THE CORPUS· — SOURCES·DOWNLOADABLE IN FULL
Check our work
Every claim on this site traces to a source. This is the list of them — all of them — with the identifier, the verification status, and a link that resolves to the document itself. Where we have not been able to verify something, this table says so in the same row, rather than in a footnote nobody reads. You can download the whole thing and audit it against the registries yourself. We would consider that a favour.
registry-verified — the identifier was retrieved from the issuing registry and every stored field agrees with that record. — entries currently carry this status.
url-only — a document without a DOI (a statute, an official report, a guideline) carrying a canonical link on the issuing body's own domain. — entries.
unresolved — we hold a record but cannot yet offer you a link that resolves. — entries, and we are not hiding the number. These are the working front of the library: some are print-only or pre-digital, some are blocked behind registries we currently cannot reach, and a few were found in audit to be defective and have been withdrawn or repaired. Our corrections log records what the audit found.
asserted-unverified — flagged as verified in an earlier pass without a recorded basis. — entries. Cleared pending re-verification.
The counts above are generated live from the published dataset on every load — never hand-entered. Historical note: a 90-entry integrity audit conducted 2026-08-09 compared every stored registry record field-by-field and found all ninety matched on title, year, journal, and identifier. That is a fixed historical result, distinct from the live counts above.
The full corpus, including verification status, in machine-readable form. Metadata is released CC BY 4.0; the source documents themselves remain with their publishers. If you find an error, we want to hear about it and we will log it publicly with attribution.
| Source | Kind | Year | Identifier | Status |
|---|
Most reference sites ask you to accept their bibliography on faith. The list is long, the claims sound plausible, and there is no practical way for a reader to test whether the sources say what the site says they say. That gap is where bad reference works live, and this field — long under-served and largely unregulated — has been especially poorly served by it.
Publishing the corpus with its verification status attached inverts that. It makes our gaps visible alongside our strengths, it lets a skeptical reader check any individual claim in about thirty seconds, and it lets a determined critic audit the entire collection in an afternoon. We think a reference work that can survive that is worth more than one that has never been asked to.