Sources and methods

• The Paleobiology DatabaseCC BY 4.0

  Every occurrence, date, classification and formation name. If something here says an animal lived in a place at a time, this is why.

• GPlates Web ServiceOpen, run by EarthByte

  The reconstructed coastlines, computed once per period and stored, so browsing costs the service nothing.

• Wikipedia and Wikimedia CommonsPer image, shown on the card

  The picture on a card, its licence and its author, and the description under the name. Each picture is licensed on its own terms, printed beside it.

• WikidataCC0

  Matches a taxon name to the right Wikipedia article, and supplies the article count behind the eye-catching ordering.

• OpenAlexCC0

  How much has been published about a taxon, used only for ordering.

• iDigBioPer record

  Photographs of actual catalogued specimens in museum collections.

• MorphoSource and SketchfabPer model, shown beside it

  The 3D scans, where a museum has published one of a specimen.

None of these are ours, all are free to use, and most run on donations or grants. The app caches hard and paces its requests so browsing here costs them as little as possible.

A fossil occurrence is a record. Almost everything built on one is a reading of it. These are the readings this site makes, and where each stops being true.

• Found in North America

  Rests on The country a fossil was dug up in, grouped by the continent it sits on today.

  Stops at A modern address, not an ancient one. A Triassic fossil from Morocco is filed under Africa, though Morocco was then joined to eastern North America. The maps show where the land was; the label does not.

• The coastlines on the maps

  Rests on The GOLONKA rotation model, the same one behind the Paleobiology Database's paleocoordinates, so shorelines and fossil dots describe one Earth.

  Stops at One model among several that disagree, drawn at a single date mid-period while continents move throughout. Read a shoreline as roughly where the sea was.

• The continental configurations on Tectonics

  Rests on Written by hand from published reconstructions. They are the app's own reading, not a database field.

  Stops at They decide exactly one thing: which region a search continues into when the chosen one runs out of fossils. The maps do not depend on them.

• Most eye-catching first

  Rests on An opinion, and the only one in the app: a coarse weighting over taxonomic class, nudged by how much has been written about the name.

  Stops at It says a ceratopsian is more worth looking at than a foraminiferan, which is a fact about people. It only reorders: nothing is hidden, no count changes, and every other ordering is a measurement.

• Likely alongside

  Rests on Taxa dug out of the same Paleobiology Database collection, which is one site and one horizon.

  Stops at Not proof of coexistence: a collection can span a bed metres thick and thousands of years. It is evidence of the same place and roughly the same time.

• The lighter cap on a diversity bar

  Rests on Taxa the database reports as implied rather than sampled: known before and after a period, so alive during it, but not found in it.

  Stops at Inferred from a range, not dug up, and drawn separately so the difference stays visible.

• A formation's rank

  Rests on Distinct taxa a search found in it.

  Stops at It measures the rock and the fieldwork together. A formation ranks highly by being rich, by being reachable, and by having been worked for a century.

• A last appearance

  Rests on The youngest occurrence on file.

  Stops at When something stops being found, not when it died out. The gap is why an extinction is normally given as a range.

• The dated events on Tectonics

  Rests on Compiled by hand from the published literature, and the app's own selection of what is worth listing.

  Stops at Not a database field, and not exhaustive. Every date is a range because every one of them is: a rift takes millions of years, and several are argued over. India meeting Asia is dated anywhere from 60 to 35 million years ago.

• The dashed line through a scatter of points

  Rests on A power law fitted by least squares to the points on that chart, on logarithmic axes.

  Stops at A description of that cloud and nothing more. It corrects nothing, and claims nothing about how many species existed. Its job is to give you a line to judge a point against.

• The artwork on a card

  Rests on Whatever illustrates the taxon's Wikipedia article, credited to its author on the card.

  Stops at A life reconstruction is an artist's interpretation of a skeleton, and colour and soft tissue are almost never preserved. A photograph of a specimen is evidence; a painting of a living animal is an argument.

The one thing most worth carrying around this site: an empty result does not mean nothing lived there. It means nothing has been found, recorded and entered into a database, which are three separate hurdles. Fossilisation is rare and collecting is uneven, and both favour ground that is dry, exposed, reachable and near a university. Whole periods and continents are thin here for reasons that have nothing to do with what was alive.

It is also why the chart on Taxonomy draws the number of fossils underneath the number of kinds. The two curves usually have the same shape, and that is the finding.

No account, no cookie, no analytics, no tracking of any kind. Two things are stored, both in your own browser and neither ever sent anywhere: your settings, and anything you save. Clearing site data clears both, and neither follows you to another device. That is the trade for not having accounts, and the Saved panel offers a download so you can keep a copy.

Searches are answered by this server, which caches the result so the next person asking the same question does not cost the databases a second request. What is cached is the answer, keyed by the question, with nothing identifying you.

Worth naming rather than glossing: the rate limiter holds your network address in memory for a few minutes to count requests. It is not written to disk, not logged, and not tied to anything you looked at.

The endpoints the pages call are open and return JSON. A script gets the same answers a browser does, and the same rate limit.

GET /api/explore?region=North America&interval=Late Cretaceous&view=index
GET /api/diversity?group=Ammonoidea&rank=genera
GET /api/taxon-profile?name=Triceratops

A refused request returns HTTP 429 with a Retry-After header; honour it rather than retrying at once. Every response carries X-RateLimit-Remaining, so a script can pace itself before being told to.

If you need the whole database rather than slices, take it from the Paleobiology Database directly: it publishes bulk downloads.

EpochExplorer.Earth@gmail.com

Worth writing about: something wrong, something that reads as more certain than the record supports, or a picture that should not be here. This republishes other people’s photographs under their own licences; if one is yours and the credit is wrong or you want it removed, say so and it will be. One person reads this, so a reply may take a few days.

Any result set here downloads as CSV, and anything with coordinates also downloads as GeoJSON for QGIS and similar tools. Look for Export beside the results, on the formations ranking, and under the diversity table.

Every file carries its own credit and licence terms, as a comment block in a CSV and as fields in a GeoJSON. Attribution is a condition of that licence rather than a courtesy, so leave those lines in place, and cite the database rather than this site in anything you publish.