Fixed-period fits give a smaller 39.63-day signal after zero-point correction in 115 Gliese 48 velocities

We tested the reported 39.6299-day signal of Gliese 48 in public radial-velocity tables. We fit free sine and cosine terms, an offset, a linear slope, a 51.5-day rotation term, and Gaussian white scatter. On the same 115 CARMENES DR1 observations, the fitted semi-amplitude is 2.90 m/s before the nightly zero-point correction and 0.59 m/s after it. Subtraction of the published common nightly correction gives 0.46 m/s. Under identical weights, the archive correction and the common correction project onto the target period with amplitudes 2.50 and 2.57 m/s. For corrected velocities, addition of the target raises BIC by 4.53. A fixed-period null simulation gives a tail probability of 0.096. The nominal 95% amplitude profile contour reaches 1.11 m/s under this model, with no guaranteed coverage for more general noise. A correlated-noise check also favors omission of the target. The 65 released HIRES velocities give a fitted amplitude of 0.24 m/s. This work shows that the target-period signal in the public CARMENES release depends on calibration. The released tables differ from the larger discovery extraction and contain no near-infrared velocity series. The result does not establish whether the planet exists. The package preserves data, source hashes, code, fit results, simulations, and figures.

Gliese 48 · Ross 318 · GJ 48 · CARMENES · HIRES · radial velocity · nightly zero point · calibration · reproducibility · planet validation