How we source

Popular history of medicine is full of myths and contested "firsts". Tabula treats accuracy as a gate, not a nicety: the site will not build if any rule below is broken.

The rules

1. Every factual claim is a numbered note with its own sources.
2. Two sources per claim, at least one of them scholarly: peer-reviewed history of medicine, a university press, or a major institution such as a national library, museum or learned society. A primary source is cited where one exists.
3. General encyclopaedias don't count (Wikipedia, Britannica). They can lead us to sources, but they are never cited as one. Specialist peer-reviewed reference works can count.
4. We search for the myth on purpose. For every "first", date and attribution we look for debunkings, priority disputes and alternative dates, and record what we found.
5. Dates carry their precision (exact, circa, range, century). Day-and-month dates need a citation and a calendar (Julian or Gregorian).
6. Status labels are public. Established means no significant scholarly dispute. Contested means credible scholars disagree, and the note says who and about what. Revised means a popular claim that is wrong, kept so the myth gets corrected rather than spread.
7. A verification log records, for every story, the sources consulted, the myth checks run and a verdict. If a claim is edited, the log goes stale and the build fails until it is re-verified.

Who verifies

Verification is done by Claude, an AI model, working to the method above, with the sources linked on every note. AI-assisted verification reduces errors but does not guarantee there are none. Some key scholarship is in print-only books, and claims we could check only through summaries are labelled read via summaries. Sources whose byline or title we have not fully confirmed are labelled details pending, and their links are authoritative. The data model also records human specialist reviews, if one is added.

Found a mistake?

Every note has a "Suggest a correction" link, with an email option if you don't have a GitHub account. Accepted corrections are published in the corrections log.