The Divisor: the Same Leak, Counted

The proton's own correction carries the construction, and this paper takes it from the AP30 file directly, quoted at its own hedges. The manifold has 21 geometric channels and resolves into 4 spacetime dimensions — three spatial plus one temporal (AP10, AP20). Each channel expresses through each dimension: 21 × 4 = 84 dimensional expressions. The dynamic coupling leaks isotropically into the 3D substrate, and the leakage per dimensional expression per solid angle is 1/(84π) — the 84 in the leakage denominator being the same 84 as the static count. AP30's own sentence is the principle: the leakage is normalised by the very structure it leaks through.

Count the window's leak the same way. The structure the flip's leak runs through is not the arena — a distinction's static count is its orientation pair through the four dimensions: 2 × 4 = 8 dimensional expressions, each leaking under the same lemma. The divisor is 1/(8π). One construction, two supports: the proton's correction is a thing held in the whole arena, support 21; the flip's is a distinction, support 2, because the window's configuration space is the pair. The support difference is the scope itself — thing versus distinction — so the divisor's derivation costs the paper one sentence it already owed.

Why four dimensions, not three. Because "held" is the scope's own word, and to be held is to persist. The pair's three spatial expressions are the orientation's projections; its temporal expression is the holding — a difference with no temporal expression is an instant, not a window, and an instant prices nothing under this paper's own definition. So the count runs through all four by the scope, not by analogy. The transfer also preserves AP30's own split exactly: there too the static count runs through four while the radiation goes into three-dimensional space's solid angle. (For the record only, and never as a reason: a three-dimensional count would give 6π = 18.85, outside the window — but the argument above stands with the window covered, as KS-FLIP.2 requires.)

Two earlier readings retire from the derivation chain, on the record: 8π as "four conditions times the delta's native circle," and an identification of the four dimensions with the four conditions {S, B, R, C}. The first is true arithmetic and not the forcing. The second is not in AP30 — the file says three spatial plus one temporal, and the construction never needed the identification; it was deleted, not repaired, when the file check caught it.

Order and geometry are independent axes, and canon proves it in one paper. AP30 carries a first-order term wearing π (the 84π expression-leakage) and a second-order term wearing none (the recursion α²×21×16/1836, the leakage of the leakage, no solid angle, its coefficient owed under KS-30.3). The π tracks geometric expression-leakage, not the power of α. The arena's 1/(1+α) is first order without π; AP30's maintenance is first order with π; AP30's recursion is second order without π; the Flip is second order with π — an expression-leakage at ledger depth two. No canon rule stands against it, because canon's own pair shows the axes come apart. This paragraph was written before the grade was decided, as review demanded.

Inheritance, on the face. Lemma 4's leakage is assumed perfectly isotropic, and AP30's KS-30.2 is live on exactly that assumption: shown anisotropic, the 1/(84π) correction changes — and with it, by the same lemma, the 1/(8π). The Flip inherits KS-30.2, and AP30's own section 8 notes its 5 ppb residual is consistent with a small anisotropic contribution, so the shared edge is genuinely live. Accordingly, WP-FLIP.1 section 9's sentence that the Flip "touches no locked value anywhere" is superseded: it still touches no locked value — the proton row, the G row, the clock and the cosmology stand exactly as they were — but it shares a locked paper's construction and inherits that paper's empirical switch, and says so.

Grade: DERIVED, wearing the support clause, the temporal-expression argument, the KS-30.2 inheritance, and the shared transfer claim of section 4. The solid-angle normalisation inside the lemma is taken on the lemma's word at both sites; the uniformity depends only on it being identical at both, which it is by construction.

The elimination, shown as the record of a killing. After the structure above was described — the words first, the pair before any arithmetic — the alternatives were run against the number. Required divisor: 25.1499. Window at the audited bar (7.4×10⁻⁷ mₑ, 0.29 ppm): [22.096, 29.183].

The constructed assembly forces the support into [1.76, 2.32]: only 2 survives. Metrology also holds a dated tiebreak it does not currently need: roughly a threefold improvement in the bar separates 8π from 9π at ~2.4σ. Data eliminated candidates here; data generated nothing. The 2 was described before it was counted, and the record shows the order.