# n209 — The repairs: three scripts fail closed, two papers rescoped, the manifest updated

**Date:** 2026-07-11 · Executing the corrections tier after the audit ledgers (n206–n208).

1. **n113 (Mercury) repaired per audit gate 29.1**: the declared first-order solver is now implemented and primary — it reproduces Δβ = 2.1630·ξ at 0.0% linearity residual — and the Newton path **fails closed** (verified: raises NonConvergence at the fitted ξ = 0.2312 instead of printing a sign). The §25 warning prints permanently: the calibration stays VOIDED until a stable nonlinear branch exists.
2. **n123 (compact winding) repaired per audit gate 47.3, first half**: the U2 gate can now fail — and it does: `FAIL: no charge ladder (windings unwind for N = [1,2,3])`. The executable now agrees with the audit instead of its own old summary. The even six-cycle construction remains owed.
3. **n136 (KZ relic) fails closed** on zero-defect samples (was: log(0) → NaN → fail-open) and names its supersession by kz-rust (n200).
4. **Papers rescoped and rebuilt**: paper I's abstract now carries the §2 attribution (the constraint owns the pole; any positive cost reproduces Coulomb; the metric prices it — no uniqueness claimed); paper II's abstract carries the §§40/43 scope correction (the keV band is dimensional anchoring; the executables do not reproduce the quoted geometry; the band is conjecture pending gates 47.3–47.4).
5. **n175's quoted numbers withdrawn on the status page** ([no executable — until reproduced]); n159 likewise flagged via the manifest.
6. **The manifest now tracks both directions**: 2 defects remaining in queue (n48 re-registration; n175/n159 reproduction), 3 repaired with dates — printed on every suite run.

**Remaining in the corrections tier: nothing.** The queue is now purely constructive: the n48 ω-re-registration with multi-pole comparison, the n175/n159 executables, and the three decisive constructions.