The martyrology remembers more than the library

Claim (verbatim)

Around 865 Usuard, a monk of Saint-Germain-des-Pres, finished the martyrology that won: an abridgement of the whole prior tradition - the Hieronymian name-lists, Bede, Florus of Lyon, Ado of Vienne - trimmed to one commemoration string per day for reading in chapter, dedicated to Charles the Bald, and so successful that seven centuries later the Roman Martyrology of 1584 is still substantially Usuard. It commemorates thousands of saints by name, place, and day. The Bibliotheca Hagiographica Latina (1898-1901, with Fros's Novum Supplementum of 1986) indexes every known Latin narrative about a saint - vita, passio, translatio, miracula. Lay the census over the library and the difference is a measurement: a large fraction of Usuard's commemorations belong to saints with no BHL entry at all - no story, not a paragraph, nothing but the calendar line itself. These are mostly not books that burned; they are cults that never had books, name-and-place entries inherited from the Hieronymian's lists and carried for a millennium on liturgical momentum alone. The martyrology is an attestation census of the whole cult economy; the BHL is a census of its textual sector; the subtraction counts the saints who were venerated on memory's bare credit. Prediction restated: at least a third of a random sample of Usuard's commemorations resolve to no Latin dossier whatsoever.

Prediction clause (verbatim)

Prediction: drawing a fixed random sample of 200 distinct named commemorations from Usuard's martyrology in Jacques Dubois's edition (Le martyrologe d'Usuard, Subsidia Hagiographica 40, 1965) - one saint per commemoration, group entries of more than three names excluded, identity fixed as name plus place plus feast day as Dubois prints them - at least one third of the sample will match no numbered item in the Bibliotheca Hagiographica Latina or its Novum Supplementum under any spelling variant Dubois's apparatus supplies (primary clause: the >=1/3 zero-BHL rate; the verdict follows it). A text that merely mentions the saint inside another saint's dossier counts as no entry; the test voids as a coverage failure if more than 20 of the 200 cannot be disambiguated to a single saint.

Kill-dataset (verbatim)

Kill: Usuard's martyrology in Dubois's edition (Subsidia Hagiographica 40) sampled name by name against the BHL index sanctorum (the 1898-1901 volumes plus the 1986 Novum Supplementum) and the Bollandists' online BHLms index - a lookup with the match rule pinned in the prediction.

Novelty / leakage triage

anticipated in the literature — this exact test has never been run

That swathes of the martyrology's commemorations attach to saints who never had a text is the field's standing claim, and the nearest printed run exists: Delehaye's scholia in the 1940 Propylaeum annotate the Roman Martyrology - Usuard's direct descendant - entry by entry with each saint's documentation or the lack of it, and the Philippart-Trigalet per-century counts show how few commemorated saints carry dossiers. But no one has drawn a controlled sample from Dubois's Usuard and tabulated a zero-BHL rate under a pinned match rule, and the outcome is genuinely open: Usuard's abridgement is biased toward well-documented saints, so the >=1/3 clause could fail. At resolution the Propylaeum scholia must be treated as the adversary instrument, not as discovery.

• J. Dubois, Le martyrologe d'Usuard: texte et commentaire (Subsidia Hagiographica 40, Brussels, 1965)
• H. Delehaye et al., Martyrologium Romanum ad formam editionis typicae scholiis historicis instructum (Propylaeum ad Acta Sanctorum Decembris, Brussels, 1940)
• G. Philippart & M. Trigalet, 'Latin hagiography before the ninth century: a synoptic view', in J.R. Davis & M. McCormick (eds.), The Long Morning of Medieval Europe (Aldershot, 2008)

Predictions

No prediction registered yet.