The hymns had an engine; the classics had luck

Claim (verbatim)

Classical Tamil is India's second classical literature, and it nearly failed to arrive. When U. V. Swaminatha Iyer and C. W. Damodaram Pillai went hunting in the 1880s, the Caṅkam corpus — the Eṭṭuttokai anthologies and the Pattuppāṭṭu — survived in a scatter of late palm-leaf copies, often a handful of usable witnesses per text, some works recovered from what amounted to a single line of transmission; Swaminatha Iyer's autobiography is a catalogue of last copies found in attics and monasteries. The devotional canon shows the opposite profile: the Tēvāram and the Nālāyira Tivviyappirapantam, installed in temple recitation with endowed singers, survive in witnesses by the hundred. The differential is the mechanism. Liturgy is a copying engine that never idles; a courtly-scholastic curriculum copies only while the school lives, and the Caṅkam texts had already fallen out of curriculum once, surviving on commentarial residue. Eva Wilden's census of the surviving Caṅkam manuscripts makes the comparison computable, witness by witness. Prediction: counting distinct pre-1900 manuscript witnesses per work, the median for the Eṭṭuttokai anthologies will be under ten, at least one canonical anthology will rest on three or fewer independent witnesses, and the median witness count for the Tēvāram hymn corpus will exceed the Caṅkam median at least tenfold (primary clause: the tenfold median gap; the verdict follows it). Kill: Eva Wilden, Manuscript, Print and Memory: Relics of the Caṅkam in Tamilnadu (Berlin, 2014) for the Caṅkam witness census, with the witness registers in the U. V. Swaminatha Iyer edition prefaces, the Descriptive Catalogues of the Government Oriental Manuscripts Library (Madras), and the Digital Tēvāram of the IFP/EFEO (Pondicherry, 2007) for the devotional side.

Prediction clause (verbatim)

Prediction: counting distinct pre-1900 manuscript witnesses per work, the median for the Eṭṭuttokai anthologies will be under ten, at least one canonical anthology will rest on three or fewer independent witnesses, and the median witness count for the Tēvāram hymn corpus will exceed the Caṅkam median at least tenfold (primary clause: the tenfold median gap; the verdict follows it).

Kill-dataset (verbatim)

Kill: Eva Wilden, Manuscript, Print and Memory: Relics of the Caṅkam in Tamilnadu (Berlin, 2014) for the Caṅkam witness census, with the witness registers in the U. V. Swaminatha Iyer edition prefaces, the Descriptive Catalogues of the Government Oriental Manuscripts Library (Madras), and the Digital Tēvāram of the IFP/EFEO (Pondicherry, 2007) for the devotional side.

Novelty / leakage triage

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

Wilden's census documents the Cankam's threadbare transmission (works surviving on a handful of late witnesses) and the contrast with the temple-sustained devotional corpus is a standard trope of Tamil literary history; but the two-sided witness-count comparison - per-work medians for the Ettuttokai against the Tevaram with a tenfold gap - was not located as ever having been computed on catalogued witnesses.

• E. Wilden, Manuscript, Print and Memory: Relics of the Cankam in Tamilnadu (Berlin, 2014)
• Digital Tevaram / Kaninit Tevaram, ed. V.M. Subrahmanya Ayyar, J.-L. Chevillard & S.A.S. Sarma (IFP/EFEO, Pondicherry, 2007)

Predictions

No prediction registered yet.