Triangulating the drowned ocean

Claim (verbatim)

Guṇāḍhya's Bṛhatkathā — the vast lost story-ocean in Paiśācī — survives only through its descendants: two eleventh-century Kashmiri verse abridgements (Kṣemendra's Bṛhatkathāmañjarī, c. 1037, and Somadeva's Kathāsaritsāgara, c. 1070), the earlier Nepalese Bṛhatkathāślokasaṃgraha of Budhasvāmin, the Jain Vasudevahiṇḍi of Saṅghadāsa, the Tamil Peruṅkatai — plus a Sanskrit rendering credited to the Gaṅga king Durvinīta in copper-plate praśastis, attested and itself lost. Lacôte's Essai (1908) argued that the two Kashmiris worked not from Guṇāḍhya but from a shared Kashmiri intermediary. That hypothesis has an order signature: descent through a common intermediary locks shared structure, so the two Kashmiri texts should agree on the sequence of shared episodes far beyond either one's agreement with Budhasvāmin — and lost-text philology becomes a computable stemma problem on surviving witnesses. Prediction: aligning the narrative episodes shared across the Kathāsaritsāgara, the Bṛhatkathāmañjarī, and the Bṛhatkathāślokasaṃgraha, the rank correlation of episode order between the two Kashmiri texts will exceed each one's correlation with Budhasvāmin's text by at least 0.3 Kendall's tau, and no substantial shared block will show a Kashmiri text siding with Budhasvāmin on order against the other Kashmiri (primary clause: the tau gap of at least 0.3; the verdict follows it). Kill: the printed editions and e-texts — the Kathāsaritsāgara (Durgāprasād's Nirṇaya Sāgara edition; GRETIL e-text), the Bṛhatkathāmañjarī (Kāvyamālā edition), and Budhasvāmin's Bṛhatkathāślokasaṃgraha (Lacôte's edition; Mallinson's Clay Sanskrit Library text, The Emperor of the Sorcerers, 2005) — with the episode concordances in Lacôte's Essai sur Guṇāḍhya et la Bṛhatkathā (1908) as the alignment baseline.

Prediction clause (verbatim)

Prediction: aligning the narrative episodes shared across the Kathāsaritsāgara, the Bṛhatkathāmañjarī, and the Bṛhatkathāślokasaṃgraha, the rank correlation of episode order between the two Kashmiri texts will exceed each one's correlation with Budhasvāmin's text by at least 0.3 Kendall's tau, and no substantial shared block will show a Kashmiri text siding with Budhasvāmin on order against the other Kashmiri (primary clause: the tau gap of at least 0.3; the verdict follows it).

Kill-dataset (verbatim)

Kill: the printed editions and e-texts — the Kathāsaritsāgara (Durgāprasād's Nirṇaya Sāgara edition; GRETIL e-text), the Bṛhatkathāmañjarī (Kāvyamālā edition), and Budhasvāmin's Bṛhatkathāślokasaṃgraha (Lacôte's edition; Mallinson's Clay Sanskrit Library text, The Emperor of the Sorcerers, 2005) — with the episode concordances in Lacôte's Essai sur Guṇāḍhya et la Bṛhatkathā (1908) as the alignment baseline.

Novelty / leakage triage

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

Lacote's shared-Kashmiri-intermediary hypothesis is the classic position, and the qualitative comparison of the three versions' arrangements (the two Kashmiris against Budhasvamin) was worked through by Lacote and Speyer. No one has quantified episode-order concordance as a rank-correlation test across the three witnesses.

• F. Lacote, Essai sur Gunadhya et la Brhatkatha (Paris, 1908)
• J.S. Speyer, Studies about the Kathasaritsagara (Koninklijke Akademie van Wetenschappen, Amsterdam, 1908)

Predictions

No prediction registered yet.