Unit 83 (counted test of P21): results of the registered run, mn73w

Unit 83: results of the registered run (attempt 1), 27 and 28 October 2026. The registered counted test of P21, the placement limb of Law V of the ARC Theory, registered on OSF on 22 October 2026 as mn73w (DOI 10.17605/OSF.IO/MN73W) and public before it ran. P21 is supported under the registered test, by the programme's own run (Unit 83 (study identifier 41-D-COUNTED-v0.1)), on the access axis: in a counted, administered revision loop on two fixed-weight engines of two providers (Claude Haiku 5.5 and GPT-6 Luna), pooled with equal weight, correction placed inside each round led a monitor that saw only the finished register by +0.5560 (95.5 per cent interval +0.5509 to +0.5611) on average, a lead the value's definition makes likely whenever the corrector repairs at more than half the rate at which the monitor discriminates and which by itself counts for nothing; and the per-round lead widened by +0.0100 (95.5 per cent interval +0.0092 to +0.0108) a round over eighteen rounds, more than the registered tolerance of 0.0075. Mechanism axis NOT EVALUABLE: the monitor and the twin, shown the same bytes, did not agree within the registered interval of the negative control. The mark is computed with an entry kept out of the register scored 0, the registered value; recomputed without a label, as the design requires, the widening is +0.0132 a round (95.5 per cent interval +0.0122 to +0.0142), above the tolerance, with such an entry scored minus 0.5, and +0.0067 (+0.0060 to +0.0074), inside the band, with it scored plus 0.5. The result is family-dependent, and one call was reissued under a disclosed post-registration recovery amendment (RA01). It is one owner's run and has not been replicated. The files: the report as written, its figures, the amendment disclosure, a results record valid against the kit schema with its provenance, the registered words, the sealed run archive (every request and reply) with its SHA-256 manifest, and an OpenTimestamps receipt. The README says how to check them. Added on 29 October 2026, after the run: a post-run diagnostic of the negative control, in the files whose names begin POST-RUN-DIAGNOSTIC. It is not part of the registered analysis and changes no mark and no axis. Michael Darius Eastwood conceived and directs this research programme and is the author of this work. Across the programme, he has used more than six AI systems in parallel, under his own instructions, to stress-test his arguments, identify possible errors, and assist in preparing draft text from his own outlines. He determines what is adopted, revised or rejected and takes responsibility for the published content. These systems are tools, not authors.