AI-generated conjecture · a wager, not a finding
← All conjectures · Philosophy, theology & the schools
Foreign names rot fastest
Status is derived only from the shepherd-authored triage/prediction data above -- community submissions and claims are a separate overlay and can never change it (see the participation panel below).
Claim (verbatim)
The Martyrologium Hieronymianum is what a text looks like after five centuries of copying without comprehension. Compiled in northern Italy in the mid-fifth century and surviving through a Gallic recension of around 600 (the oldest witnesses - the Echternach codex, Bern, Wolfenbuettel - stand within a century of that), it is, for each day, a bare string of names and places swept in from Greek, Syrian, and African calendars. Strings without stories cannot be error-corrected: scribes split one martyr into two, promoted place-names into persons, dragged entries onto neighboring days, and re-imported whole lists as duplicates. Delehaye and Quentin's Commentarius perpetuus (Acta Sanctorum Novembris II.2, 1931) is the monumental autopsy, diagnosis by diagnosis. The information-theoretic claim: the decay was not uniform but tracked the scribe's distance from the name's home. Gallic and Italian entries could be checked against living cult and local knowledge - redundancy that corrects noise - while the African and Eastern lists, foreign names in foreign spellings copied in Frankish scriptoria, accumulated damage unchecked. The martyrology is thus a five-century channel-noise experiment with a built-in control group, and the Bollandists' commentary has already scored every trial; nobody has simply counted it by region.
Prediction clause (verbatim)
Prediction: for the pinned sample of the January and July day-lists, classifying each name-entry by home region and scoring it corrupt where the Delehaye-Quentin Commentarius diagnoses doubling, splitting, place-turned-person, or dislocation, the corruption rate of African-plus-Eastern entries will exceed that of Gallic-plus-Italian entries by at least 2:1 (primary clause; the verdict follows it); entries the Commentarius leaves undiagnosed count as clean, biasing against the claim; the test voids if the two months yield fewer than 200 classifiable entries.
Kill-dataset (verbatim)
Kill: the Martyrologium Hieronymianum in the De Rossi-Duchesne edition (Acta Sanctorum Novembris II.1, 1894) with the Delehaye-Quentin Commentarius perpetuus (Novembris II.2, 1931) - a per-entry regional corruption census over the pinned months, scored strictly by the Commentarius's own diagnoses.
Nobody has run this test. The kill-data is named above. If you can run it — or you know the paper that already settles it — claim the kill or submit the prior scholarship. Kills and prior scholarship are credited here, by name, as they come in.
Provenance
Run: Fresh agent generation · model: claude-fable-5
Fresh blind generation by claude-fable-5, 2026-07-17, hagiography wave anchored to the Bollandist instruments: every kill names BHL/BHLms, the Acta Sanctorum, a martyrology edition (Dubois's Usuard, De Rossi-Duchesne/Delehaye-Quentin Hieronymian, Quentin's Bede), or the comparative repertoria (BHG/BHO, Pinakes), with a countable operation - attestation-vs-text subtraction, items-per-saint inequality, codex co-occurrence, witness-count asymmetry along version chains, witness age-profiles, unica censuses, genre stratigraphy, cross-repertorium translation tallies, corruption and name-retention ledgers, edition-date gradients, sole-edition and sole-witness counts. Disjointness from the owned w08 medieval-religion ground checked by grep and by hand: w08 #17 owns the BHL-version-count-vs-translationes correlation - items 5 and 9 here reuse BHL/AASS as instruments but under different operations (original-vs-rewrite witness asymmetry; Frankish genre stratigraphy without any relic-mobility correlation); w08 #18 owns miracle-type distance mixes (untouched); w08 #35 owns breviary lesson-length compression - item 6 here measures witness age-profiles across the Legenda aurea bottleneck instead; w11 #24/#27/#28 own the Greek-side Synaxarion/Metaphrast operations - item 10 here stays on repertorium cross-references (translation direction), not witness sweeps; w23 #13 owns female patronage via DBBE - item 8 here is a Latin attestation-vs-text subtraction. No other packet names Usuard, the Hieronymian, Voragine/Fleith, Mombritius, BHG/BHO-as-instruments, Gregory's Gloria martyrum, the MLA, or the AASS calendar cliff (grep across all fresh packets, 2026-07-17). Candidates considered and dropped: suppressed-cult text fates (no countable pre-print corpus - Guinefort-type cults are known precisely through inquisitors, not dossiers); a BHL never-printed-backlog census (folded into items 14-15); a Latin parallel to the Metaphrastic sweep (too close to w11 #28's mechanism). Honesty flags: MODERATE on the Gloria martyrum 63 Patroclus chapter number and anecdote detail (story standard in the cult-and-text literature via Van Dam, but the census clause depends only on Krusch's index, not on this example); MODERATE on item 7's 5% unica rate and item 15's 4% sole-edition rate (deliberately risked far-from-zero thresholds; the operations stand regardless); MODERATE on the exact count of Bede's authentic entries ('hundred-odd', guarded by the void clause) and on the Saint-Pere-en-Vallee attribution of the burned Chartres legendary (color, not load-bearing). Coverage guard held throughout: BHLms indexes the Bollandist-catalogued collections (Belgium-France-Rome skew), so all witness-count clauses are read as counts-in-BHLms, never as absolute survival; items 4 and 16 carry 'Kill (partly not yet built)' because their decisive joins (dated-version tables; destroyed-shelfmark concordances) are real published scholarship not yet assembled into datasets.
Novelty / leakage triage
anticipated in the literature — this exact test has never been run
The instrument has already scored every trial: the 1931 Commentarius perpetuus diagnoses doubling, splitting, place-into-person, and dislocation entry by entry across the Hieronymianum, and the mangling of foreign names is a recurring theme of its diagnoses. What no one has done is classify the entries by home region and count - the 2:1 corruption ratio is unrun arithmetic over a printed diagnosis table, with the undiagnosed-counts-clean rule biasing against the claim. Register-before-compute: the regional classification layer is the only new construction.
- G.B. De Rossi & L. Duchesne (eds.), Martyrologium Hieronymianum (Acta Sanctorum Novembris II.1, Brussels, 1894)
- H. Quentin (recension) & H. Delehaye (commentary), Commentarius perpetuus in Martyrologium Hieronymianum (Acta Sanctorum Novembris II.2, Brussels, 1931)
Predictions
No prediction registered yet.
Weigh in
No community feedback yet.
New here? Create an account first
Create an account or sign in and your feedback is tied to you — you can track it, get replies, and claim this conjecture so others know you’re working on it. Prefer not to? Just leave your take below as a guest — only the name you type is shown.
Add your take
Posted immediately (spam is removed). Community feedback is never an adjudicated verdict and never changes this conjecture's triage label or status above.