A famous name is a magnet for orphans

Claim (verbatim)

In the Chinese Buddhist catalogues the number of translations credited to the second-century Parthian pioneer An Shigao (安世高) grows across time - the opposite of what loss alone would do. Dao'an's Zongli zhongjing mulu (道安 綜理眾經目錄) of 374, preserved inside Sengyou's Chu sanzang jiji (僧祐 出三藏記集, Taisho 2145), listed only a few dozen An Shigao titles; by the Kaiyuan shijiao lu (開元釋教錄, 730, Taisho 2154) the attributed count has swelled past a hundred, and modern philology (Zürcher; Nattier's Guide to the Earliest Chinese Buddhist Translations) accepts only a small core as genuinely his. The mechanism is orphan-adoption: an early anonymous translation, once its true shiyi (失譯, translator-lost) status becomes inconvenient, drifts onto the most famous available pioneer, because a canon prefers a named father to an unknown one. The signature is a monotone growth of attribution against a stable genuine core - and it should be a property of the early pioneers specifically, not of the bureau-era masters whose output was documented in real time. Prediction restated: the Kaiyuan lu credits An Shigao with at least three times as many works as Dao'an did, while a bureau-era translator's count barely moves across the same catalogue span. Attribution inflation is a force that pulls toward fame, and it is strongest where documentation was weakest.

Prediction clause (verbatim)

Prediction: tracking the count of works attributed to An Shigao (安世高) across the catalogue chain in CBETA, the Kaiyuan shijiao lu's (開元釋教錄) An Shigao count will exceed Dao'an's fourth-century count (as preserved in the Chu sanzang jiji) by at least a factor of three, while the count attributed to the bureau-era translator Xuanzang (玄奘) grows by less than 20% across the same catalogue span (primary clause: the >=3x growth for An Shigao against <1.2x for Xuanzang; the verdict follows it); attributions are counted from each catalogue's own translator-headed entries.

Kill-dataset (verbatim)

Kill: the CBETA XML-P5 corpus texts of Sengyou's Chu sanzang jiji (T. 2145, carrying Dao'an's list) and the Kaiyuan shijiao lu (T. 2154), counting attributed titles per named translator in each, with the SAT Daizokyo full-text database as the cross-check on the extant genuine core.

Novelty / leakage triage

already answered in the literature

This primary clause is guaranteed by the published record. The monotone growth of An Shigao attributions — Dao'an's fourth-century list (preserved in Sengyou's Chu sanzang jiji) of a few dozen titles against the Kaiyuan lu's count swollen well past a hundred — is precisely the tabulation Zürcher and Nattier put in print, and Nattier's Guide fixes the small genuine core against which the inflation is measured. Xuanzang's contemporaneously documented, stable corpus is the standard textbook contrast. The >=3x-for-An-Shigao / <1.2x-for-Xuanzang comparison is therefore already tabulated (An Shigao side) and guaranteed (Xuanzang side) in the very scholarship the conjecture itself cites; only the exact recomputation in CBETA remains. Leaked.

• E. Zürcher, The Buddhist Conquest of China (Leiden: Brill, 1959)
• Jan Nattier, A Guide to the Earliest Chinese Buddhist Translations: Texts from the Eastern Han and Three Kingdoms Periods (Tokyo: IRIAB, Soka University, 2008)

Predictions

No prediction registered yet.