{
  "session": 62,
  "scope": "Derived boundary map over all73 parsed master text/data pages (15 LP1 +58 LP2). Preserve master tokens and add only explicitly aligned source evidence from Sessions54,59,60,61. This is not a full image census.",
  "inputs": ["liber-primus__transcription--master.txt", "source_event_ledger.json", "separator_pixel_results.json", "three_dot_review_results.json", "typography_review_results.json"],
  "representation": "Every non-rune gap retains ordered literal non-whitespace tokens, line/page position and separately deduplicated physical events. Co-located red events remain separate. Three inline '-' discrepancies receive an observed-three-dot overlay. Split3+1 and1+3 groups remain paired fragments with unresolved interpretation, not automatic periods.",
  "known_controls": [
    {"name":"second_koan","images":["14.jpg","15.jpg"],"key_runes":"ᚠᛁᚱᚠᚢᛗᚠᛖᚱᛖᚾᚠᛖ","method":"Subtract repeating key mod29, continuous across both pages, initial phase0.","unenciphered_F_rune_offsets":[49,58],"source_note":"Published skip positions65,76 count characters in the space-containing input, one-based. Normalized retained input confirms they are rune offsets49,58. Baseline reproduced before defining these counterfactual boundary checks."},
    {"name":"LP2_page56","images":["73.jpg"],"method":"Subtract successive primes minus1 mod29; initial prime index0.","unenciphered_F_rune_offsets":[56]},
    {"name":"LP2_page57_direct","images":["74.jpg"],"method":"Direct reading; no key counter. Diagnostic example of a control unable to distinguish counter-reset hypotheses."}
  ],
  "policies": ["continuous", "reset_at_observed_three_dot", "advance_one_at_observed_three_dot", "reset_at_observed_four_dot", "reset_at_audited_red_mark", "reset_at_page_boundary"],
  "rules": [
    "Use only individually aligned physical events for the punctuation policies. No unaudited literal period is promoted to an observed four-dot event.",
    "At a selected gap, act before the next rune. End-of-stream marks cannot affect plaintext. Page resets apply only inside a multi-page control.",
    "Keep known keys and F exceptions fixed; do not refit phases, keys or exceptions to rescue failed policies.",
    "A reset occurs once per selected gap; coincident physical events stay separate in the ledger.",
    "Report rune mismatch counts, first mismatch, active event counts and counter changes. Counter changes need not change periodic key values.",
    "No-event and no-key cases are uninformative, not evidence supporting a boundary rule.",
    "Check baseline against retained known prose and re-encrypt. Compare rune values without changing wording to improve a score."
  ],
  "limits": "These controls test whether a proposed rule can be universal across the selected known methods. They do not prove an unsolved section shares the same key schedule. No unsolved decryption or key search in this session."
}