Half the Anthology hangs from one book

Claim (verbatim)

The Greek epigram tradition survives through two filters that disagree, and their disagreement is countable. The Palatine Anthology - some 3,700 epigrams in fifteen books, assembled in the tenth century out of Constantine Cephalas' collection - survives in exactly one manuscript, the codex Palatinus (Heidelberg, Pal. gr. 23, with its detached final portion now Paris. suppl. gr. 384), unknown to scholarship until Salmasius recognized it in 1606/7. Until then 'the Greek Anthology' meant Maximus Planudes' edition of 1299 (his autograph survives: Venice, Marc. gr. 481; first printed by Janus Lascaris, Florence 1494): about 2,400 epigrams in seven books under his own taxonomy, plus some 388 epigrams absent from the Palatine - the Appendix Planudea, printed since as book 16. Planudes was a monk editing for use, and he cut on morals: Strato's paederastic Musa Puerilis (AP Book 12, 258 epigrams) he suppressed nearly entirely, and erotica elsewhere he dropped or laundered. The two-filter geometry means, first, that the Byzantine thirteenth century actively deleted a documented slice of its own inheritance - a rare case where the censor's exact cut-list survives; and second, that a third or more of the extant corpus has a single point of failure: had the Palatinus perished - and it spent the Thirty Years' War as loot and the Napoleonic era split between Rome and Paris, where part of it remains - the erotic, skeptical and satirical registers of a millennium of Greek epigram would exist only as Planudes' expurgation.

Prediction clause (verbatim)

Prediction: computing per-book retention rates - the share of each Palatine book's epigrams also transmitted by the Planudea, per the concordances of the standard editions (Beckby's Anthologia Graeca; the Bude Anthologie grecque; Paton's Loeb) - AP Book 12's retention rate will be at most one third of the mean retention rate of Books 6, 7 and 9 (dedicatory, sepulchral, epideictic), and the Palatine-only corpus (epigrams with no Planudean transmission) will exceed one third of the Palatine's total (primary clause: the Book-12 suppression ratio at <=1/3 of the comparison mean; the verdict follows it); disambiguation: epigram identity follows the editions' concordance numeration, doublets counted once; coverage guard: void if concordance tables cannot be tabulated for any of the four books. Confidence: direction certain, exact retention figures MODERATE, thresholds cushioned.

Kill-dataset (verbatim)

Kill: the Palatine-Planudea concordances in Beckby's Anthologia Graeca and the Bude Anthologie grecque, with the codicological anchors (Pal. gr. 23, Paris. suppl. gr. 384, Marc. gr. 481) confirmed in Pinakes/diktyon - a per-book retention table.

Novelty / leakage triage

already answered in the literature

Cameron's monograph and the standard editions contain the answer: the Palatine-Planudean relation is tabulated in Beckby's concordance tables, Planudes' near-total suppression of Book 12 is documented with counts in Cameron's account of the Planudean recension, and the gross arithmetic - some 3,700 Palatine epigrams against some 2,400 Planudean, 388 of those absent from the Palatine - is stated in the introductions; the per-book retention table re-derives printed concordances and both pinned clauses are guaranteed by them.

• A. Cameron, The Greek Anthology from Meleager to Planudes (Oxford, 1993)
• H. Beckby (ed.), Anthologia Graeca, 4 vols (2nd ed., Munich, 1965-1968), concordance tables

Predictions

No prediction registered yet.