Browser / publication-scale preserved results:

Icarus 1566 — original H2-087 blind SOURCE_ONLY lane
- R20 FV: 10.0605278916 arcsec/century
- R40 FV: 10.0623924511 arcsec/century
- R80 FV: 10.062877461882765 arcsec/century
- post-seal R80 ruler: 10.063089091251317 arcsec/century
- R80 relative difference: about -0.002103%
- stable response / D: 1.998473, 1.999585, 1.999874 for R20/R40/R80

Named-case transfer browser rows
- Mercury, Venus, Moon, and Europa expose preserved H2-089 PAIR_TOTAL_EVERYWHERE comparison rows.
- The browser marks reconstructed absolute values with '~' when only rounded archived relative-error rows are being displayed.
- Venus and Europa are low-eccentricity conditioning controls; do not interpret their raw residuals as clean physical falsifications.
- Moon is an isolated central-source relativistic comparison; raw observed lunar apsidal motion includes large classical perturbations.

Frozen chronology / mechanics guardrail
- H2-087 blind Icarus was sealed before the GR/context ruler opened.
- H2-089 later selected M_pair = M1 + M2 as the working universal two-domain bookkeeping rule by FV-only mass-ratio homogeneity stress.
- H2-092 structural response multiplicity = 4; controls = 3 and 2.
- One conservative FrostTransit handoff remains one D-scale work ledger. The factor four multiplies response, not stored energy.

The default Python wrapper runs a smaller historical validation profile. Use `python lab_test.py --full` for the publication-scale H2-087 profile, `python lab_test.py --pair` for H2-089 pair-total validation, and `python lab_test.py --secondary` for H2-092 structural-response validation.
