A thousand monks, zero leaves

Claim (verbatim)

Srivijaya ran one of medieval Asia's great Buddhist study centers, and we have an eyewitness ledger of its book culture. The pilgrim Yijing stopped there in 671 and returned in the late 680s to write and translate; his memoirs — composed in Srivijaya and sent home — report over a thousand monks whose learning matched India's, and record his own supply run back to Guangzhou for assistants, paper, and ink to copy texts in the southern capital. Three centuries on, Atiśa spent about twelve years (traditionally 1012-1024) there with Dharmakīrtiśrī of Suvarṇadvīpa, whose Abhisamayālaṅkāra commentary, the Durbodhāloka, carries a colophon placing its composition in the city of Śrīvijaya under Cūḍāmaṇivarmadeva, around 1005. A copying economy importing Chinese paper and ink by the shipload, sustained across at least four centuries, implies output in the tens of thousands of fascicles. But the survival mechanism was purely export: humid-tropics palm leaf and paper die within a century or two without recopying institutions, and Srivijaya's monasteries died with the polity. The entire textual output of Srivijayan Buddhism should therefore survive only where it was carried — Tibet and China — at a rate of roughly one work per century of operation. Prediction: a census of author-attributed texts of Srivijayan Buddhism surviving anywhere will find fewer than five works, every one preserved by export — at least two works under gSer-gling-pa's name in the Tanjur, including the Durbodhāloka with its Suvarṇadvīpa composition colophon, plus Yijing's two Srivijaya-written memoirs in the Chinese canon — and exactly zero manuscripts or fascicles of any date before 1300 surviving in Sumatra or anywhere in the archipelago in local custody (primary clause: the fewer-than-five works census with zero locally surviving manuscripts; the verdict follows it). Kill: the Tōhoku Complete Catalogue of the Tibetan Buddhist Canons and the rKTs Kanjur-Tanjur database (rkts.org) searched for gSer-gling-pa attributions, with J. Takakusu's translation of Yijing, A Record of the Buddhist Religion as Practised in India and the Malay Archipelago (Clarendon, 1896), and Peter Skilling's studies of Dharmakīrtiśrī of Suvarṇadvīpa as control.

Prediction clause (verbatim)

Prediction: a census of author-attributed texts of Srivijayan Buddhism surviving anywhere will find fewer than five works, every one preserved by export — at least two works under gSer-gling-pa's name in the Tanjur, including the Durbodhāloka with its Suvarṇadvīpa composition colophon, plus Yijing's two Srivijaya-written memoirs in the Chinese canon — and exactly zero manuscripts or fascicles of any date before 1300 surviving in Sumatra or anywhere in the archipelago in local custody (primary clause: the fewer-than-five works census with zero locally surviving manuscripts; the verdict follows it).

Kill-dataset (verbatim)

Kill: the Tōhoku Complete Catalogue of the Tibetan Buddhist Canons and the rKTs Kanjur-Tanjur database (rkts.org) searched for gSer-gling-pa attributions, with J. Takakusu's translation of Yijing, A Record of the Buddhist Religion as Practised in India and the Malay Archipelago (Clarendon, 1896), and Peter Skilling's studies of Dharmakīrtiśrī of Suvarṇadvīpa as control.

Novelty / leakage triage

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

That Srivijaya survives archaeologically and epigraphically but left no local manuscript corpus is a commonplace of Indonesian scholarship since Coedes, and gSer-gling-pa's (Dharmakirtisri of Suvarnadvipa's) works in the Tanjur plus Yijing's Srivijaya-composed memoirs in the Chinese canon are catalogued export survivals any competent reader can name. But the primary clause is a bounded census - author-attributed texts of Srivijayan Buddhism surviving anywhere, fewer than five, all by export, zero pre-1300 local manuscripts - and that enumeration over a scope that must first be defined has not been assembled and published as such. The zero-local-manuscripts half is near-guaranteed by the known material record, but the works-count is the conjecture's own arithmetic. The neighborhood is thoroughly worked; the specific census is not.

Predictions

No prediction registered yet.