Fifty-three volumes, four survivors, one copy each

Claim (verbatim)

Constantine VII's excerpting bureau dismembered the historians - some 26 authors from Herodotus to George the Monk - into fifty-three subject collections, a number the surviving proem of the Excerpta de legationibus itself announces. Of the fifty-three, four survive in any direct transmission, and their witness geometry is a portrait of how palace books die. The De legationibus reached the sixteenth century in a single Escorial codex, which burned in the fire of 1671; it lives only through the apographs made from it in the 1570s. The De virtutibus et vitiis rests on one manuscript, the codex Peirescianus (Tours, BM 980). The De sententiis is legible solely as the lower text of the palimpsest Vat. gr. 73, recovered by Angelo Mai's reagents and edited by Boissevain. Only the De insidiis has two partial witnesses (Scorialensis Omega I 11 and Paris. suppl. gr. 607). The mechanism: an imperial reference set was never a copying network. One exemplar per volume, in one library, with no classroom or liturgical demand, is in-degree one by design; such a set does not fade by neglect - it waits, intact, for a single shelf-scale event. The 53-to-4 attrition with median witness count one is the cleanest laboratory the Greek tradition offers for the survival mathematics of unreplicated state compilations, and every number in it is checkable against Pinakes' witness records and the Berlin editions' prefaces.

Prediction clause (verbatim)

Prediction: in Pinakes, the surviving Excerpta Constantiniana collections will number at most four; their combined independent pre-1500 Greek witnesses will number at most five; the median witness count per surviving collection will be one; and at least one surviving collection will be transmitted solely as a palimpsest undertext while at least one is transmitted solely through post-1500 apographs of a destroyed codex (primary clause: at most 4 collections, at most 5 pre-1500 witnesses, median 1; the verdict follows it); disambiguation: a witness is an independent pre-1500 Greek manuscript carrying the collection - apographs, whether of surviving or destroyed exemplars, are excluded from the count and reported separately; coverage guard: the attested total of fifty-three must be verifiable in the De legationibus proem in de Boor's edition, else void.

Kill-dataset (verbatim)

Kill: Pinakes/diktyon witness records for the four collections, checked against the prefaces of the Berlin editions (de Boor 1903 and 1905; Buettner-Wobst/Roos 1906-1910; Boissevain 1906) - a collection-level witness census set against the proem's own fifty-three.

Novelty / leakage triage

already answered in the literature

Every clause of the primary is stated in the Berlin prefaces and the surveys: the proem's fifty-three, the four surviving collections, the lone Peirescianus (Tours, BM 980) behind the De virtutibus, Vat. gr. 73's undertext as the sole De sententiis, the two partial De insidiis witnesses, and the De legationibus alive only in the 1570s apographs of the Escorial codex burnt in 1671; Lemerle rehearses the 53-to-4 arithmetic and Nemeth re-tabulates the witnesses. A Pinakes re-count adds no number the prefaces have not printed.

• C. de Boor (ed.), Excerpta de legationibus (Berlin, 1903), praefatio
• Th. Buettner-Wobst & A.G. Roos (eds.), Excerpta de virtutibus et vitiis (Berlin, 1906-1910), praefatio
• U.Ph. Boissevain (ed.), Excerpta de sententiis (Berlin, 1906), praefatio
• P. Lemerle, Le premier humanisme byzantin (Paris, 1971), chapter on the Constantinian encyclopedism

Predictions

No prediction registered yet.