The Central Asia wave: two new desks on the Silk Road

28 September 2026 · Nabu news

The library turned east this week: the corridor between Iran and China — Khotan, Sogdiana, Bactria, the Tarim oases, the Mongol steppe — went from nearly empty to a working wing, and two new desks opened to serve it: the Turkologist and the Mongolist, the library’s twenty-sixth and twenty-seventh research axes.

The Iranian east. Three sources anchor the Middle Iranian lane. Skjærvø’s Khotanese Manuscripts from Chinese Turkestan in the British Library (Zenodo, CC BY 4.0) contributes 2,357 manuscript records with 14,574 transliterated lines — the first bulk Khotanese text anywhere in the library. The upstream file is a self-declared draft with about a hundred unclosed tags; rather than let one broken record poison its neighbors, the parser isolates each manuscript record as its own fragment and recovers it independently, so damage stays local and censused. Beside it, Oxford’s Invisible East Digital Corpus (Zenodo, CC BY 4.0) brings 1,316 catalog records of documents from the medieval Islamicate East — New Persian, Sogdian, Bactrian, Judeo-Persian, Middle Persian, Arabic — 481 of them with folio transcriptions, plus typed dates and located toponyms that feed the timeline and places lanes directly. And the Kaʿba-ye Zartošt inscription (ŠKZ) arrives sentence-aligned in all three of its versions — Middle Persian, Parthian, Greek — Šābuhr I’s res gestae as the library’s first Parthian document, fifty lines each readable against its two siblings.

The Turki manuscripts. The Turkologist’s desk opens on ATMO — the Annotated Turki Manuscripts from the Jarring Collection Online (Indiana University and Lund University Library, CC BY-SA 4.0): eighteen 18th–20th century Eastern Turki manuscripts transcribed line by line, Perso-Arabic text with a full Latin transliteration on every line, and even the marginalia and endorsements served with their provenance marked. Beside it sit new Wiktionary extracts for Old Turkic, Old Uyghur and Chagatai, and the first Turkic treebank — Old Turkic clausal examples under Universal Dependencies.

The Secret History. The Mongolist’s desk opens on John C. Street’s romanized text of the Secret History of the Mongols (v24, 2013), hosted by Monumenta Altaica — use granted by the site’s maintainer by email, with Street’s own copyright notice carried verbatim on every surface, in the spirit of the notice that the late professor “always wanted free propagation of his works.” One document, 3,523 lines at Street’s own line numbering, with section markers and footnote pointers preserved as annotations. The kaikki Manchu extract (2,577 entries) joins the same desk’s Mongolic-Tungusic lane, and Tocharian A and B dictionary extracts land alongside for the Tarim’s other language family.

And the Slavic south. One outlier rode the wave: Scripta Bulgarica (Bulgarian Academy of Sciences, CC BY-NC-SA), 124 medieval Bulgarian monuments — the Bitola inscription, charters, damaskini — each text carrying its printed-edition citation as provenance. The fragile upstream was mirrored snapshot-first, so the texts are safe whatever happens to the hosting.

A rhyme dictionary learns its descendant. The Qieyun (601 CE) and the Guangyun (1008 CE) — the two poles of the Chinese rime-dictionary tradition — are now cross-linked small rime by small rime: 3,359 curated correspondences from the restoration project’s own concordance table, so a lookup in one dictionary shows its counterpart in the other. The numbering shift the Guangyun’s added rimes introduce is exactly why the upstream table, not arithmetic, is the authority — and the twenty-six rimes present only in a rival restoration are counted honestly rather than guessed.

The adoption ledger. Housekeeping with teeth: the source registry is now alphabetically ordered by construction, and every source carries an explicit adoption date, backfilled from the project’s own history. nabu list --by-adopted reads the library’s growth as a timeline — newest arrivals first — and a standing census now flags any source whose fetched metadata carries date-, place- or genre-shaped fields that were never mined into the library’s axes, so unmined ore stays visible until it’s either extracted or explicitly declared out of scope.