Antarctic data fields · Terra Chartarum

Why these fields exist

The Antarctic material is an argument about maps made when there was nothing to observe, when what was observed was disputed, and when the ground itself was moving. That argument only holds if the difference between a guess, a report, a sighting and our own drawing is carried by the data rather than by the prose beside it. These are the fields that carry it.

Evidence class

What kind of claim a record makes. The stored values do not change; only the labels a page shows may.

Value	What it means
`conjectured`	Drawn because theory implied it should exist, with no report behind it
`inherited_cartography`	Taken from an earlier map rather than from any new information
`reported_not_observed`	Represented from testimony that was not independently confirmed
`direct_observation`	Seen by a named expedition on a stated date
`instrumental_fix`	A position from a celestial observation and a timekeeper
`dead_reckoning`	A position inferred at the time from course, speed and elapsed time
`scholarly_reconstruction`	Reconstructed by later scholarship
`editorial_interpolation`	Supplied by Terra Chartarum to join records the sources leave apart
`later_confirmation`	Evidence that a claim held
`later_disproof`	Evidence that it did not

Two of these are worth holding apart carefully. Dead reckoning is a historical act: a navigator inferred a position and recorded it, and that inference is evidence. Editorial interpolation is ours, made now, and is never evidence of anything except our own drawing decision.

Observation is not accuracy either. A record can be a direct observation and still be wrong; what the class asserts is that somebody looked.

Geometry provenance

Every line, outline and point declares where its coordinates came from. There are six answers, plus one for a record that has no geometry at all.

Records drawn by Terra Chartarum, whether generalisation or interpolation, can never reach the public tier. That is enforced in the build rather than promised here.

not_spatial is a real answer and not a gap. A record about a plate nobody has examined stays without geometry; the alternative is a point that looks like knowledge.

Confidence

high, medium, low, contested, unresolved.

contested and unresolved are not weaker versions of low. contested means scholarship disagrees, and unresolved means the question is open. Neither is a defect in the record, and a record at either may not be presented to a public reader as established.

Review state

raw, normalized, reviewed, approved, published. A record leaves raw only once its source locator names something a person actually read. Naming a source is not reading it, and the two are different fields for that reason.

Later status

confirmed, modified, disproved, unresolved, not_applicable. Held apart from the original evidence, so that a feature later removed from the charts keeps both its claim and its disproof. A disproved feature is a record of what was reported and believed, and is never stored as merely an error.

Discovery vocabulary

The words around 1820 do the most damage when they are used loosely, so they are defined before they are used: sighting, first sighting, first mainland sighting, ice front observation, first landing and confirmed land are each a different question. The dataset has no field called discovery, because that word bundles seeing, recognising, reporting and being believed, and carries a priority argument with it.