Sources

How entries are made

The borders are scholars' work. Every shape on the map comes from the datasets listed below — Cliopatria and, for the German states, Historical Basemaps. No AI draws or alters them.

The text is written with AI. The glossary articles and the short territory descriptions in the map were researched and drafted by an AI research assistant, working from the sources named on each entry: for most, the subject's English Wikipedia article; for some, reference works such as Britannica, Encyclopaedia Iranica and World History Encyclopedia.

What the build checks. Each entry is matched to the territory it describes on the map, and the match is verified before a page is published. Dates the research does not state are never filled in — the page says what the map draws instead. Passages the research itself marked as unverified are held back rather than published.

What it does not do. The entries are not written or reviewed by professional historians, and not every sentence has been checked by a person. AI can get dates, names and details wrong. Treat an entry as a well-sourced starting point, follow its sources for anything that matters, and if you find a mistake, tell us at hello@theworldthroughtime.com — corrections go into the research and are republished.

The borders on the map

• Cliopatria v0.12 CC-BY-4.0

  Seshat Global History Databank

  Roughly 14,000 dated polygons covering the whole world from AD 1. Every territory on the map comes from here unless the panel says otherwise.

• Historical Basemaps GPL-3.0

  André Ourednik

  Fixed-year snapshots, used only for the optional German states overlay between 1715 and 1814, where Cliopatria draws a single Holy Roman Empire blob.

• Natural Earth 1:110m land Public domain

  Natural Earth

  The coastline silhouette drawn behind the territories in equal-area mode.

• Natural Earth 1:50m shaded relief Public domain

  Natural Earth

  The optional relief under the Mercator projection, reprojected and cut into tiles served from this site rather than a third-party tile service.

Code and type

• Leaflet 1.9.4 BSD-2-Clause

  Volodymyr Agafonkin and contributors

  Drives the Mercator backend only, served from this site rather than a CDN. The equal-area backend is plain SVG with no dependencies.

• Equal Earth projection Published method

  Bojan Šavrič, Tom Patterson and Bernhard Jenny (2018)

  The default projection. Implemented directly from the paper's coefficients, so areas on screen are true to areas on the ground.

• Geist and Geist Mono SIL Open Font License 1.1

  Vercel

  Self-hosted, so no request ever leaves the page for a font.

Method

• Douglas–Peucker simplification Published method

  David Douglas and Thomas Peucker (1973)

  Reduces the source geometry to a 0.25° tolerance (~28 km). The source's own point spacing is ~41 km, so nothing visible is lost.

• Sutherland–Hodgman clipping Published method

  Ivan Sutherland and Gary Hodgman (1974)

  Clips polygons to the map's bounding rectangle at build time.

Reference works cited in the glossary

Most entries are anchored on their subject's English Wikipedia article; some also draw on independent reference works — Britannica, Encyclopaedia Iranica, World History Encyclopedia, UNESCO, university repositories and specialist works. Some regional research files cite their sources once for the whole region rather than per entry. Everything consulted is listed below, grouped by publisher.

What the data cannot do

This map draws real borders from published datasets rather than hand-drawn shapes. It also has real limits, and they are worth knowing before you rely on it.

Borders are sampled, not continuous

The sources record a territory's shape at particular dates and hold it constant until the next recorded change. Moving the slider to 1631 shows the last configuration the data knows about at that point — not a reconstruction of that specific year. Every panel names the span its polygon actually covers.

Before 1648, borders were not lines

Medieval and early-modern territory was a bundle of overlapping rights — jurisdictions, tolls, advocacies, church lands — held by different people over the same ground. Drawing that as a crisp polygon is already a simplification. Treat pre-Westphalian outlines as approximations of where authority was concentrated.

Not every territory is shown

In the 18th century the Holy Roman Empire contained roughly 1,800 entities, of which about 294 were significant states — 243 with imperial immediacy and 51 free imperial cities. No open dataset maps all of them. Where the primary source lumps the remainder together it appears here as a single region honestly labelled Holy Roman Empire — smaller territories, not as one country.

