The catalogue of scales the concert hall forgot

Claim (verbatim)

Indian art music is usually told as deep continuity, but its modal inventory turns over. The thirteenth-century Sangitaratnakara of Sarngadeva - the great synthesis just before the Persianate reshaping of North Indian music - catalogues in its Ragavivekadhyaya a large system of ragas: grama-ragas, uparagas, ragas proper, bhashas and their kin, well over two hundred named modal entities. The performed canon of modern Hindustani and Carnatic music also runs to a few hundred ragas - but mostly different ones. If raga transmission were continuous, the medieval catalogue's names would populate the modern stage; instead the systems are largely name-disjoint, because a raga is a performance category, alive only while sung, and a name notated in a treatise but dropped from performance leaves no sounding trace to recover. The mechanism is the same silence that swallows any oral repertoire, sharpened by a documented system-change: the marga/desi and grama-raga framework of Sarngadeva gave way to the mela and thata frameworks of the following centuries. Prediction restated: fewer than a fifth of the ragas named in the Sangitaratnakara survive by name in the modern performed canon, and the still-older grama-raga names of Matanga's Brhaddeshi survive at a lower rate still - modal mortality rising monotonically with the age of the treatise.

Prediction clause (verbatim)

Prediction: matching the raga-names catalogued in the Sangitaratnakara's Ragavivekadhyaya against a standard modern performed-raga census (Hindustani and Carnatic), under 20 percent of the Sangitaratnakara raga-names will appear in the modern canon, and the grama-raga names of Matanga's Brhaddeshi will match the modern canon at a still-lower rate. Primary clause: Sangitaratnakara-to-modern name-survival is under 20 percent. Disambiguation: matching is by raga-name after standard Sanskrit-to-modern orthographic normalization, and a modern raga counts as a match only on name identity, not on speculative scale-equivalence. Coverage guard: if the Sangitaratnakara raga-list cannot be extracted to at least 150 names, the test voids.

Kill-dataset (verbatim)

Kill: the GRETIL/SARIT e-text of the Sangitaratnakara (Adyar Library edition) for the medieval raga-list and Matanga's Brhaddeshi (Trivandrum Sanskrit Series 94; IGNCA edition) for the grama-ragas, against a standard modern raga census (the raga indexes of the Hindustani and Carnatic reference literature) - count name-matches by treatise.

Novelty / leakage triage

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

Radical name-discontinuity between the medieval catalogues and the performed canon is the published consensus - Powers' New Grove treatment of the raga systems and Widdess's monograph on the early ragas both stress that the grama-raga framework died and that continuity even of names is thin, and te Nijenhuis tabulates the treatise lists - but no located source runs the Sangitaratnakara's roughly 264 names against a modern performed-raga index to a percentage, and the count sits genuinely near the 20-percent boundary (names like Vasanta, Hindola, Varati and Gurjari do persist), so the threshold is a live test rather than guaranteed arithmetic. The Brhaddesi-lower monotonicity clause is likewise asserted qualitatively, never counted.

Predictions

No prediction registered yet.