# n228 — Entry 2 recheck: the knee-anchor formula loses its derivation route; the hypothesis stands naked

**Date:** 2026-07-11 · **Resolves n226 AT-RISK #2.** Sequential computation, five family points × two seeds.

## The finding

γ = α·mc²/ħ was derived through the one-monitor identity Q = α/α_G — the *single* monitor's overdamping equated to the EM/gravity strength ratio. Under P3″ that route is gone by construction, and the search for a two-ledger replacement came back empty: along the equal-cone family, λ_e(λ_g) is affine with **no invariant** — the ratio drifts 0.28 → 0.91, the difference drifts systematically (std ≈ 35% of range), and the affine offset is seed-dependent (−0.76 vs −1.08). No natural cross-sector constant exists at leading order. (A candidate difference-invariance from the two n224 points was tested and dissolved at five points — recorded so nobody rediscovers it.)

## Verdict

**RESCOPE downward, honestly:** γ = α·mc²/ħ is now a **pure registered hypothesis** — the fine-structure form is a bet, no longer a supported arithmetic. What survives intact: the window's *span* (m_p/m_e between the two anchor bets), all shape predictions (zero-at-rest, ω² onset, the ~6% multi-pole knee), and the falsification logic (a measured knee anywhere is the discovery; its location then *selects* the two-ledger normalization rather than confirming a derivation). The future route to re-deriving the location runs through Gate D leg 2 (the normalizations tie the sectors); until then the 3.7 keV number is a naked wager, labeled as such.

Entry 2 and the parameters γ-row updated. AT-RISK queue: 7 remaining.

**Class: [A-model, five-point family scan]; the demotion [rule 5 — old claim: derived-through-Q; failure: the one-ledger route died with the ledger; replacement: hypothesis + Gate D leg 2 route].**