A Geometric Law for Disk+Gas Galaxies

Two frozen coefficients, calibrated on half the sample, tested blind on the other half

BeeTheory.com · Real data · 21 May 2026

Result first

For bulgeless (disk+gas) galaxies, the plateau velocity can be predicted from two purely geometric reference quantities — the ring velocity V_ring and the extent R_max/R_ring — through a law with two fixed coefficients. Calibrated on half the galaxies and applied blind to the other half, it predicts the plateau to a median error of 17% (44/71 within 20%). The visible ring carries most of the signal; the geometric term refines it.

1. The inputs — all geometric, from the reference model

Each galaxy is reduced to one visible ring (disk+gas at its mass-weighted radius). Three reference quantities follow, none of them fitted to the rotation curve:

SymbolMeaning
M_ringtotal disk+gas mass
R_ringmass-weighted radius of the ring
V_ring√(G·M_ring/R_ring), the ring’s own circular velocity
R_max/R_ringextent: how far the galaxy is seen relative to the ring radius

2. The law and its coefficients

The deficit between the observed plateau and the ring velocity, expressed as a relative excess, follows a simple two-coefficient form fitted on the calibration half:

ΔV / V_ring = -0.101 · ln(R_max/R_ring) + (-0.222) · ln(V_ring) + 1.428

and the predicted plateau is V_plateau = V_ring · (1 + ΔV/V_ring). The two coefficients on the geometric terms are -0.101 (extent) and -0.222 (ring velocity); the constant is 1.428. The negative extent coefficient means more extended galaxies show a smaller relative excess — their diffuse mass is already well captured by the ring.

3. Blind test

The coefficients are then frozen and applied to the other half of the galaxies — never seen during calibration. No per-galaxy tuning.

Blind test predicted vs observed plateau
Predicted vs observed plateau velocity for the 71 blind galaxies. Solid line is equality, dashed lines ±20%. Median error 17%.

The blind prediction lands within 20% for 44 of 71 galaxies and within 30% for 56. The agreement is carried mainly by V_ring: the visible ring alone fixes the order of magnitude of the plateau, and the geometric term adjusts the rest.

What holds, honestly

The strong claim is the blind test: two frozen coefficients, no per-galaxy freedom, predict the plateau of unseen galaxies to ~17%. The honest qualifier: most of that success comes from V_ring (the visible ring), with the geometric extent term adding a modest correction. The residual scatter — the part neither V_ring nor R_max/R_ring captures — is real and reminds us the reduction to a single ring loses some information. This is a calibrated empirical law for disk+gas galaxies, not yet a first-principles derivation.

Method note

All velocities are real SPARC data (Lelli, McGaugh & Schombert 2016). Reference quantities (M_ring, R_ring, V_ring, R_max) come from the fixed per-galaxy table. The 143 disk+gas galaxies were split in two by name: 72 for calibration, 71 held out for the blind test. The coefficients shown were fitted only on the calibration set.

BeeTheory.com — A geometric law for disk+gas galaxies · Data: Lelli, McGaugh & Schombert 2016 · Initial generation: 21 May 2026 with Claude.ai · © Technoplane S.A.S. 2026