One shared radial slope gives no clear shape gain in 32 SPARC test galaxies with fixed mass profiles

We test whether one shared radial slope improves relative rotation-curve shapes under fixed SPARC mass profiles and geometry. We remove each curve's mean log speed and use a fixed radial acceleration relation as the baseline. A fixed name-hash split gives 65 fitting galaxies and 32 test galaxies. The fitted slope is -0.01970. On test galaxies, the root mean squared shape error rises from 0.03579 to 0.03672 dex. Each galaxy has equal weight. Twelve of 32 galaxies improve. The mean squared error gain is -6.70e-5 dex squared, with a 95 percent whole-galaxy bootstrap interval of [-1.80e-4, +5.59e-5]. Planned mass, acceleration-scale, and sample checks also give no clear gain. We provide a supplied-profile normalization check, an independent scalar calculation, frozen data, code, figures, and complete prediction tables. The original relation used SPARC. Only the added correction has separate fitting and test galaxies. Unknown mass-profile errors and radial covariance remain outside this conditional test.

SPARC · galaxy rotation curves · radial acceleration relation · relative curve shape · galaxy test split · reproducible benchmark · fixed mass profiles