The table of contents outlived the book

Claim (verbatim)

Marcianus graecus 299, the tenth-century Venetian codex that is the flagship witness of Greek alchemy, opens with a pinax - a table of contents drawn up when the anthology it transmits was still whole. The codex then lived the ordinary life of a heavy Byzantine book: across centuries of use, rebinding and excerption, quires and leaves went missing, losses documented in the modern codicological literature (Saffrey's description of the Venice manuscript; the CMAG notice; Berthelot's account of the Saint-Mark manuscript in the CAAG Introduction, which prints its table). The pinax therefore now promises texts the codex no longer contains: an index that outlived its books, a manuscript keeping the receipts of its own decay. Because Parisinus gr. 2325 (thirteenth century) and Parisinus gr. 2327 (dated 1478) descend from the same Byzantine anthology by partly independent lines, some of the pinax's ghosts can be recovered from Paris; a residue can be recovered nowhere, and that residue is a measured sample of loss inside the tradition's best witness - titles we can still read pointing at texts we cannot. The mechanism is the ordinary physics of the codex: the list is written once and survives as a unit, while the listed texts sit distributed through hundreds of vulnerable leaves, so the census outlasts the population it counted. Prediction restated: at least one pinax entry in ten answers to no surviving text in M itself, and at least two pinax-promised works survive nowhere in the Greek tradition at all.

Prediction clause (verbatim)

Prediction: matching the pinax of Marcianus graecus 299 entry by entry against the texts the codex actually transmits and against the wider Greek tradition as edited in CAAG, at least 10 percent of pinax entries will answer to no surviving text in M itself (primary clause: the >=0.10 ghost-entry share within M; the verdict follows it), and at least two pinax-promised works will prove absent from the entire Greek tradition, the Paris manuscripts included; if the pinax cannot be resolved to at least 30 discrete entries, the test voids as a coverage failure rather than resolving.

Kill-dataset (verbatim)

Kill: the pinax of Marcianus gr. 299 as printed and discussed in the CAAG Introduction (Berthelot-Ruelle, 1887, open scans), controlled against the CMAG codex description and Saffrey's codicological study - an entry-by-entry match of promised titles against M's extant contents and the CAAG-edited corpus.

Novelty / leakage triage

already answered in the literature

The pinax-against-contents confrontation of Marcianus gr. 299 is already published work, not an un-run census: Berthelot's CAAG Introduction prints the Saint-Mark table and discusses the announced treatises the codex no longer carries; the CMAG notice and Saffrey's codicological study document the lost leaves and quires behind those absences and identify which announced items lack texts in M; and the Bude prolegomena state which promised texts the Paris line restores and which survive nowhere in the Greek tradition. The pinned 10-percent share is re-expression of a published entry-by-entry inventory.

Predictions

No prediction registered yet.