[redacted][unpublished]`, `64-lamps-on-th…

These three research agents are running in the background. I'll wait for their results before synthesizing the consistency report.

Noted — scene extraction complete. Waiting on the canon digest and the scenes 25-62 back-reference extraction before synthesizing.

Noted — back-reference extraction complete. Still waiting on the canon digest agent before synthesizing the full report.

Confirmed with direct quotes. Now I have everything needed for the final report.

Canon consistency report: scenes 63–68

1. Hard contradictions

• Bone-setter distance vs. binding-list distance. Scene 67 unpublished redacted sentence: unpublished prose redacted."* Twelve miles ≠ forty miles. Also scene 67 unpublished redacted sentence: unpublished prose the binding note's own "eight weeks in a redacted" — the two figures the binding list itself uses for this beat (forty miles, eight weeks) don't match what actually landed on the page (twelve miles, ~three months). Fix: either change scene 67 unpublished's distance to "forty miles" (and adjust "he'll be here by dark," which doesn't work for forty miles on a fen track) or update SCENES.md/canon to record twelve miles and reconcile the healing.md "forty miles off" framing, which this scene is supposed to dramatize.

2. Deferred questions resolved casually / binding-list breaks

• None found. The redacted is cast (scene 68 unpublished) but not framed as "the seed" of a redacted; redacted's competence is sourced to her trade ("my first warden's rule…") not her sealed background; the redacted-refuses-water beat (scene 66 unpublished) draws no clever in-scene comment — Edwin notes it privately in writing, unremarked by anyone else; the engine is never invoked at the mouth in 63–68; the silent-soil/Deep-adjacency question is untouched. All compliant.
• Numbering drift, not content: SCENES.md line 84–86 cites "the redacted on a Fenholt beast after redacted (scene 66 unpublished)." The actual casting is in scene 68 unpublished; scene 66 unpublished only sets up the fen-margin fight and the Seal-back-at-midnight beat. Either the binding list's scene number is stale or the beat has moved. Flag for whoever reconciles SCENES.md against final numbering.

3. Day/date/hour errors

• None found. Every clock reference in 63–68 checked against timeline.md's Day 29/30 entries and matched exactly: redacted's arrival 14:20 (scene 63 unpublished "twenty past two"), church-tower working 15:30–19:30 (scene 64 unpublished "half past three" / "half past seven"), redacted's bell 18:00 (scene 64 unpublished "at six"), 290 ewes on the green 20:10 (scene 65 unpublished), full dark 19:40 (scene 66 unpublished "twenty to eight"), dogs at withy-bed post 23:25, second gate 23:52 ("eight minutes to midnight"), road held to 01:10, mist at 04:00 (scene 67 unpublished "as promised"), sluice-keeper found 09:00, noon count. "It's the thirteenth" (scene 63 unpublished) = Day 29 correctly.

4. Sums, distances, counts, times that don't add up

• Item 1 above (bone-setter distance/redacted duration)** is the one clear numeric fault.
• Sheep-count possible collision: scene 64 unpublished redacted sentence: unpublished prose unpublished redacted sentence: unpublished prose If they're different locations (a general staging pen vs. Sennet's own flock's final destination), no conflict — but the draft doesn't clarify which, and a reader will likely conflate them. Worth a one-line disambiguation.
  redacted sentence: unpublished prose unpublished, Day 30) — verify against the established delve date. The leakage-count series in canon world files runs …Cornmonth 19, Harvestmonth (break 13th/14th), **delve (21st), Fallowmonth 11 (posted 30th)…. If "the 21st" there means 21 Harvestmonth (7 days after Day 30), it contradicts "eighteen days." It is consistent, however, with the muster date named in scene 61 unpublished, "the second of Fallowmonth" (16 remaining days of Harvestmonth + 2 = 18) — so scene 68 unpublished's figure likely tracks that date correctly, and the ambiguity is in which canon entry ("delve 21st") the digest was reading, not necessarily an error in the scene. Recommend the author/checker confirm directly against timeline.md's dated delve entry before treating this as clean — I could not get a fully pinned citation for "delve (21st)"'s month in the time available.
  redacted sentence: unpublished prose redacted + one other thing, ~20 uncounted, ~200–400 loose) are internally consistent across scenes 66/67 and match canon's Day 30 timeline entry verbatim — no faults.
• redacted redacted sentence: unpublished prose unpublished) implies he started ~22:00 on the 13th, but scene 64 unpublished has him "at table writing names" from around 18:00 (the bell). That's closer to 18 hours than 14. Minor, worth a look.

5. New facts scenes establish that canon does not yet record

• The full nine-steadings list (scene 65 unpublished redacted sentence: unpublished prose redacted's, the two Hobbs, the redacted redacted sentence: unpublished prose unpublished only asserted the count of nine; this is the first on-page enumeration and should be backfilled into world/geography.md.
  redacted sentence: unpublished prose
• Edwin's lifetime tally for the redacted redacted sentence: unpublished prose
  redacted sentence: unpublished prose unpublished, Peri to Edwin) — establishes an off-page teaching scene between Edwin and Mrs Tansy, witnessed by Peri, not previously on record; Mrs Tansy's only established line (scene 07) is about her own flock in a wet spring, which this doesn't contradict but does supplement. Should go into characters/minor.md under Mrs Tansy.
• New minor characters/props introduced in scene 65 unpublished: Tess and Burl (Mattie's hands?), Tess's cow (likely the "one cow" in the casualty count), Mattie's "far bay" of eleven acres of hay, deliberately left unguarded. None of these appear in characters/minor.md yet.
  redacted sentence: unpublished prose redacted date for the first time in dialogue and should be logged in timeline.md.

redacted sentence: unpublished prose unpublished) are consistent with any earlier-established patrol dates — no canon reference to "the tenth" for causeway checks was found either confirming or denying this, so it reads as new, uncontradicted material rather than an error.