Planned before any run: the planted-convention check's pass rate on v3 at N = 1,850¶

Audit item 5. The Stage 1 draft registers outcome-neutral check 5: on the development window, one run with each half's own comment marker planted in a quarter of its refinements must pass on both halves. Its one pilot (v1, bf16, random halves of 183 changes each, one seed) passed on the weaker half at a lower bound of +0.005, and on a partition the halves hold 125 and 81 development changes. A positive control that halts a working apparatus is not "sufficient" in the sense the registered-report guidelines use, so its pass rate is measured before it is registered.

Runs. partition_run.sbatch with PLANT=0.25 (placebo_corpus.py --plant, the symmetric calibration's markers and seed, planted over every window of each half), on the first ten admissible partitions (seeds 2, 3, 4, 5, 7, 8, 9, 10, 11, 12) at training seeds 1 to 10 and N = 1,850, corpus v3. A run passes when its verdict is pass: both halves' 95% lower bounds above zero.

Rule, fixed now. Check 5 is registered as it stands only if all ten runs pass and the pass probability projected from the ten runs' per-half estimates and interval half-widths (normal, per half, product over halves) is at least 0.95. Otherwise it is redesigned before registration: the plant raised to a half and the same ten runs repeated, and failing that the check read over several runs. The rates are logged whichever way they fall.