The biggest single day — fifteen new sources, from Ganjoor to the Patrologia Graeca
2 November 2026 · Nabu news
Two development windows closed today, and together they are the largest single-day growth in the library’s history: 2,609,689 documents / 121,507,867 passages across 269 live sources (census 2 November 2026), up 1.7 million passages in a day.
The classical Persian library, whole. The operator of Ganjoor, the Persian poetry reference site, granted reuse of its data the day of the ask, and within hours its classical cone was on the shelves: 212 poets / 126,165 poems / 1,453,355 couplet-passages, from Rudaki and Ferdowsi through Nezami, ʿAṭṭār, Rumi, Saʿdi and Hafez to the Qajar divans, every poem dated by its poet’s life and most placed. Beside it landed the library’s first New Persian dictionary — English Wiktionary’s Persian lexicon (20,350 entries whose etymologies reach back through Middle and Old Persian) — so the new texts arrived with their lookup layer in the same window.
Byzantine Greek, a shelf from nothing. The Calfa/GREgORI Patrologia Graeca brought 33 volumes of Migne read by machine — 6.8 million lemmatized Greek tokens of Photius, Psellos, Skylitzes, Zonaras, Choniates, Gregoras and Palamas, uncorrected OCR declared as such — where the library previously held no medieval Greek at all.
The Semitic and biblical wave. The ETCBC’s Targum corpus (47 manuscripts of Aramaic targums with gold lemmas, the Cairo Genizah fragments among them) and Syriac corpus (31 Peshitta books with morphology), the Göttingen Hexapla database (Aquila, Symmachus and Theodotion beside the Septuagint, 169,420 tokens), and the 1688 Bucharest Bible — the first complete Romanian Bible, in its original Cyrillic — which also brought a Romanian branch to the lect registry.
Inner Asia and the excavated past. BDRC’s page-aligned Tibetan transcriptions (1,160 etext volumes / 99,571 woodblock and manuscript pages, each page tied to its scan, fetched text-only) and the first Warring States and Han bamboo-slip manuscripts: graph-level transcriptions of the Guodian and Peking University Laozi, the Anhui Odes and the Tsinghua chapters, every line aligned to its received text.
Dictionaries and instruments. Cleasby–Vigfusson and Zoëga give Old Norse its dictionary layer (65,158 entries); the 1914 Haga biographical dictionary adds 54,558 Japanese entries; the Göttingen RuneS database brings 613 runic finds read glyph by glyph; the Paramāra inscriptions and Comenius’s Czech devotional prose round out the wave. Two long-standing lookup defects fell in the same window: Welsh queries now reach the GPC under every spelling of the language’s code, and the OSHB lexicon’s 1,562 Aramaic entries answer Aramaic queries instead of hiding behind the shelf’s Hebrew label.
The morning window. Before the wave, today’s first gate delivered the demo image’s release rail (the public image is now multi-arch — amd64 installs work — with a workflow that builds, smokes and publishes it at release moments), a silver scansion lane (Latin and Greek metrical analysis validated at 88.1% and 94.6% agreement against held gold scansions, projected nowhere until deliberately run), the Church Slavonic Elizabeth Bible (76 books / 36,000 verses, verse-aligned to the library’s Bible works), the Penn Parsed Corpus of Historical Yiddish (63 hand-parsed texts, 1478–1959), and a column-anchored bridge from the Celtic dictionary shelf into Holder’s Alt-celtischer Sprachschatz — whose volume I was fetched page by page from the Bavarian State Library’s open IIIF service the same day.
The lect registry grew to meet the material: one aggregated package of 28 anchors is under review — a twelve-leaf Pacific spine, Vietnamese with its Nôm-era stages, Classical Mongolian and Written Oirat, Tangut, Yiddish with its Weinreich stages, and Romanian — each with researched classification notes. As always, the numbers above are the generated census of the live catalog, not hand-kept claims.