A 79.22-day fit gives a 0.96 m/s amplitude in 111 paired LHS 1140 velocity differences

We test how a signal near the disputed LHS 1140 d period depends on velocity processing. We match 111 ESPRESSO exposures from the 2020 cross-correlation reduction to the 2024 line by line reduction. The latter data are recovered from the original vector PDF of Figure B1, with a numeric check on seven printed rows. The difference cancels any additive signal that enters both measurements in the same way. On 95 nights, a fit at the imported period of 79.22 days gives a semi-amplitude of 0.957 m/s. Under independent Gaussian night errors, the projected 95% interval is 0.520 to 1.394 m/s. Simple time and line-width terms give similar amplitudes. In a 70 to 120 day scan, the largest gain is at 77.718 days; none of 10,000 Gaussian null scans reaches it under the main covariance. Added exponential time covariance weakens the evidence. With an added 2 m/s noise amplitude and 7 or 30 day time scale, the fixed-period intervals include zero and the scan p estimates are 0.530 and 0.432. These are fixed sensitivity cases, not fitted noise estimates. The new paired data and analysis measure dependence on the full processing chain. They cannot identify the correct reduction or exclude a planet. Cadieux et al. already found no convincing evidence for candidate d. This study uses the same spectra and provides no new independent observations. The package includes source data, recovery checks, provenance, code, fit results, simulations, and figures.

LHS 1140 · radial velocity · ESPRESSO · paired spectra · line by line · cross correlation · temporal covariance · vector PDF data recovery