How this atlas works
Anima Mundi tries to put the world’s religions, myths and unexplained experiences in one place, so that historians, theologians, scientists, investigators and curious readers can look at the same material together. That only works if you can see how everything in it was made. This page sets that out, including the parts we are least sure of.
Where the information comes from
Every case, tradition, god, person and place links to its sources: scholarly books and articles, encyclopedias, museum records, official reports, and primary documents such as declassified government files. Where a primary document could not be opened online, it is cited rather than quoted, and listed for retrieval. Documents are linked to where their owners publish them, not copied onto this site.
The atlas holds about 4,700 case files, 101 traditions and 146 gods, with people, routes, myth motifs, rites and primary sources linked between them. Depth is uneven: some entries rest on a whole literature, others on a single account. Each page says which.
How it was built, including the use of AI
Anima Mundi was designed and edited by its owner, and much of its writing, data entry and code was produced with the help of an AI assistant (Anthropic’s Claude), working from the sources cited on each page. We say this plainly because it matters for how you read the atlas: an AI can misread a source, compress a nuance, or state something with more confidence than the source allows.
What we do about that:
- Every claim is meant to trace to a cited source, and the source is one click away.
- Sources that could not be opened are not cited from memory; they are listed for retrieval.
- Uncertain and contested points are labelled as such, never smoothed over.
- Corrections are welcome and acted on (see below), and changes to assessments are recorded.
If you find an error, we would rather know. Use Report an error on any page.
Credibility labels
Each case carries one label describing the state of its record, not whether the event “really” happened:
- Official record: documented by a government, court or institution at the time.
- Unexplained / well-attested: several good sources, and no explanation that fits the details.
- Explained: a specific explanation has been shown for this case.
- Disputed: the facts or their explanation are actively contested.
- Single account: one witness or one source.
- Unverified claim: asserted, with little or no independent support.
- Hoax: shown to have been fabricated (a confession, a forgery demonstrated, an investigation’s finding).
- Legend / folklore: a story handed down, valued for what it shows about belief.
How explanations are weighed
For many cases the atlas gives its own reading of how likely each explanation is. This is an editorial judgement, set out openly so that it can be challenged.
- An evidence profile. Seven factors are scored from 0 to 5: the number of witnesses, their independence, instruments and sensors, physical traces, how soon the event was recorded, investigators’ access, and the consistency of the accounts.
- Every serious explanation, with a share. Each kind of case has its own set (for an aerial sighting: something known misidentified, a natural phenomenon, perception and memory, secret human technology, a hoax, something genuinely anomalous, and so on). The shares add up to 100, each with its reasoning.
- Undetermined is its own share. When the record is too thin or too mixed to choose, that uncertainty is shown as “Undetermined”. It counts neither for nor against the account.
- Uncertainty is not fabrication. A “hoax” share needs evidence of fabrication. Where a hoax share rests only on doubt, it is moved to Undetermined; an unproven specific suspicion keeps at most a small share.
- Generic explanations are not assumed. An ordinary explanation that has not been shown for the specific case (misidentification, a natural phenomenon, perception, the surroundings, a cultural story) is capped at 40% for official and well-attested cases and 50% for disputed cases, single accounts and claims. The remainder goes to Undetermined, and the page says so. This is a deliberate choice: the atlas treats general explanations as hypotheses to be shown, not as defaults. Reasonable people weigh this differently, and experts can propose other figures.
- Official verdicts are shown separately. They are reported for comparison and weighed as claims, never used as inputs.
- Researched and automatic. 597 cases have a researched assessment written from the case file, the scientific literature and the sources. The rest have an automatic estimate from their evidence profile, labelled as such until they are researched.
What is automatic
- Automatic estimates of explanations for cases not yet researched (above).
- Case types: recurring kinds of case found by grouping the descriptions statistically (48 groups). They are a way to browse, not a classification anyone has endorsed.
- Subcategories such as craft shape or kind of haunting are tagged by rules from each description. Unclear cases are left as unknown, and a wrong tag can be reported.
- Proposed connections in Discover are ranked from the atlas’s own records. They are invitations to test, not findings.
Dates, places and pictures
- Founding dates of traditions are approximate, and traditional dates are marked as traditional.
- Places are as precise as the sources allow. For legends and regional traditions they mark an approximate centre.
- Pictures come from Wikipedia and Wikimedia Commons, with their credits. Where no faithful image of a god exists, the atlas shows an emblem rather than an invented likeness.
- Undated items appear on the map only from 1900, so that they do not sit in a past they cannot be placed in.
What the atlas cannot know
- Coverage is uneven. English-language and Western sources are over-represented; many traditions are described from outside; oral traditions are known mostly through others’ records.
- Summaries simplify. A paragraph about a tradition cannot carry its internal diversity or its own self-understanding. Follow the sources.
- Records of strange events are records. Including a case documents that an account exists, not that it is true or false.
- Assessments are judgements. They are made openly and can be wrong.
Experts, corrections and moderation
The atlas is meant to be added to by people who know these subjects better than it does.
- Perspectives: signed views on any case, tradition, god or theory, with evidence, what would change the author’s mind, and the strongest case against their view. Contributors can be verified through their ORCID record.
- Proposed changes to an assessment, credited by name when accepted.
- Open questions: what is still unknown about an item, and who has taken it on.
- Report an error, on every page.
Moderators check that a contribution is on topic, sourced and civil. They do not judge whether they agree with it. Contributions are published under Creative Commons Attribution 4.0, with the author named.
Citing and reusing
Every page has Cite this, and every signed perspective has its own permanent link and citation. The atlas’s data can be downloaded from Open data. Please cite the original sources for any claim you rely on; the atlas is a guide to them, not a replacement.