A fixed activity covariance gives a nominal 95% bound of 1.18 m/s on the 23.064-day HARPS–ESPRESSO signal difference in 167 L 98-59 velocities

We compare the reported 23.064-day signal of L 98-59 f between HARPS and ESPRESSO. We use the public Table B1 velocities from Cadieux et al. (2025). Both distributed tables retain six epochs that the article lists as excluded. We remove only these epochs and preserve an explicit 167-row sample. We fit circular signals for b–f, three instrument offsets, and a fixed published activity covariance. The common signal amplitude is 1.441 m/s. The HARPS-minus-ESPRESSO sine/cosine pair has a norm of 0.438 m/s. Its difference score is 2.255 for two coefficients, with a nominal conditional tail of 0.324. The joint nominal 95% difference ellipse includes zero. Its largest norm is 1.17321 m/s, which gives a conservative bound of 1.18 m/s. All 23 tested cases with correlated noise have nominal difference tails above 0.05. After the noise amplitudes are fitted again, 1000 null draws per noise model give tails of 0.095 for independent errors and 0.289 for the activity covariance. We preserve the original tables, source hashes, code, full coefficient covariances, residuals, figures, and null simulations. The period and covariance came from these same data. The bound holds them fixed and does not include their uncertainty or selection. This test does not establish equal signals, a planetary cause, or independent confirmation.

L 98-59 · radial velocity · instrument comparison · Gaussian process · data audit · conditional confidence bound · HARPS · ESPRESSO