An open, measured atlas of federal regulation

What this is

eCFR Atlas measures the Code of Federal Regulations and publishes the results: how many words each agency's regulations run to, where jurisdiction overlaps, how often each part changes, and the full text of every part with a citation you can copy. It is optimised for lookup, citation and export rather than for browsing.

It exists because the obvious version of this site is easy to build wrongly. Its predecessor published word counts it had inferred rather than measured, and stored them in the same place as the real ones — so “we could not measure this” and “we measured 104,642” became indistinguishable a layer downstream. Everything about the current design, down to the shape of the database columns, is arranged so that cannot recur.

Where the data comes from

All source material comes from the Electronic Code of Federal Regulations public API at ecfr.gov. The eCFR is a joint project of the Office of the Federal Register (National Archives and Records Administration) and the Government Publishing Office. Related material is published through govinfo.gov.

Federal government works are not subject to copyright in the United States. This project adds measurements, structure and presentation; the underlying regulatory text belongs to the public.

This build reflects the eCFR as of September 30, 2026.

Three disclaimers

How to cite

For a figure: eCFR Atlas, “{page title}”, measurement current as of 2026-09-30, ecfr.fixit.works.

For regulation text: cite the CFR itself, and verify against the official sources above. The copy-citation control on every part page produces a citation with the eCFR link and the source date already in it.

Privacy

This site is static HTML served from a CDN. It sets no cookies, runs no analytics, embeds no third-party scripts, and loads no external fonts or images. The only thing it stores in your browser is your light/dark theme preference, and only if you change it from the default.

Contributing

The pipeline, the API and this site are open source. Corrections to the measurement rules are especially welcome: if you think the tokenisation or the exclusion list is wrong, the argument is worth having in public. See the API page for programmatic access and methodology for the current rules.