (all but two of the second batch are distichs, and they do reach exactly to the bottom of the page). So he skimmed through his exemplar again from the begin­ ning, adding a few he had missed the first time round (for example, having originally passed from 60 to 62, he now took 61). The implication is that Planudes had before him a sequence of riddles arranged in exactly the same order as the sequence in AP XIV. What now of Cougny VII 31 and 23, known to us (in the Anthology tradition)19 only from L? Both could, of course, be additions from another source, but for a variety of reasons this seems unlikely. In the first place, the riddle sequence in L is the middle element in a tripartite anthology of epigrams and oracles (ff.38F -384r). First 15 oracles, second 19 riddles, third 27 epigrams from monuments to charioteers. Now by a lucky chance the whole series of charioteer epi­ grams, just as they were transcribed direct from the original monu­ ments,20 is preserved in Planudes' later, definitive, anthology of 1299 (Marc. 481, ff.43v-45v, split up in modern editions between AP xv 41- 50 and APlan. 335-378). Thus it is possible to see that the series in L is just a selection from this fuller series, with no additions from else­ where. The oracles are described as being taken £.1< rijc 8€oco4>tac. The Theo­ sophia is a late fifth or early sixth-century Christian work which quotes a number of pagan oracles carefully chosen for their possible appro­ priateness to things Christian. It does not survive entire, but we do 17 In fact only the first couplet of 57 and 58 is written in this place. The second couplets follow after Cougny VII Z3. It is clear from the symbols in the margins that the omission was acddemal, and for the sake of simplidty I have restored Planudes' intended order. 18 Gallavotti, op.cit. (supra n.16) 46, claims that Cougny VII Z9 is written between AP XIV 56 lines l-Z and lines 3-4 of the same poem. This is a curious slip: Cougny vn Z9 is AP XIV 56.3-4. Buffiere inexcusably perpetuates the error (critical note ad loc.). 19 Cougny VII 31 is known, with a couple of unimportant variants, from Athenaeus: see below, p. 348. 20 See my full discussion in a forthcoming book, Porphyrius the Charioteer. 346 PSELLUS AND THE DATE OF THE PALATINE ANTHOLOGY have manuscripts containing fuller excerpts than those in L (pub­ lished in H. Erbse's very full and elaborate edition, Fragmente griechi­ scher Theosophien [Hamburg 1941]). Once more, L's series is simply an excerpt from this one source. A priori, the probability is that L's riddle sequence derives from just the one source. Whether Planudes drew on three separate sources for this tripartite anthology in L, or just one source where the three parts were already (in fuller form) united, is a question that need not concern us for the moment (it may be in­ soluble).21 But we ought to think twice about postulating more than one source for each part. In the second place, Cougny VII 23 is not in fact a riddle. It would be odd for Planudes to have gone to the trouble of consulting another source for his riddle sequence and then add a poem which was not a riddle at all. On the other hand, it would not be odd for the compiler of the original riddle series to have included Cougny VII 23, since the sequence in AP XIV includes no fewer than three poems which, though mildly puzzling, are no more riddles than Cougny VII 23: nos. 8, 10 and 15. In this connection, it is particularly instructive to see that L includes XIV 9, which is not a riddle but an arithmetical puzzle. Now AP XIV (like the anthology in L) is tripartite: it contains an approximately equal number each of riddles, oracles and arithmetical problems. Basically, the oracles occupy the middle of the book (65-100), and the problems the end (116-146), though between 101 and 115 there is a mixture of riddles and oracles. Basically, the rest of the book consists of riddles, though most of 1-13 are problems (1-4, 6-7, 11-13), as also are 48-51-not to mention the three that are neither riddles nor prob­ lems (8, 10, 15). That is to say, the book is not divided cleanly and straightforwardly (like the tripartite anthology in L) into its three component parts. Some attempt has been made to provide short mixed sections, without however breaking up all the material in this way. Now