The fourth canon: Tibet takes the twenty-third desk

28 July 2026 · Nabu news

The library’s Buddhist desk has read the dharma in Pali, Sanskrit, and Chinese; as of today it reads it in Tibetan. The complete Derge canon — the Kangyur’s 1,200 texts and the Tengyur’s 3,362, some 1.36 million woodblock lines — arrives in Esukhia’s Public-Domain exact-representation of the Degé blocks, cited by Tohoku number at folio-line grain, mistakes of the carvers preserved as carved.

Around the canon, the full apparatus of a working desk, the twenty-third: 84000’s English translations (388 texts at the Reading Room’s own citation grain), the Old Tibetan documents — OTDO’s 413 critical editions spanning the Dunhuang scrolls, the imperial pillar inscriptions, and five texts in Old Zhangzhung — the Annals and Chronicle annotated and paired line-by-line with Dotson’s translation, the SOAS gold corpus, and a lexicon shelf anchored by the Mahāvyutpatti, the ninth-century imperial glossary whose 9,379 entries bridge Sanskrit, Tibetan and Chinese — so define bodhisattva and define bodhisattvaḥ now reach the same entries through a stem-aware Sanskrit lookup.

The phase’s charge was that Buddhist alignment “work to the fullest extent,” and the honest version turned out better than the easy one. Document-grain translation edges join every Kangyur text to its 84000 translation; the cross-canon parallels graph (195,287 edges) already pairs 261 Pali suttas with their Chinese āgama counterparts; and where a first look said the citation schemes were irreconcilable, a second look found 84000’s TEI carries 33,073 inline Degé folio anchors — so show <kangyur-text>:1b --parallel renders the Tibetan woodblock lines beside their English folio block, coverage honestly labeled, in either direction. What was refused is as deliberate as what was built: no fake line pairing, no translation-equivalents dressed up as cognates.

Desk cards now name their languages with counts, language cards name their desks, and the census stands at 968,182 documents / 68.3 million passages in 121 language codes, 1.4 million dictionary entries across 103 shelves.

Local, license-honest, rebuildable — from Sumer to the Land of Snows.

← All news  ·  Atom feed