About · Methodology

Evidence before conclusion

Every statement on a record is tied to a kind of evidence. The kinds are never blended into a single score.

Kinds of evidence

Published archaeology
Peer-reviewed publication, archaeological report or recognised professional source.
Official record
State or institutional archaeological record.
Historical source
Map, antiquarian account, manuscript or historic publication.
Folklore / oral tradition
Recorded cultural tradition or oral account. Evidence of memory, not of events.
Beyond the Bend observation
Direct field observation by the Beyond the Bend team.
Aerial observation
Observation derived from drone or aerial imagery.
Computational result
Generated from a geographic or astronomical calculation. Inputs are recorded.
Hypothesis
Interpretation requiring further evidence.

Confidence is not truth

Where confidence matters — especially in astronomical work — it is assessed separately for each question: geometric agreement, measurement uncertainty, statistical significance, archaeological plausibility, intentionality. Intentionality is treated as unknown unless independent archaeological evidence supports it. “Unknown” is a legitimate and useful result.

Folklore

Folklore and mythology are approached as historical and cultural sources: evidence of how landscapes were remembered, named and interpreted over time. They are not treated as proof of archaeological events, but neither are they dismissed when they may preserve useful patterns of memory, place and tradition. Each folklore record keeps the recorded tradition separate from its modern interpretation.

Locations

Each record states how precisely it may be located publicly:

Vulnerable or unpublished archaeology may be shown approximately or withheld. Records also state the monument's condition:

Destroyed monuments remain records; they are drawn differently on the map, not removed from it.

Calculations

Solar directions are calculated with published, versioned algorithms. Each result records its inputs, assumptions (including horizon altitude and refraction), algorithm version and limitations, so it can be reproduced. A geometric agreement between a monument and an astronomical event is reported as geometry, not as evidence of intent. When patterns are searched for across many monuments and events, the number of tests and the result expected by chance are reported alongside.