Residual Gravity from the Time-Dependent Virial Theorem - P2

Gravity has memory. When matter moves, spacetime curvature does not adjust instantly. It relaxes toward equilibrium over a finite timescale — and that delay is what we have been interpreting as invisible mass. This two-part work derives the effect from first principles and then subjects it to strict blind tests across independent surveys. Part II evaluates the closed predictive chain — visible matter → predicted clock → predicted residual — without ever fitting target residuals, halo masses, or lensing amplitudes to the data being evaluated. Blind audits across SPARC, THINGS, KiDS, GAMA, WALLABY, ultra-diffuse galaxies, and merging clusters confirm: A universal dynamical floor near 36.6 Myr: across 277 valid galaxy–criterion readings, no well-sampled transition-region estimate falls below this reference scale. Baryons predict the clock: visible matter recovers the observed relaxation timescale with Spearman ρ = 0.939, while a randomized null collapses near ρ ≈ 0.1. Fixed channel separation: dynamics responds with amplitude A_dyn = 1; weak lensing independently confirms the geometric prediction A_lens = 19/15 ≈ 1.267 — measured to within 0.02σ in the KiDS sample — reducing the residual χ² by a factor of 28.2 relative to baryons alone. Consistent signal across regimes: morphological type, HI kinematics, dark-matter-free galaxies, and cluster mergers all behave as expected: relaxed systems converge toward the reference scale; disturbed systems return longer relaxation times. No new particles. No fitted halos. One geometric delay, two observational channels, and a predictive chain that holds across four independent surveys and seven distinct audit regimes. The apparent dark matter component emerges as a finite relaxation memory of the gravitational field.