The drowned codex speaks Latin

Claim (verbatim)

Archimedes reached the Middle Ages in three Byzantine books. Codices A and B stood in the papal library at Viterbo, where William of Moerbeke translated from both in 1269 - his working autograph survives as Vat. Ottob. lat. 1850; codex C, the palimpsest prayer book of 1229, was identified by Heiberg in 1906, vanished again, and returned at auction in 1998 to be imaged (Netz, Noel, Wilson and Tchernetska's 2011 publication). The deaths are dated: B disappears after the papal inventory of 1311; A disappears after the 1560s, surviving only through its Renaissance copies. Now the geometry: On Floating Bodies stood in B and C but never in A - so from 1311 until Heiberg read the palimpsest, the founding treatise of hydrostatics existed nowhere in Greek reach, and its sole complete carrier for six centuries was Moerbeke's Latin, the text Tartaglia printed in part in 1543 and Commandino corrected in 1565, the text through which Stevin and Galileo's century knew Archimedes on flotation. Modern editors still read parts of it through the Latin where the palimpsest is illegible. And the control case is exact: the Method and the Stomachion stood only in C - no translation exit at all - and accordingly ceased to exist for those same six centuries, returning only by palimpsest accident in 1906. Translation is carrier-diversification, and one friar's literal Latin carried a treatise of Archimedes alone across a six-hundred-year gap in which every Greek witness was dead.

Prediction clause (verbatim)

Prediction: compiling the work-by-codex incidence table over A, B and C with the Moerbeke translation from Heiberg's second-edition prolegomena and the 2011 palimpsest publication, On Floating Bodies will show absence from A and every A-descended Renaissance copy, presence in B and C, zero accessible Greek witnesses between 1311 and 1906, and the Moerbeke Latin as its only complete carrier in that gap (primary clause: the incidence-and-gap geometry of On Floating Bodies; the verdict follows it); secondarily, both C-only works (Method, Stomachion) will show no medieval translation and no printed edition before 1907, and at least three treatises of the corpus will each depend on a single medieval carrier; Renaissance Greek manuscripts count as A-descendants per Heiberg's stemma, citations do not count as carriage, and the test voids if the cited editions leave the stemma assignment contested for more than two works.

Kill-dataset (verbatim)

Kill: Heiberg's Archimedis opera omnia (second edition, 1910-15) prolegomena, Clagett's Archimedes in the Middle Ages I-II, and The Archimedes Palimpsest (2011) - a work-by-codex incidence census with a dated carrier-gap computation.

Novelty / leakage triage

already answered in the literature

Heiberg's prolegomena print the work-by-codex incidence, and the primary geometry is the stated centerpiece of the literature: On Floating Bodies absent from A, present in B and C, with Moerbeke's Latin (Ottob. lat. 1850) its only complete carrier from B's disappearance after the 1311 inventory until Heiberg's 1906 palimpsest reading - stated in Heiberg, in Clagett, and again in Netz-Noel; the single-carrier count is tabulated shelf-arithmetic. One control leg is factually off: the Stomachion did have a translation exit - an Arabic fragment published by Suter in 1899 - so the secondary 'no medieval translation and no printed edition before 1907' fails for it; the primary clause is untouched.

• J.L. Heiberg, Archimedis opera omnia, 2nd ed. (Leipzig, 1910-1915), prolegomena
• M. Clagett, Archimedes in the Middle Ages, I (Madison, 1964) and II (Philadelphia, 1976)
• R. Netz & W. Noel, The Archimedes Codex (London, 2007)
• H. Suter, Der Loculus Archimedius oder das Syntemachion des Archimedes, Abhandlungen zur Geschichte der Mathematik 9 (1899)

Predictions

No prediction registered yet.