About Record Atlas
A map of where genealogical records are, who holds them, and what it costs to look — searchable by every name a place has ever had.
The problem
Genealogy is not short of websites. Cyndi's List has three hundred thousand links. The FamilySearch Wiki has a page for every country on earth. Almost nobody uses either, because a list answers what exists and the question a researcher actually has is my great-grandfather was born in Gallignana in 1863 — where is that record, and can I see it tonight?
Three things stand between those two questions, and all three are geographic.
The name changed. Gallignana is Gračišće. Fiume is Rijeka. Pressburg is Bratislava. You cannot search for a place whose name you do not know, so the find box here matches every name a place has ever had — and it does this for the whole world, not only where the shelf is deep. 45,985 places carrying 115,922 other names they have been written under: type Breslau and it says Wrocław, Czernowitz and it says Chernivtsi, Memel and it says Klaipėda. Of the three problems, this is the one that does not need a shelf to answer, so it is answered first and everywhere.
The jurisdiction changed. That 1863 baptism is in a Croatian archive. A 1930 one from the same village is filed under Italy, because between the wars the village was Italian. This is not a curiosity — it is the single commonest reason a record that exists cannot be found, and this map has a table for it.
"Online" is eight different things. Free images; images behind a free login; an index with no images at all; part free and part paid; images behind a paywall; a catalogue card for paper nobody has photographed; paper you can only read in the building; and — for nine places in ten — nobody has looked yet. Most directories flatten all of that into a link. Here it is the colour of the dot, and the eighth colour is the commonest one on the map.
And a fourth thing, which took this atlas a while to admit: the record is not always a register. Australia has no parish-register series. Civil registration there is state by state and closed for a lifetime. What exists instead is 2,014 digitised newspapers, free and full-text, in which births, marriages, deaths and obituaries were printed in volume — so for most Australian ancestors before 1950 the newspaper is the record. A map that only counted registers would have called that country empty and been wrong about it.
Whose flag flew over it
The jurisdiction problem has a companion, and as of today it lives here rather than beside here. 245 pages of flag history — who governed which ground, when, and the banner they flew — were built as a separate project and have been merged in whole. They are unchanged: this was a move, not a redesign.
The two halves answer the same question from opposite ends. This atlas knows, for 230 researched eras across 41 regions, which archive a regime's paper actually went to — the part nobody else publishes. The flag pages know who was flying what, for 210 countries rather than 41 regions. Each era on a flag page now shows what this atlas holds for that ground in those years, fetched live from a file this site publishes for anyone to use.
They are side by side, not yet one thing. The flag timelines are written into their pages as markup, so a search here cannot see them. Freeing that data is the next piece of work, and until it is done this page will not pretend otherwise.
Where it came from
Out of a one-family archive. The Defranceschi Archive's map answers three questions about one family's ground: where the record books are, where a named person can be put, and where the family is buried. Two of those are family facts and cannot generalise. The first one can, and this is it, taken out and pointed at the world. There are nine such archives, and what they have in common is the reason this atlas exists.
The Croatian depth here — 1209 places and 6,528 volumes, of 40,266 dots in all — is that archive's work, imported once as a snapshot. Everything family-specific was dropped on the way in, including the thing its map is coloured by: how far one researcher had got. That is a true and useful fact if you are that researcher. To a stranger it means nothing, so the colour here answers a stranger's question instead.
What it does not know
It is early and it says so, and here is the number that says it: 94.9% of the ground on this map has never been surveyed. There are 40,266 dots standing in 101 countries, and only 1,907 of them, across 46 countries, are places somebody has actually walked volume by volume. The rest are administrative districts — the units records are filed in, correctly placed and completely unread — and they are drawn hollow so that the difference cannot be mistaken. A place outside a mapped region says so in its own panel rather than guessing.
The regions were chosen by where research is actually being done, not by what would look impressive on a map. Each one was written because somebody was stuck on that ground and needed to know whose archive to write to — which is why the list runs Central Istria, Southern Istria, The Kvarner — Senj, Vinodol, Rijeka, Carnia, Friuli, Dalmatia, The Imotski frontier, Campania — Caserta and Naples, Eastern Sicily — Catania, Calabria, Posen — Prussian Poland, The eastern Cape, England and Wales, Queensland, Santa Fe — Rosario, Galicia, Bukovina, Bohemia, Moravia and Silesia, Hungary — the Kingdom's core, Transylvania and the Banat, East Prussia, Pomerania and Danzig, Ireland — the whole island, Scotland, Alsace and Lorraine, Silesia, Livonia, Estonia and Courland, Congress Poland — the Russian partition, Carniola and the Slovene lands, Bosnia and Herzegovina, Vojvodina and the Banat frontier, Quebec and French North America, Acadia — the Maritimes, The Military Frontier — Vojna krajina, The Netherlands, Denmark, Norway, Sweden, Finland, Portugal, Belgium and the Southern Netherlands rather than the tidy sweep of Europe you might expect.
A grey dot means nobody has surveyed it. It does not mean nothing is there. Keeping those two apart is worth more than coverage, because a map that quietly implies completeness sends people home when they should keep looking.
The same honesty applies to the institutions. Of the sources in the directory, a little over three hundred are listed because a national register or an open dataset named them and nobody has opened them since. Two hundred of those have now at least had their door rattled — alive or dead, a page or an application, open or refusing robots — which turns a list of URLs into a queue somebody can work. It does not turn any of them into a surveyed source, and the directory does not say it does.
Declining after looking is a result; declining without looking is a gap. Fifteen sources have been assessed and turned down with the reason written out, and this project does not defeat a bot gate or send a user-agent it is not, however easy that would be. An archive that refuses a robot is stating terms, and an atlas that asks archives to trust it cannot be the thing that ignores them.
How it is built
A static site — no server, no database, no account. The data is plain JSON in a public repository, and a build gate refuses to publish it if a source has no date on it, if a provider does not resolve, or if a place's coordinates fall in a different country from the one it claims. That last rule caught this project filing the Italian comuni of Carnia as Croatian on its first import.
Country outlines are Natural Earth, public domain. The gazetteer is GeoNames, CC BY 4.0, downloaded rather than scraped — the providers this project would otherwise have ingested wholesale refuse robots, and the honest answer to that is to build on the data that is actually offered rather than to take what is not. Maps are OpenStreetMap. The historical regions are drawn by hand, deliberately roughly, because they exist to say which filing system a place fell under rather than to adjudicate a border.