Linear run fits give a 0.36 m/s channel difference at 4.1244 days in 112 Barnard's Star velocity pairs

We test the blue-minus-red velocity in all 112 published MAROON-X pairs for Barnard's Star. An identical shared velocity cancels from each pair. Separate straight lines for nine observing runs give a weighted residual sum of 259.938 for 94 degrees of freedom. This exceeds the independent quoted-error model. Four fixed sine and cosine pairs test the published planet periods. The 4.1244 day difference has a fitted amplitude of 0.360 m/s. Its conditional F reference probability is 0.002926, or 0.01170 after the Holm adjustment for four periods. This reference assumes independent Gaussian epoch errors with one fitted variance scale. Removal of the 2022 Apr run gives 0.309 m/s and an adjusted probability of 0.01018. An exact run sign test gives 4 of 256 patterns at or above the observed largest projection at the four periods. That test permits dependence within each run under independent, centrally symmetric run errors. Earlier periods from separate ESPRESSO data give a 0.362 m/s difference and the same sign count. The result has material limits. Constant run baselines give an adjusted F probability of 0.07481. Removal of the three largest standardized baseline residuals gives 0.209 m/s and 0.09981. The periods and models were tested retrospectively. The evidence does not establish a new planet, a physical period, or the cause of the difference. The contribution is a paired control dataset and a reproducible fit with its sensitivity tests. All source values match independent PDF and HTML parses. Code, source provenance, covariance matrices, results, and figures are preserved.

Barnard's Star · MAROON-X · radial velocity · channel difference · paired control · run sign test · exoplanets · measurement error · reproducible analysis