How this archiveis made.

A fragment is one object, place, text or idea that says something about what humans made, found or lost. Each of the six halls gathers fragments from every continent, and a coverage table, generated with the archive, shows where the collection is still thin: by hall, by domain, by continent and by language.

We do not look for the most famous piece. We look for the one that carries a human gesture: a notch cut in bone, a hand pressed on rock, a line in a colophon. No domain named in the founding idea is left out, from the first tools to artificial intelligence, and wars and losses are told with the same care as achievements.

Every fragment cites at least two sources, and at least one of them is scholarly or institutional: a peer-reviewed article, a museum record, an excavation report, an official heritage listing. A fragment that does not meet this rule cannot be published: the archive is built by a validator that refuses it.

Dates are given with their uncertainty (about, between, before) and coordinates are marked as approximate when they are. When sources disagree, the fragment says so and gives the range of published estimates rather than choosing one in silence.

Each fragment exists in English and in French, with the same content; a fragment missing one language is not published. A text written in one language is translated, then read again by a person against the sources before publication.

Names of places and peoples keep the forms used by the communities concerned, with the usual English or French form alongside.

What is established and what is interpreted are never mixed. Established facts sit behind a solid line; readings and hypotheses sit behind a dashed one, labelled as such. The archivist's note, written in the past tense of a civilization that has gone, is a way of telling, not a claim of new knowledge.

The archive does not yet have a historian or an advisory board. Until it does, every fragment carries the words “Scholarly review in progress”, and its rigour rests on the sources the validator requires and on a careful human reading in each language.

When a specialist reviews a fragment, their name and the date of the review will be shown on it. If you are a researcher or a curator and would like to review fragments in your field, we would be glad to hear from you.

No image generated by artificial intelligence is ever presented as a historical document. Photographs and scans come from museums, archives and open collections, and each one carries its licence and a link to its original record.

Software may help draft or translate a text. When it does, a person checks every sentence against the sources before publication, and remains responsible for it.

One image is not a document: the picture of a statue in a shaft of light on the threshold page is a decorative image, chosen and supplied by the founder of the project. It is never used as a fragment and is listed as such in the rights and credits.