Record Atlas
3.50M volumes

3,502,257 volumes in the Atlas

Local
What this site collects, and what it does not

Privacy

A static site with no account, no login and no database.

What Record Atlas itself collects

Nothing. There is no account, no sign-in, no form and no database. The site is static files served by GitHub Pages. What you type into the find box, which layers you switch on and which year you set stay in your browser's address bar and are never sent anywhere.

If you choose to keep a copy online

On My research you can keep an encrypted copy of your research online, with no account. The six words it gives you are made in your browser and are never sent to us. Your research is encrypted in your browser, and only that scrambled copy is sent to our storage, which cannot open it. We hold no names, no email address and no way to recover a copy: if you lose the six words, the copy is gone. Your browser keeps the six words (and the time of the last save) under one key, ra-sync-v1, so "Sync now" works without retyping; "Stop syncing on this device" removes it, and "Delete everything I have stored" removes the copy from our storage. The Tree Canvas keeps where your cards sit, in your browser (ra-canvas-v1), and with the six words on, in the same encrypted copy: the link, file name or archive each tree came from, and a short code for each card, never the tree file or a name, date or place from it (what is kept). This is off unless you switch it on, and the rest of the site works exactly the same without it.

If you ask Claude to read a line

On Reading the record you can paste a line of a register, or photograph one, and ask Claude to draft a reading. This is the one place where something you type leaves your browser. It happens only when you press a button. The text (or the picture, shrunk and stripped of its hidden data in your browser) goes to our small service at ai.recordatlas.org, which sends it to Claude, an AI made by Anthropic, through Anthropic's programming interface, and returns the draft. Our service keeps counts only: how many tries an address, a six-word identity and the whole site have used today, so that nobody can run up a bill. It keeps no text, no picture, no answer and no name, and its request logs are off.

Anthropic's published policy for this kind of use says it deletes what is sent within 30 days, keeps content that its safety systems flag for up to two years, and does not train its models on it without permission (its policy; its API retention page). Policies change, so read those if it matters to you. Please do not paste or photograph details of people who may be alive.

To stop automated abuse the service asks Cloudflare's bot check (Turnstile) to confirm you are a person. That script loads only when you first click in the box, and Cloudflare sees your address and browser as it does for any site that uses it. If you have set up sync, your six words give the service a separate identity for your allowance; the service cannot open your saved research with it. Everything else on that page, the word finder and the date decoder, runs in your browser and sends nothing.

What your browser fetches from elsewhere

  • Map tiles from Esri (the default map, with its place names, the satellite view, and the dark grey tiles the Night dots map shows once you zoom in) or OpenTopoMap (the terrain view), depending on which basemap is showing. The Night dots map's land outlines (Natural Earth) come from this site, and only when you choose it. Those services see your IP address and which tiles you asked for, as any map on the web does.
  • Leaflet, the mapping library, from the Cloudflare CDN.
  • Fonts from Google Fonts.
  • GitHub Pages serves the site and keeps standard server logs.

Following a link to a provider takes you to that provider's site, under their terms and not these.

Advertising

This site carries a single advertisement in its footer, served by Google AdSense. Google and its partners may use cookies to serve and measure ads. In the EEA, the UK and Switzerland, Google's own consent message appears before personalised ads are served, and you can change that choice at any time.

You can review Google's practices at policies.google.com/technologies/partner-sites.

Gifts

The Support page links to a payment page that Stripe hosts. The payment happens there, under Stripe's terms and privacy policy: this site never sees your card, your name or your email, and stores nothing about a gift. Nothing is loaded from Stripe until you follow the link.

Counting visits

This site counts page views with GoatCounter, the same counter every archive in this family of sites uses. It sets no cookies, gives you no identifier, and keeps no personal data — it records that a page was loaded, roughly where in the world from, and which page sent you. Nothing is shared with any advertiser, and nothing is joined to the advertising above.

There is no consent banner for it because there is nothing to consent to: it collects no personal data in the first place. If you would rather not be counted at all, any tracker blocker will stop it, and the site works identically without it.

Living people

This map is about records and the institutions that hold them, not about people. It carries no names of individuals, living or dead. The one-family archive it grew out of keeps a strict rule about living people; this site sidesteps the question by never holding a person in the first place.

Getting in touch

Corrections, including anything you want removed, go through the public issue tracker.