About the data

What counts as a record?

Entities, events, properties, assignments, claims, causal hypotheses, myth interpretations, relationships, and dates can each carry their own status, confidence, explanation, and sources. This prevents evidence for one statement from being borrowed silently by another.

How chronology works

Dates are assertions, not formatting hints. A range remains a range; competing chronologies remain separate. The database uses astronomical integers for correct ordering, while the interface converts them back to BCE or CE and never displays a year zero.

Definitions are not assignments

A property page answers “What does this term mean?” A property assignment answers “Why does this record receive that term?” See New Kingdom Egypt and Peleset.

Myth is not archaeology

Traditional narrative, archaeological observation, and modern symbolic interpretation are presented as different layers. The Trojan Horse view demonstrates why a plausible association is not automatically an established fact.

Confidence and source criticism

Confidence is always stated in text and never encoded only by color. Source pages expose reliability notes and semantic backlinks. Metadata and links do not grant permission to redistribute copyrighted publications.

Who reviewed this?

Source authors wrote the cited scholarship. Imported research collections contain the assessments shown on record pages. A dataset workflow label such as “reviewed for v0.1” is not documentation of specialist review.

The reading journeys were prepared with Codex on 6 September 2026 from the linked research records and selected institutional sources. They are editorial syntheses awaiting specialist review. No specialist reviewer has been attributed to these new introductions.

Record pages expose a named specialist and review date when supplied; otherwise they say “not recorded.” Evidence confidence concerns a specific claim or assessment, not editorial approval or an automatic measure of scholarly consensus.

How the reading view groups evidence

Explicit account groups bring separate descriptions of the same occurrence together while preserving each original record and its dates. A group count is not a source count, and sources shared between accounts are not independent corroboration. Broadly similar changes at different places remain separate.

Timeline context separates intervals over 200 years and modern research observations from events and shorter phases. This is a browsing aid; a shorter range can still be uncertainty rather than duration. Associated regions are navigation tags, and research collections are labeled separately.

Site markers use supplied or identified site locators. Regional associations without site-level placement are kept in a separate list. Approximate political extents and research areas have different meanings.

Editorial changes · 6 September 2026

• Introduced the four-city investigation, source-led reading journeys, place introductions, glossary and reading notebook.
• Made unsupported interpretations explicit before confidence badges and exposed source notes.
• Corrected Cindy Kocik’s name in the 2023 drought citation against the publisher record. Original release data remains preserved.
• Grouped identified accounts, separated long-running context, and removed invented regional marker scatter from the map.
• Separated dataset maintenance status from specialist review attribution. New editorial text awaits specialist review.

Annotated starter reading · Terms and chronology