# Evidence for job #3394 (findings #563, #806, #809) on `paper/moire-primes.md` Served sha256 bab9e4356999b16d8f4b3d1e457496bea9238bd8ecdba2ba2f8e15340402698a, 57465 B, 887 lines (the `content_sha` of all three findings; also return #1257's `moire-primes.md`). ## #563, the coefficient at p = 4999, by direct count The parenthesised list in section 6 is the `coef` column of return #1257's own `checks1432.out` (sha256 e2151044a3b1d3047e2ccb263143dd256057c842f0df69ddfd21fc6159f7bdf7), which prints `17 K 54 coef 1.8114`, `101 K 1226 coef 1.7865`, `p=1009: K= 79661 quotient= 12.7677 quotient/ln p= 1.845914` and `4999 K 1564687 coef 1.8748`. So 1.81, 1.79, 1.85 are that file's numbers and only the fourth was stale. Recomputed here by sieve of Eratosthenes to 4999^2: pi(4999^2) - pi(4999) = 1564687 = the K checks1432.out prints, and (p^2 - p)/(K - 1)/ln p = 1.874847 -> **1.87**. ## #806 + #809, the seam-z span, against the producer's stdout `research/fold-profile-13-hotspot-sweep.js` (served, sha256 0ebd512033257c82aca3130512fd917c239b81f0e23bcd25e7200d5c43472177) re-run here: `node` v22.23.3, rc 0, 81 stdout lines, sha256 88704cc3dd7d4eb817c65aab07a21d47c606bf3ca2cabf090238cc09427e2b37. Its two seam tables give, at the half-widths the file names: - T19 (W = 9,699,690), half-width 3,000: z = **-1.42** (then -0.07 at 30,000, -0.13 at 300,000); - T23 (W = 223,092,870), half-width 30,000: z = **+0.26** (then -0.68 at 3,000, -0.53 at 300,000). The z column prints an explicit sign only for negative values, so the fresh stdout shows `0.26` where the file's embedded table writes `+0.26`; it is the same value. The six cells are the span the sentence calls "a range over six measurements", and the pair the file quoted (-1.41, +0.13) is the *older* table the same file records at its lines 349-352 (`-1.41, -0.19, -0.09 at T19 and -0.65, +0.13, -0.50 at T23`), which the 400-control version superseded. ## GUARDS On both the served text and the revision: `fences = 0`, `code-sha256 = 0`, `out-sha256 = 0`. The file has no fenced content and no hash-bound OUTPUT block, so nothing has to be re-embedded and the revision's own diff is two replaced lines. The producers were still run, because they are what the changed numbers cite. ## Left open (not this job's path) `GLOSSARY.md` line 66 carries the same stale span -1.41 to +0.13 for the same producer; finding #809 says so and it is outside `paper/moire-primes.md`. Not touched here.