60 new nights with independent 1 m/s errors cannot raise five HARPS annual information ratios above 0.1 on fixed 30° La Silla grids at e=0.8

We test 60 future nights for eight public HARPS series with low annual eccentric orbit information. The period is 365.25 days and the eccentricity is 0.8. The score is the least amplitude information over all periastron phases and orientations, divided by ideal information at the same total weight. The data have 1,186 historical nightly records from 4,335 exposures. We test future nights from November 2026 through October 2030 at La Silla. Each future night has a total independent error of 1 m/s. The model has old era offsets, a free future offset, and a common trend. Each 30-degree grid has one fixed candidate time per eligible night. Five series have upper bounds below 0.1 for every allowed 60-night allocation: GJ 221, GJ 691, GJ 9255, HD 175607, and HD 47186. Selected schedules for GJ 163, GJ 191, and GJ 3341 have lower bounds above 0.1. Each selected schedule gives at least 1.81 times the score of uniform dates. No allowed allocation can improve any selected score by more than 2.21%. We reconstruct upper bounds without a solver and bound phase minima with exact derivative inequalities. The package includes data, code, dates, figures, and replay instructions. A separate 20-degree test holds measurement errors fixed. It shows a large effect from added seasonal coverage. The annual schedules score below uniform dates for six series at 350 days and seven at 380 days. These are conditional information limits. They are not planet detections or detection probabilities.

radial velocity · observation planning · annual orbits · eccentric orbits · HARPS · information bounds