# n250 — The maximum photon frequency: the ceiling exists, and it is frosted

**Date:** 2026-07-12 · **Question (the human's): the maximum frequency of a photon.**

## The answer in three layers

**1. A ceiling exists, structurally — a hard prediction.** The photon is a wave of the substrate (the cycle-current modes, n249), and no glass carries waves above its band edge: **ω_max ~ c/ℓ_s.** Continuum QED has no such ceiling; this theory *cannot avoid one*. That is a genuine structural discriminator, independent of every crisis.

**2. The value is unpriced but bounded from below by the theory's own card.** The GRB-dispersion bound (ℓ_s < 4.3×10⁻²⁷ m) forces **E_max ≥ 4.6×10¹⁹ eV (~46 EeV)**; if ℓ_s were Planckian, E_max ~ 10²⁸ eV. The most energetic photon ever observed (LHAASO, ~2.5 PeV) sits at least four orders below the *minimum* ceiling — every photon in nature is comfortably legal. (Noted and refused under F10: the 46 EeV floor lands in the UHECR/GZK decade — a numeric adjacency with no mechanism; the bound is one-sided and the coincidence is not claimed.)

**3. The approach to the ceiling is frosted, not sharp.** Two structural effects stack as ω → ω_max: (i) the dispersion is quadratic-subluminal (prediction 9's form) — the group velocity falls toward zero at the edge: *the last photon does not fly; it stands*; (ii) computed tonight on the true glass: band-top modes **localize** (participation ratio 6.9 vs 83 at the band bottom) — near-ceiling photons are Anderson-trapped by the vacuum's own disorder before the edge is reached. **The vacuum's ultraviolet is frosted glass**: the effective transparency ceiling sits *below* ω_max, and the highest-frequency light in this universe ends its life standing still in a cell of the vacuum.

## What would test it

Unreachably high directly — but the *shape* of the approach is prediction 9's card: energy-dependent arrival delays growing quadratically, with the theory's specific addendum that transparency itself degrades before dispersion diverges (a spectral cutoff softer than the naive band edge in any source spectrum reaching high enough). Kill: unchanged from prediction 9 (any confirmed linear or superluminal dispersion); plus the structural one — a demonstrated photon *above* a future two-sided determination of c/ℓ_s.

**Class: the ceiling's existence [structural, A]; the floor value [arithmetic on the card's bound]; the localization [A-model, computed]; the GZK adjacency [refused, F10].**