# n62 — pre-registration: the interferometry squeeze (before any code)

**The confrontation.** Neutron interferometry (COW 1975 and descendants)
measures the gravitational phase m Phi t / hbar of the LIGHTEST matter
ever phase-tested, at the ~1% level. In this model the clock/phase
coupling to the potential is exact only in the strong-monitoring regime;
below it, the closed sector contaminates with the measured 1/(domination
ratio) law (n55c: s = -0.18, +0.177, +0.215 at ratios 2.6, 10.3, 20.6).
Consumption scales with mass (EP), so a neutron sits at the smallest
gamma*F ever probed. If the regime curve is universal, existing
interferometry data FORCES the audit rate up — and collides with n59's
Lambda-knee link, which wants it cosmologically slow. One of them dies.

## Registered measurement (toy, today)
The REGIME CURVE: s_op(R), R = gamma*C''/|E0''|, measured at
R = 2.6 ... ~200 by sweeping gamma at F = 0.2 with the full three-term
dynamics (n55c machinery). Gates:
- C1: |s_op/0.25 - 1| fits A/R with a single constant over the sweep
  (fit fracRMS < 10%); report A.
- C2: the COW threshold: R* = A/0.01 — the domination ratio required
  for 1%-accurate phase coupling. Report R*.
- C3 (the squeeze, stated as structure — units honest): since R scales
  as gamma * m, the lightest phase-tested matter binds gamma from
  BELOW: gamma >= R*(toy) x (closed-sector scale)/m_neutron. Any such
  bound pushes the decoherence knee far above cosmological rates:
  REGISTERED CONSEQUENCE — the naive Lambda-knee identification (n59)
  and the interferometry regime condition are mutually exclusive.
  The fork, declared now: either (i) n59 dies in its naive form (the
  knee is fast, the lab band may be reachable after all, Lambda needs
  another origin within the model), or (ii) maintenance pricing is
  nonlinear in F at small F (a new mechanism, needing its own
  registration), or (iii) the model is dead between two existing
  datasets. n62 cannot pick between (i)-(iii) — it measures the curve
  that makes the fork sharp.

## Honesty clauses
- The toy supplies the SHAPE (the 1/R law and R*), not physical units;
  the squeeze is a structural incompatibility, quantitative only up to
  the units mapping.
- E0''/C'' also depends on the trap geometry; A is reported for the
  standard n55 trap and the sensitivity to trap width is reported
  alongside (w = 6 vs 9).

## RESULTS (same day; runs in the session record)
- C1 as registered: FAIL at 13.8% — pulled by the deep-contamination
  points (R = 2.6, 5.1) where the asymptotic law has no business
  holding. Over the asymptotic range (R >= 10) the law is clean:
  dev = A/R with A = 2.810, and — stronger than anything gated —
  A IS TRAP-GEOMETRY INDEPENDENT TO FOUR DIGITS (2.810 at w = 6,
  2.811 at w = 9). The contamination constant is universal.
- C2: the COW threshold R* = A/0.01 = 281: one-percent-accurate
  gravitational phase coupling requires a domination ratio of ~281.
- C3, the squeeze, now sharp: R scales as gamma x mass, and the
  neutron is the lightest matter whose gravitational phase is verified
  at the percent level. The audit must therefore run ~281x above the
  closed-sector scale of the LIGHTEST tested matter — pushing
  gamma_drain far up. CONSEQUENCE RECORDED: the naive Lambda-knee
  identification (n59) is untenable alongside the interferometry
  regime condition. The fork resolves most naturally to branch (i):
  n59-naive dies, the audit is FAST, and the laboratory decoherence
  knee (prediction 1) moves TOWARD accessible bands rather than away —
  existing quantum interferometry, read through this model, is
  evidence that the knee is findable. Lambda then needs another origin
  within the model. Branches (ii) nonlinear pricing and (iii) death
  between two datasets remain open at the units level.