How an entry earns its place.

Evidence before narrative.

Each entry begins with the claim it must support, then identifies the relevant evidence: fossils and their geological context, comparative anatomy, development, biogeography, observed population change, or molecular data.

A compelling story may organize evidence, but it may not outrun it. We distinguish direct observations from interpretations and reconstructions.

Dates are measurements, not decoration.

Numerical ages are reported at a precision appropriate to their source. “About,” ranges, and uncertainty are retained where the record does not justify a single date. Geological boundaries follow current international stratigraphic usage when relevant.

• Relative dating establishes sequence through stratigraphy and correlation.
• Radiometric methods estimate numerical ages with stated assumptions and uncertainty.
• New calibration, provenance, or phylogenetic evidence can change an interpretation.

Trees are hypotheses.

A phylogeny is a testable model of relationship, not a portrait gallery and never a ladder of progress. Nodes represent inferred common ancestors; living tips are cousins, not stages on the way to another species.

Simplified diagrams may omit lineages for readability. Omissions are disclosed by context, and uncertainty is described where competing topologies matter.

Revision is part of the product.

Substantive corrections should update both language editions. Sources are recorded per long-form entry, and major editorial pages preserve a shared reference register.

AI may assist illustration, drafting, or translation, but it is not treated as a source. Claims must be traceable to human-authored scholarship or institutional collections; generated reconstructions are identified as interpretive artwork.