How We Know
Every claim on this map — a founding date, a governor's name, whether a building was still standing on one particular day — traces back to a real source. Here's the discipline behind that.
The one rule that shapes everything else
Roman Maps is frozen at a single moment: 11 August 117 CE, the day the emperor Trajan died at Selinus in Cilicia. It is not a timeline and it does not blur centuries together. Every record on the map carries a plain extant_117ce flag — standing on that exact day, or not.
That single rule forces real judgment calls, and we make them the same way every time. A building finished before 117 is true. A building finished after — the Pantheon as it looks today (126 CE), the Colosseum's later renovations, Hadrian's Wall (begun 122) — is false, even though it is often the more famous version of the building. A building destroyed before 117 — Pompeii, buried in 79 CE — is also false, but we still show it, because it was still a real place in Roman memory. A building under construction right at the snapshot gets a note explaining exactly how far along the work was, rather than a flat yes or no. When the evidence doesn't clearly settle it, we default to false and say why in the description — the safer wrong answer is the one that undersells a monument, not the one that shows something too early.
Where the map itself comes from
The base geography — coastlines, rivers, provinces, the road network — is assembled from open scholarly and cartographic datasets: Itiner-e and Pelagios for Roman roads, the DARE (Digital Atlas of the Roman Empire) project for province boundaries and the ancient gazetteer, Natural Earth for coastlines and physical geography, and OpenStreetMap contributors for the building-by-building street plans of the 40 archaeological sites you can walk through at full zoom — Ostia, Pompeii, Ephesus, Palmyra, and the rest. None of that is invented; all of it is checked against the ancient footprint before it ships, and every layer's source is named in the map's own attribution strip.
Two kinds of source, kept separate
Click any curated place and you'll usually find two different source blocks, and they answer two different questions. Sources is where we found the facts — modern scholarship, excavation reports, museum records, the reference works that establish a date or an identification. In ancient writing is different: a passage from a Roman-era author, inscription, or papyrus that mentions the place directly — Suetonius on the Basilica Julia, a milestone naming a road station, a graffito naming a house's owner. Not every place has one; ancient authors wrote about famous temples and forums far more than they wrote about ordinary frontier forts, and we don't invent a citation to fill the gap.
As of today, 469 of 469 curated places (100%) carry a modern source, and 172 of them (37%) quote an ancient author or inscription directly, and 158 of 230 of our highest-confidence records (69%) now carry that ancient-source citation — a number we track as a standing target, not a finished total, because it only ever goes up as research continues.
Confidence, honestly labeled
Every curated record carries a confidence rating — high, medium, or low — based on how firmly its location and identification are established in the record. High means excavated, securely identified, and precisely located. Medium means the identification is generally accepted but the exact spot or date has some give. Low means "somewhere in this valley" or "named in a text but never excavated" — we still show it, because a documented uncertain location is more honest than pretending the place doesn't exist, but we say plainly that it's uncertain. This label never appears as jargon on the place card itself — no visitor should have to learn what "confidence: medium" means just to read a description — but it drives which places get a precise pin versus an approximate one, and it's always available in the place's own data record for anyone who wants to check our work.
Images
253 of 469 curated places (54%) carry a real image — a scholarly reconstruction drawing, a 19th-century engraving, or a modern photograph of the actual ruin, sourced from Wikimedia Commons and credited by name underneath. We do not use AI-generated images, stock photography standing in for a specific site, or modern photographs that show visibly modern infrastructure where an ancient one should be — a wrong picture is worse than no picture, so a place with no verified image gets a plain tinted rail instead of a guess.
What's still incomplete
This map is built in public, in ongoing research shifts, and we'd rather say what's missing than paper over it. Some provinces have no securely dated 117 CE governor in the surviving record at all — we leave those blank rather than guess a name. Inscriptions and literary mentions run out for entire categories of place (tombs, villas, shipwrecks) where ancient authors simply had no reason to write about them. 31 places still have a description under 60 words, and that number is a live work queue, not a hidden one — npm run metrics prints it fresh every time it runs, the same script that generated the counts on this page.
When we get something wrong, we don't quietly delete it — we correct it in a follow-up commit and note the correction in the project's own running shift log, the same standard a Wikipedia talk page holds itself to. If you find an error, the map's data files are what actually ship — there is no separate "true" answer hidden somewhere else.
Read more
Roman Maps shows the empire frozen at 11 August 117 CE, the day Trajan died — its moment of greatest territorial extent. Explore the full map →