First the text, the melody much later or never

Claim (verbatim)

The earliest antiphoners are silent books. The Compiegne antiphoner (Paris, BnF, lat. 17436), copied around 870, gives the texts of the office and mass chants for the whole year and not a single note of music - it is one of the unnotated witnesses on which Hesbert built the Corpus Antiphonalium Officii, a chant book that records what was sung but not how. Neumes do not appear in the surviving record until a generation or two later, and pitch-readable notation of the office not until the eleventh and twelfth centuries. So for the core office repertory the first attestation is textual and the melodic attestation is late, sometimes by two centuries, and for a residue of chants named in these early text-antiphoners no securely readable melody survives at all. The mechanism is that the text was fixed and administratively circulated before a notation existed to carry the tune, so the database's melody record begins long after the repertory's documentary record, and the gap between them is a measurable stratum of melodies that were sung, then named in writing, and only much later - or never - fixed as pitches. Prediction restated: for office chants whose earliest witness is an unnotated text-antiphoner such as Compiegne, the median gap to the first pitch-readable melodic witness exceeds 150 years, and a nonzero fraction acquire no diastematic melody at all.

Prediction clause (verbatim)

Prediction: for office chants first attested in the unnotated text-antiphoners of the Corpus Antiphonalium Officii (Compiegne, BnF lat. 17436, and its congeners), the median interval between first textual attestation and first pitch-readable (diastematic) melodic witness in the Cantus Database exceeds 150 years, and at least 2 percent acquire no diastematic melodic witness at all (primary clause: the greater-than-150-year median text-to-melody lag; the verdict follows it). First melodic witness counts only as heighted, pitch-readable notation, not adiastematic neumes; coverage guard: the lag is a property of surviving witnesses and later discoveries can only shorten it, so a confirmed median above 150 years is robust, the test voiding if datable melodic witnesses are found for fewer than 200 of the text-antiphoner chants.

Kill-dataset (verbatim)

Kill (partly built): the Corpus Antiphonalium Officii inventory of the unnotated text-antiphoners (Compiegne and congeners) as the textual-attestation baseline, against the earliest diastematic melodic witness per chant in the Cantus Database; the operation is a text-to-melody attestation-lag distribution.

Novelty / leakage triage

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

The stratigraphy is textbook - Compiegne c. 870 is unnotated, Hesbert's CAO and the Antiphonale Missarum Sextuplex rest on such text-books, and pitch-readable notation of the office belongs to the eleventh and twelfth centuries - but the pinned statistic is a per-chant median interval between first textual and first diastematic witness across the CAO repertory, a date-lag distribution nobody has computed; the surveys' qualitative 'gap of one to two centuries' does not settle a median over thousands of chants, nor the 2-percent never-notated residue. This is the discipline's named case of un-run date-lag-median arithmetic.

• R.-J. Hesbert, Corpus Antiphonalium Officii I: Manuscripti 'cursus romanus' (Rome, 1963)
• R.-J. Hesbert, Antiphonale Missarum Sextuplex (Brussels, 1935)
• K. Levy, Gregorian Chant and the Carolingians (Princeton, 1998)
• D. Hiley, Western Plainchant: A Handbook (Oxford, 1993), notation chapter

Predictions

No prediction registered yet.