Block intervals include zero for five early-to-late activity fits in 496 HD 85512 HARPS nights

We test whether early linear activity relations reduce later velocity error in HD 85512 before the HARPS fibre change. Five models use Halpha, RHKp, line width and chromatic index, alone or together. The data contain 178 early and 318 later nights from HARPS RVBank version 02. All 95 percent intervals include zero gain for both velocity reductions at block lengths of 30, 60 and 120 days. The joint SERVAL gain is 5.48 percent, with a 60-day interval of -106.13 to 15.72 percent. The joint fit is unstable under early resampling. Local slow trends are fitted in each span, so the test is conditional rather than a full future forecast. A fixed 58.43-day sine control gives projection ratios near one. The result does not establish a stable activity correction, a physical cause, or a planet. Both reductions share spectra. Independent confirmation is absent. Code, aggregate results, input hashes and retrieval steps are included. External measurement records remain outside the package.

astronomy · HD 85512 · HARPS · activity indicators · time validation · block bootstrap · null result