# n222 — The glass speed test: disorder makes the two gravitational polarizations co-propagate

**Date:** 2026-07-11 · **Follows n221.** Modulated-pattern speeds (the real q²-slope quantities), both polarizations, distorted diamond, three seeds per point, phase- and direction-averaged.

## Result

| disorder η | c²₊ | c²₊ | speed-squared ratio (3 seeds) |
|---|---|---|---|
| 0.00 | *does not exist* | 1.756 | — (the crystal has one polarization) |
| 0.25 | 2.034 | 1.823 | 1.116 ± 0.010 |
| 0.50 | 1.988 | 1.950 | 1.020 ± 0.030 |
| 0.75 | 2.015 | 2.002 | **1.007 ± 0.021** |

**The two polarization speeds converge to equality as the substrate amorphizes** — at strong disorder they are equal within the seed spread (c₊/c₊ = 1.003 ± 0.011 in speed). The Zener obstruction, fatal for every crystal (n220), dissolves *dynamically* in the glass: not only the base-point quantities (n221) but the propagation speeds themselves equalize. What no symmetry could supply, statistics does — and the statistics were mandated by the theory's own Kibble–Zurek origin.

## The new falsifiable edge, now quantitative in shape

At finite disorder the equality is statistical: the residual polarization-speed spread (here ±2% at 216 sites) is a **gravitational-wave birefringence** that self-averages away with propagation volume. A real GW traversing ~10⁴⁴ coherence cells would see the spread suppressed by the corresponding root — the prediction's *shape*: birefringence ∝ (disorder amplitude)·(cell/wavelength)^{3/2}-class, with LIGO's polarization consistency capping the disorder. The number needs the units bridge; the shape is armable.

## Honest scope and the remaining program

216 sites; q = 2π/3 (large — finite-size speeds, but both channels at the *same* q, so the ratio is meaningful); three seeds; jittered crystal rather than true glass; **the EM channel's speed on the glass not yet computed** — that plus the re-pinning (c_g = c_EM on the amorphous substrate) and a frozen re-run of the n220 stability scan at scale (Rust, true amorphous graphs, many seeds) constitute the P3″ gate. Only after that does the proposal go back to the reviewer.

**Class: [B — multi-seed numerics, scope declared]; the convergence trend [robust across seeds]; the birefringence shape [conditional prediction, units open]. The night ends with the two deaths intact, and a direction that passed its second test in a row.**