{
  "_meta": {
    "name": "p26",
    "photo": "26.Martin-MacFarlane.jpg",
    "anchored": "2026-05-13",
    "task": "Phase 10 belt-y FF anchor capture",
    "method": "Central saturated-core centroid, warm-index side-parhelion scan, and visual confirmation against 22-degree halo diagnostic overlay.",
    "status": "belt-y FF anchor with parhelion-y tilt flag; do not treat as a new inversion-route verdict",
    "inferred_h_deg_from_anchor": 8.97,
    "note": "Left and right parhelion y picks differ by 12 px, so this should also feed the candidate parhelic-tilt analysis if that branch opens."
  },
  "sun": [464, 204],
  "r22": 323,
  "r22_source": "parhelion-derived",
  "r22_source_note": "p26 shows no continuous 22° halo arc — only parhelia. R22 was set from the parhelion offsets, so the parhelion-route residual is tautological on this photo. Audit memo §2 item 7 confirms. Pass B1 classification 2026-05-13; ineligible for parhelion-route promotion regardless of residual.",
  "geometric_validity": {
    "left": "valid",
    "right": "invalid",
    "right_invalid_reason": "right_offset = right_x - sun_x = 786 - 464 = 322 px is less than r22 = 323 px, so arccos(r22 / right_offset) = arccos(1.003) is undefined. Geometric impossibility — the right parhelion cannot lie inside the 22° halo and still be a parhelion. Audit memo §2 item 6: 'p26 right-side anchor encodes a geometric impossibility ... the residual table records this with no flag, demonstrating the table does not actively enforce the geometric constraint that defines the route.' Pass B1 fix 2026-05-13: flagged here at the schema level so the residual table can render `invalid` rather than silently coerce."
  },
  "parhelion": {
    "left_x": 132,
    "right_x": 786,
    "left": [132, 183],
    "right": [786, 171],
    "left_y": 183,
    "right_y": 171,
    "y": 177,
    "tilt_flag": "left/right y differ by 12 px"
  },
  "parhelic_belt_y": 177
}