# n246 — Violent explosions as glass probes: the reading list, and a new bound from existing chirps

**Date:** 2026-07-12 · **Question (the human's):** measurements on violent explosions — what to read. **Answer: a reading table per messenger, and one genuinely new constraint extracted from data already in hand.**

## The new bound: distant clean chirps squeeze the glass

The disorder birefringence random-walks the GW phase: Δφ ~ (δc/c)·√(L·ℓ_d)·ω/c. At the model's per-cell contrast (2%, n222 — declared as model number): a Gpc-distance merger **decoheres (1.6 rad) if ℓ_d = a₀** — but LIGO's Gpc-class events (GW190521 family) show *clean, template-matched chirps*. Therefore:

**ℓ_d < ~5 pm at the model contrast — the vacuum's disorder correlation length is bounded an order of magnitude *below* the Bohr radius** (joint form: (δc/c)²·ℓ_d < 2×10⁻¹⁵ m). This tightens n237's mm-scale GW-coherence cap by nine orders, using nothing but the loudest explosions already observed. Fed to the quench triangle: the KZ freeze-out correlation is sub-ångström — a real constraint on τ_Q once units price the chain. (No contradiction with the a₀ *monitor* cell — ℓ_c and ℓ_d are different lengths with different owners.)

## The reading table (what each messenger of a violent transient reads, and what kills)

| Messenger / observable | What to read | Theory stance | Kill |
|---|---|---|---|
| **GW waveform match vs distance** | template-match degradation ∝ √L | the glass signature — the one *positive* curve to hunt | degradation found with wrong scaling; or none found ever → ℓ_d keeps shrinking (bound, not kill) |
| **GW polarization of loud distant events** | +/× consistency per event | equal on average; scatter self-averages | coherent polarization-dependent speed = kills the statistical mechanism |
| **Multimessenger lag (GW/EM/ν) vs distance** | the *slope* of lag vs L | slope = 0 exactly (one cone); GW170817's 1.7 s must stay emission-side | any lag growing ∝ distance |
| **d_L(GW)/d_L(EM) per event** | the ratio vs distance | no trend (conversion capped) | a distance trend = conversion seen; without matching GW-side loss = kills the conversion reading |
| **GRB time-of-flight vs photon energy** | lag structure in E | *linear-in-E lag forbidden*; quadratic allowed, unobservably small at the card's ℓ_s (10⁻⁵ s at Gpc/GeV) | any confirmed linear or superluminal dispersion (prediction 9's kill, unchanged) |
| **Galactic SN neutrinos vs photons** | arrival consistency (SN1987A-class) | dressed matter and light share the cone — consistency at propagation level | a reproducible matter/light cone split |
| **Spectral features at ħγ ≈ 3.7 keV** | the temptation | **refused** — the knee acts on superposition coherences, not classical spectra; amplitude unarmed (n181); F10 hygiene | — |

## The honest summary

Violent explosions are the theory's best free instruments because they integrate the glass's per-cell physics over the longest available baselines: the birefringence, the conversion, and the cone-unification all accumulate as √L or L while every competing systematic stays local. The single most valuable future measurement in this table: **waveform-match statistics vs distance across the growing BBH catalog** — the one place the glass's signature would appear as a *trend* rather than an absence, and the bound tightens for free with every loud detection.

**Class: the ℓ_d bound [B — model contrast declared, data qualitative (clean chirps); the arithmetic exact]; the table [stances from the standing registry]; the knee-spectral refusal [F10].**