From Bactria to the Balkans to Oaxaca
11 October 2026 · Nabu news
Two wings opened and a third began. The Middle Iranian shelf is now complete as TITUS hosts it: five grant-gated corpora — the Pahlavi books, the Manichaean corpora, Sogdian, Khotanese Saka and Bactrian — joined Tocharian A in one wave, each retrieved once under its personal grant, census-fixed against the real pages, and verified to a clean parse. Not one of the 191 first-sync surprises (unknown editorial lanes, severed UTF-8, the editors’ own “not yet read” entries) aborted a sync: every one quarantined honestly, got its census, and recovered. Bactrian’s dated documents now carry real CE envelopes — era year by era year through the ruled [+222, +228] window that spans the contested epoch, down to the Tang-i Safedak inscription’s year 492 — and 417 of the Tocharian A leaves know their findspot.
The Slavic opening ran end to end across two phases. Nine wikisource cones — Serbian epic song to Old East Slavic chronicle — censused category by category with written exclusion rules (the Russian cone’s honest yield: 12 texts, after 31 pages keyed to in-copyright editions were turned away). Then the completion wave: the KorBa manually annotated Middle Polish subcorpus and the 1488 Prague Bible ground truth; KDSP’s long prose and the Protestant prints of Trubar and Bohorič in page OCR; the Serbian ELTeC novels, the Croatian school canon and the ijekavian Vuk–Daničić Bible, each work admitted on its own public-domain record; the complete Ipatiev chronicle from the 1908 Petersburg edition with the editors’ Soviet-era volumes excluded by rule; Nasovič’s 1870 dictionary opening the Belarusian shelf; and the German Slavistics canon — Miklosich to Jagić’s Archiv — as a reference shelf with its OCR quality measured per volume. Riding along: Brückner’s 1927 etymological dictionary of Polish as a real, queryable shelf, and the registry grew its West and South Slavic families with the Polish and Czech stage ladders.
And the Americas. The library’s first American wing opens with the Ticha corpus of colonial Valley Zapotec — Córdova’s 1578 Arte through two centuries of notarial testaments, each document postured honestly as original or period translation — Guaman Poma’s Nueva corónica entire from the Royal Danish Library’s public-domain edition, and the Jesuit-era Tupinambá and Old Guaraní treebanks. The registry minted the Oto-Manguean and Tupian families to carry them, and the asks are out for the biggest prize of all: the full machine-readable Florentine Codex.
By the numbers: twenty-four new text sources across the three phases, two new dictionary shelves (Brückner’s Polish etymologicum, Nasovič’s Belarusian), one new research desk, and the kotenseki crawl past a quarter of its 97,731 books. The library stands, as of this writing, at 2.35 million documents and 110 million passages across 203 corpus sources on 28 desks (the live census has since moved on) — and the whole wave rode the quarantine belts without a single aborted sync.
As always: canonical means canonical — every text verbatim from its source, every exclusion a written rule, every license the grantor’s own words.