Colophon

The single most important source feeding this catalogue is Open Library, the public-domain bibliographic project run by the Internet Archive. Where we know the ISBN of a book, we walk Open Library’s Works → Editions hierarchy to find the first English-language printing’s cover; where an author exists in their catalogue, we resolve our messy listing-supplied names against their canonical record.

Open Library is the kind of infrastructure the rest of the internet should be made of: nonprofit, public-domain (CC0), patient, librarian-tended. It does not extract value from its users. It does not gate its data behind APIs. It has been quietly compounding for almost twenty years and is one of the few comprehensive book catalogues outside the control of Amazon.

What we give back.Where our editorial work would materially improve a record — an author with no biography on Open Library where we have a substantive one, missing birth or death years, no link to a wider catalogue presence — we submit a single careful update per day via their authenticated edit API. Every edit carries an honest _commentnoting the source of the change. Our contributions enter Open Library under the same CC0 terms we received their data under — the work compounds.

The Internet Archive is currently fighting an existential legal battle with major publishers over its Controlled Digital Lending program. If you ever find this site useful, the most direct way to repay that good is to donate to the Archive. We do.

Author photographs, biographies, and the prose used as the seed for our author pages are drawn from the English Wikipedia’s biography articles and their associated infobox images on Wikimedia Commons. Structured facts about authors — lifespans, canonical name forms, public-figure identifiers — come from Wikidata, including the P18 "image" claim that backs most author portraits in this catalogue.

Wikipedia text is used under CC BY-SA 4.0; Wikimedia Commons media under each file’s respective licence (almost always CC BY-SA or public domain). Wikidata is CC0. Where we quote or closely paraphrase Wikipedia prose in an author bio, the attribution travels with the text.

When a listing arrives with an ISBN we don’t recognise, the Google Books API provides the bibliographic skeleton — title, authors, publisher, year — that we hand to our classifier. Google never sees a buyer or seller; it sees only the ISBN we’re asking about. We do not display Google’s cover thumbnails because they tend to be the cover of whatever printing happens to share that ISBN, which is rarely the one a rare-book buyer wants to see.

We do not host listings of our own. Every copy you see on offer originated on abebooks or eBay, and clicking through to a seller takes you to the original marketplace, where the actual transaction happens. We are a discovery surface, not a market.

Listings are observed via abebooks’ public search pages and eBay’s Browse API. We respect both sources’ rate limits, identify our crawler honestly in the user-agent string, and re-check listings periodically to keep the catalogue from showing copies that have already sold.

A handful of cataloguing tasks lean on language models accessed through OpenRouter: classifying books by collectibility tier, generating an English author bio where neither Open Library nor Wikipedia provides one, and normalising authorial name variations into a single canonical form. The model choices are not secret — current pipeline runs primarily use Anthropic’s Claude family for prose and Xiaomi’s MiMo for structured classification.

Where machine-generated text would mislead, we say so: any author bio synthesised by a model is tagged bioSource = llm-generated in the underlying database. Bios sourced from Open Library or Wikipedia are tagged accordingly and rendered with their attribution intact.

The site is a Next.js application backed by PostgreSQL through Prisma. Layout is composed with Tailwind CSS; the alert emails are React Email templates delivered via Resend.

The body and display face is Newsreader by Production Type, an open-source serif designed for screen reading. Caps and small annotations are set in JetBrains Mono. Both are used under their respective open licences and shipped via the Next.js font pipeline so no third-party CDN sees your traffic.

The work of a small rare-book catalogue is mostly the work of standing on other people’s shoulders — librarians, curators, archivists, and the volunteer cataloguers of Open Library and Wikipedia who have spent decades building the public-domain infrastructure this site depends on. We owe them more than a line in a colophon, but a line in a colophon is at least a start.

See also the privacy notice for how the small amount of data we do keep about you is handled.