We present a reproducible kinematic relaxation discriminator constructed from age-stratified stellar velocity dispersion in galactic disks. The method defines a dimensionless regime metric D (R) that quantifies departures from quasi-steady kinematic behaviour using publicly accessible Gaia DR3 data and an explicitly documented ADQL query. Applied to a restricted Milky Way sample of 2292 disk stars across five radial bins (7. 25–9. 25 kpc), the discriminator yields D (R) values of order 10⁻². Given per-bin statistical uncertainties and power thresholds, this constitutes a non-detection of non–quasi-steady behaviour above D ≈ 0. 03–0. 09 at 95 percent power, depending on radial bin. The best-constrained bin (R ≈ 8. 75 kpc, N⋆ = 772) yields D = 0. 023 ± 0. 009 with Dₘin, power = 0. 031. Smaller deviations cannot be excluded at current resolution. A phenomenological scaling ansatz is evaluated against independent SPARC rotation-curve data as a gradient-sign consistency test. For NGC 3198, the full radial sample shows significant rank monotonicity (Spearman ρ = 0. 70, p < 10⁻⁴), while the quasi-steady-restricted outer regression is not statistically significant at present resolution and is therefore reported as a null result. This work does not propose a modification of gravity. It provides a documented kinematic pipeline with fixed parameter governance, quantified detection floors, and explicit falsifiability criteria. All intermediate tables, data products, and regression summaries are included to enable independent verification.
Villarroel et al. (2026) studied this question.