Most of the map is blank for most of history — on purpose

This is the single most important thing to understand about this map. The source records territorial states. For most of human history most of the world's land had no such thing: much of Africa, the Americas, Australia, Siberia and the Pacific were lived in by peoples who did not draw borders. Those areas are drawn as plain land, and that blankness is a finding, not missing data. In AD 1 the whole world here holds 45 mapped polities; by 2024 it holds 190.

Filling the blanks in would be an invention, so the map leaves them empty.

Areas are approximate, and sometimes badly so

The figure in each panel is the area of the polygon as drawn. The source was digitised from map images at roughly 40 km resolution, and coarse tracing systematically inflates a compact country with a convoluted border. Measured against accepted figures: the Kingdom of France in 1789 and Germany in 2024 both land within 1%, and the Kingdom of Italy in 1871 within 4% — but the German Empire in 1871 comes out 22% too large. Use the numbers to compare and to sort, not as authoritative statistics.

A known mislabel: the Golden Horde

The source keeps applying the name Golden Horde to the western steppe until the 1690s, though the Horde itself fragmented into the Crimean, Kazan, Astrakhan and Siberian khanates and the Great Horde was destroyed in 1502. Those later polygons are left exactly as the source draws them rather than renamed, because what they are meant to represent is not clear enough to relabel honestly — but they should not be read as the Golden Horde surviving into the 17th century.

Corrections applied to the source

Two changes are made to the primary dataset, and no others. It records the post-war occupation zones under the occupying power's own name, and lets the American one run to 1966 — which would draw West Germany labelled United States of America on top of the Federal Republic for sixteen years. Those records are relabelled as occupation zones for 1945–1949, and the 1950–1966 remainder is dropped as a duplicate of the Federal Republic. The geometry itself is untouched; nothing is invented. The Soviet zone is not broken out separately, because the source draws it as part of one large USSR polygon.

Two datasets, and where each is used

• Cliopatria v0.12 — Seshat Global History Databank, CC-BY-4.0. The whole world, every year. Around 14,000 dated polygons worldwide; 1,420 distinct polities appear on this map across 404 epochs between AD 1 and 2026, of which 716 have hand-written descriptions — about 96% of what is on screen at any given year, measured by area. The rest carry the source record and a Wikipedia link, and are visibly flagged as unverified. Repository · Paper
• Historical Basemaps — A. Ourednik, GPL-3.0. A separate, optional overlay for 1715–1814 over the German lands only, where it names 16–34 territories against Cliopatria's 11–13 plus one undifferentiated blob. Repository

Both projects warn that historical boundaries are disputed and that their maps should be checked against other sources before academic use. That warning is passed on here in full.

Base map

The Mercator view's terrain is Natural Earth's 1:50m shaded relief (public domain), cut into tiles and served from this site rather than a third-party tile service. Labels and modern national borders are deliberately absent: relief and coastlines are roughly constant and help explain why borders ran where they did, whereas modern place names under a 9th-century map would mislead. Turn the base map off entirely with the checkbox, or fade it with the opacity slider.

Colour

Colour separates neighbours; it does not identify anyone. Shades are assigned so that no two adjacent territories share one, and kept stable over time so a territory does not change colour as you scrub. Names come from the labels, the hover tooltip and the legend — never from colour alone.

Licence

The German states overlay (1715–1814) is derived from André Ourednik's Historical Basemaps and distributed under the GPL-3.0. Its complete corresponding source — the layer, the upstream files, the scripts that made it and a list of the changes — is a download: source/german-overlay.zip.

The world layer is derived from Cliopatria, © Seshat Global History Databank, and stays available under CC-BY-4.0. Natural Earth's coastlines and relief are public domain; Leaflet is BSD-2-Clause and Geist is under the SIL Open Font License.

The glossary text — every article and every territory description in the map — is licensed CC BY-SA 4.0, the same licence as Wikipedia. Reuse it freely: credit "The World Through Time" with a link to the page, say if you changed it, and share your version under the same licence. Details in the terms of use.

The site's code and design are © The World Through Time, all rights reserved.