Aubreton, for whom the Palatine scribes are the compilers rather than just the copyists of AP XIV, would ascribe this idiosyn­ cratic and haphazard method of arrangement to those scribes rather than to their sources, since in his view they made additions to what­ ever sources they may have used at irregular intervals throughout (i.e. the three 'Psellan' poems, 5, 35 and 58). Yet here we have L evi­ dently drawing on a source which is not AP but which nevertheless 21 I hope to discuss the matter elsewhere before long. ALAN CAMERON 347 already contained the idiosyncratic melange of riddles and problems AP XIV 1-13. If it can be shown that L's source was fuller than AP XIV, then it would have to be a collection (or an apograph of a collection) which was earlier than AP XIV. And ifL's source contained Cougny VII 31 and 23, then it was fuller than AP XIV. For the sake of convenience let us call this postulated common source of AP XIV and L by the name X. Now the riddle sequence proper in AP XIV runs from XIV 14 to 64. With the exception of XIV 9 from the mixed sequence 1-13, all the riddles in L are from this main sequence, running from 19 to 62. Now since Cougny VII 31 stands before XIV 19 in L, presumably it also stood before 19 in X-probably between 14 and 19. Here is the poem: n ' , "1:' 1:" , " () ,,, ~ €VT avap€c a€Ka V7JVCt KaT'YjI\V av €LC Eva XWPOV, '\:"'\'(} " \'{} \:" , '" , \'() €V OE L aLe €fLaxoVTO, I\L ov 0 OVK 'Yjv aV€I\€C aL' I:",/~ 1:" 'c '\\ "\:' 1:" f ~ , aL,#, U 0 €~ WIV\VVTO, vowp a V7T€P€ L X€ Y€V€La. The answer is apparently a shipwreck. The immovable stones they fight in or among are the rocks the ships have crashed into, and the 'thirst' alludes to the undrinkable salt water the men drown in. This still leaves one or two details unclear (i.e. the numbers in line 1),22 but whatever the exact interpretation, is there a suitable context for a riddle of this form between XIV 14 and 19? There is indeed. Compare XIV 14 itself: E9''' \:' ' ~ " \:' , ~ LC aVEfLoc ovo V7J€C €P€TTOVCtV O€Ka vaVTat· 9'\:" f3 ' '.J.' ''\' € tC O€ KV €PV7JT7jC afL't'oTEpac € a€L. The answer here is certain: the double flute (av'\oc) , the ten sailors being the ten fingers and the wind the breath of the player, while the two ships are of course the two pipes. But more relevant than the solution is the form of the riddle. Like Cougny VII 31, the point turns on a specified number of ships and sailors in contact with the ele­ ments. Naturally enough there are many other examples of riddles on the same or linked themes being juxtaposed (e.g. XIV 17-19, on hunting; 23-24, on birth; 32-33, on Nessus the centaur; 40-41, on day 2Z Diels solved the problem by simply transposing the numbers. so that there were ten sailors and five ships. Professor Geoffrey Arnott draws my attention to the note in Dale­ champs' sixteenth-century translation of Athenaeus (ad 4570): "aenigmatis esse puto hunc sensum: quinque dassis praefecti. pugnam decem navibus commiserunt ... ," which is perhaps a better way of dealing with line 1. 348 PSELLUS AND THE DATE OF THE PALATINE ANTHOLOGY and night). Cougny vn 31 and AP XIV 14 would make highly appro­ priate neighbours. L omitted one of the pair, AP XIV the other. It was only in X that both stood together. Now for Cougny vn 23: Kot$p7] ']KapwLo 7T€PUPPWV II7]v€A67T€La, Eg 7Todv £fLfJ€fJavia. TpLaaKTVAoc £g€cpaav(}7]. The point, of course, is not that Penelope herself has six feet and three fingers, but that line 1 (which often occurs in the Odyssey) has six metrical feet of which three are dactyls. Its immediate predecessor in L is XIV 62, one of the last of the riddle sequence in AP, its successor XIV 9, one of the first. In X, then, it must have stood either at the very end, or at the very beginning, where Planudes found it his second time round. The beginning is much more likely, since Cougny VII 23 has nothing in common with the riddles at the end and everything in common with the material at the beginning. It has been remarked already that there are three poems in AP XIV that are neither riddles nor puzzles. All occur together at the beginning: 8, 10 and 15. If Cougny vn 23 came at the beginning in X, it would have stood before 9, sandwiched between two of these intruders. Here is 8, a similar bit of counting in verse, on the numbers on opposite sides of a die: