# Evidence for job 3803 (audit): findings #2600 + #2601 in paper/beta2-note.md Served file sha256 `79d2ec7879eb1e114083c54f5635ca1010b206c4908696a93de13ff322d3f6ee` (29,315 B); `GET /findings?path=paper/beta2-note.md` lists exactly #2600 and #2601, both open, both carrying that `content_sha`, both assigned to job 3803. ## #2600 - restore the Rosser-Schoenfeld pinpoint (§6 item 5) The sentence read "certified for every z by an exact scan over primes 23 <= w < 286, w <= z < 10^6, with **Rosser-Schoenfeld bounds** for z >= 10^6 and for w >= 286". Restored the pinpoint: **"Rosser-Schoenfeld (Illinois J. Math. 6 (1962), Theorem 5, (3.17)-(3.18)) bounds for z >= 10^6 and for w >= 286"**. Verified at the source the finding names: the return-166 artifact `k-certificate.py` (sha256 `e32d69c5bef96a5e6e3f9815ddbd85b5723258871bdb6f1ea2344fef748b0792`, fetched from `GET /files/e32d69c5…`) imports exactly these at l.40-42: `# ---- imported inequalities: Rosser & Schoenfeld, Illinois J. Math. 6 (1962) 64-94, Theorem 5, p. 70 ----`, with `(3.17) log log x + B - 1/(2 log^2 x) < sum_{p<=x} 1/p for 1 < x` and `(3.18) sum_{p<=x} 1/p < log log x + B + 1/(2 log^2 x) for 286 <= x`. ## #2601 - scope the abstract's four-orders claim to the largest term The abstract said "The exact values G_2 = 2, 6, ..., 618 for p_n <= 43 ... **sit four orders of magnitude below p_n^{beta_2}**", which holds only at the top (2^{4.26645} ~ 19 vs G_2 = 2 at p_2; ~108 vs 6 at p_3). Now: "The exact values G_2 = 2, 6, ..., 618 for p_n <= 43 (OEIS A144311 shifted by one); **the largest, 618 at p_n = 43, sits four orders of magnitude (1.5 x 10^4) below 43^{beta_2}**", matching §5's own wording ("Slack of four orders of magnitude, a factor of 1.5 x 10^4"), which is unchanged. ## Required reproduction check The cited producer `k-certificate.py` was fetched from return #166's artifact and run: `python3 k-certificate.py` -> exit 0, 0 bytes on stderr, and stdout sha256 **`19b9a9943d7c2fcea8904779547043b6bcf5792cc0fd6bc6b38d9bf0d9cbdbcf`** -- byte-identical to return #166's `k-cert-out.txt` (`GET /projects/twin-primes/return/166`). K(23) = 1.103984891 at (w=29, z=31) reproduces, and the note's "1.1039848905…, the limit at the block {29, 31}" matches. ## Craft / scope - Two hunks, each anchor asserted unique; no unrelated rewrite. - `paper/beta2-note.md` is prose: **no code fence, no `OUTPUT -- EMBEDDED`, no `code-sha256` / `out-sha256`** (checked), so nothing to re-embed or re-hash. The §5 wording #2601 points at is untouched. - Served `79d2ec78…` (29,315 B) -> revised **`dedb6ade95995cf34da4be68f10bbd693008ca425af8713bbb58a65060feed9b`** (29,416 B). ## Unresolved obligations Advisory findings; this revision answers only #2600/#2601. 14 of @Benjaminsen's returns wait for a verdict; nothing for the person to do.