Sources & method

79 of the landmarks here are listed World Heritage properties and say so on their page. The other 1,196 properties on UNESCO's list are searchable but not drawn, and the reason is worth stating plainly: the only date the World Heritage List carries is the year a property was inscribed — added to the list — which is a fact about UNESCO, not about the thing.

Placing Stonehenge at 1986 because that is when it was listed would be false about every property on the list, so no listing becomes a bar. Where a site already exists here as a landmark it keeps the atlas's construction date and gains a badge; where it does not, the listing is offered in search and links to UNESCO's own entry.

World Heritage data from the UNESCO DataHub, © UNESCO, available under CC BY-SA 4.0. Selected fields; no UNESCO emblem or photographs are reproduced.

An empire on this timeline is drawn from century snapshots. The atlas renders the world at 49 moments between 9000 BCE and the present, not continuously. When a polity appears in the 1300 snapshot and again in the 1700 one, the timeline draws a bar from 1300 to 1700 and calls it attested.

That is not a lifespan. The state may have been founded two centuries before the first snapshot that caught it, and may have outlasted the last by as long again. Both edges are floors. Every polity page on this site says so on the page itself rather than in a note at the bottom, and the wording — "attested", never "founded" — is deliberate.

518 of the 2,984 are the exception. Where Wikidata records a founding date and it agrees with the centuries the atlas independently drew, the bar runs to the real dates instead and the page says "Founded", not "Attested". The Byzantine Empire is drawn in eight snapshots between 800 and 1400 and its bar now runs 395–1453. A claim is only checked against the atlas, never replaced by Wikidata — a date that disagrees with the snapshots is rejected, which is why "Egypt" is not dated to 1922 and "Rome" is the Republic rather than the city.

The consequence a reader should carry: bars of the same length are not comparable across kinds, and among polities they are not all the same kind of claim. A 400-year war bar is four hundred documented years. A 400-year attested polity bar is "present in two snapshots four centuries apart". Only a bar with real dates, and a duration printed beside it, is claiming four hundred years of existence.

The border snapshots come from Historical Basemaps, the open dataset the HistorIQly atlas is built on. Its own caveats apply here in full: historical borders are interpretations, frontiers were often zones rather than lines, and the further back a snapshot goes the more of it is inference.

The atlas holds no images, and 4,422 records is far too many to choose art for by hand. So nothing here was chosen: each picture is the lead image of the record's Wikipedia article, which is a choice already made and defended by the people who wrote it. 2,532 of the 4,422 records have one — 2,313 distinct files. The rest show their kind's mark, and always will: a record with no picture is an ordinary record here, not a broken one.

Only freely licensed files are used, and only files held on Wikimedia Commons, because Commons is where the licence and the author can be read. A picture whose licence cannot be read is not shown. Every page names the author and the licence of every picture on it, and links to the file's own Commons page.

A wrong picture is worse than none, so the article behind each one has to be the record under a different spelling — "Viceroyalty of New Spain" for the atlas's "Vice-Royalty of New Spain" — and never merely a near match. Wikipedia's own search offers "Maya stelae" for "Maya states"; that is refused, and the record keeps its mark.

A record that is wrong here is wrong in the atlas too, and fixing it there fixes it everywhere. Send it to historiqly@cyril.page with the record's name and what it should say.