Saved artifacts for the measurement-contribution study

Inventoried 318 final primary fits across 53 stretches, with no original-recording reads, native predictions or new fits. The published model comparisons remain 53 globe / 0 flat / 0 unresolved. This inventory asks which exact saved arrays can support another diagnostic.

18 stretches have all six required saved primary artifacts ready for a matched projection comparison. The downloadable inventory records each authoritative result origin, declared artifact/hash, resolved public copy, array dimensions, time-context checks and any reason preventing reuse.

“Not saved” refers to a residual/Jacobian artifact, not an unfinished model comparison. A missing artifact is never replaced by a lower-cost historical fit or by a repeat. Any later replay at saved parameters needs a separately frozen finite allowance.

The inventory checks receiver layout and a consistent relative-time origin; it does not independently establish the instrument clock. It changes no measurement, interval, parameter or published fit label.

Finite plan · Measurement-contribution scope · Current findings.