The BIC gain of a 62-day GJ 776 signal changes across tested HARPS models

We test whether the known-planet period branch changes the evidence for a selected 62.21-day radial-velocity candidate of GJ 776 (HD 189567). We use all 252 HARPS nightly means from the prior SCI search and 30 public ESPRESSO observations from its later validation. Changing the second planet from its 37-day annual alias to its published 33.6-day branch does not remove the third-orbit fit gain. On the published branch, with eccentricity limited to 0.95, the BIC change for a third orbit is -29.0 with white scatter. It is +0.2 after we add a cubic time trend and a defined filtered line-width term. Negative values favor the third orbit. The later ESPRESSO data do not favor the two fixed candidate curves from these fits when the known planets have free phases. A free 62-day sinusoid improves the base ESPRESSO fit, but loses that gain with line width. We also reproduce the prior validation scales and show that its base and line-width clipping variants remove different observations. The nuisance model also changes the result. Unger et al. (2021) already reported the 61.9-day line-width signal and its possible activity relation. Our contribution is a reproducible model and selection audit of the later SCI candidate. Local optimization, high eccentricity, short ESPRESSO coverage, and conditional errors limit these tests. They do not prove a stellar cause or exclude all planets near 62 days.

HD 189567 · GJ 776 · radial velocity · HARPS-RVBank · ESPRESSO · stellar activity · model comparison · annual alias · clipping · reproducibility