Computational analysis of historical manuscripts, in the browser.
A quire is the bundle of folded parchment leaves from which every medieval codex was physically built — the unit manuscripts are made of.
QuireLab pairs a viewer for the IIIF image APIs that hundreds of libraries publish with a statistical laboratory for the texts themselves. The flagship dataset is the Voynich Manuscript (Beinecke MS 408, Yale): QuireLab reproduces its well-established statistical fingerprints — conditional character entropy, Zipf compliance, word-length structure, per-section vocabulary — against known-language baselines, entirely client-side. No server, no accounts, nothing uploaded.
It makes no decipherment claims. It makes the numbers behind the debate reproducible and explorable.
- Text Lab — H1/H2 conditional entropy (Bennett-style, space-joined stream), Zipf rank–frequency with fitted α, word-length distributions, top words, character frequencies, per-section Voynich statistics, and paste-your-own-text comparison.
- Manuscript Viewer — load any IIIF Presentation v2/v3 manifest (Yale, Gallica, British Library, …), browse thumbnails, lightbox with keyboard navigation and full-resolution links.
- About & Methods — tokenization and cleaning rules, entropy definitions, credits.
npm install
npm run dev- Voynich transliteration: ZL 3b (Zandbergen & Landini, May 2025) from voynich.nu, building on the EVMT project.
- Images: Beinecke Rare Book & Manuscript Library, Yale University, via public IIIF.
- Baselines: Alice's Adventures in Wonderland and De Bello Gallico I–IV (public domain, Project Gutenberg).
- Key literature: Montemurro & Zanette, PLoS ONE 2013; Bennett 1976.