Block tests leave two fixed GJ 887 velocity predictions unresolved in 75 HIRES nights

We compare two fixed GJ 887 velocity predictions in 75 independent HIRES nights. A joint fit to 282 HARPS nights supplies waves at the published 50.77-day planet period and the 38.7-day rotation period. Their HIRES weighted error gains are +6.19 and -4.81 percent. The difference is 11.00 percentage points, but a conditional 30-day block range is -14.63 to +34.39 points. Block alignment tails range from 0.0466 to 0.0880. Small period changes affect the gains. Conditional residual injections recover the planet-period mean in 66 to 75 percent of trials at a threshold set from the rotation-period mean. The fixed rotation wave is not a full activity model. The result does not confirm or reject the published planet or establish a new signal. This follows an earlier campaign transfer control and adds uncertainty and comparison tests. Code, aggregate evidence, source hashes, and external-scratch retrieval steps are included. External catalog records are not bundled.

astronomy · GJ 887 · HIRES · HARPS · independent observations · stellar activity · null result · block uncertainty