Kepler DR25 supplemental fits give median radius-ratio errors of +4.1% at input b<0.5 and −10.2% at b≥0.8 in selected recoveries with measured MES 7.1–10

This study tests radius-ratio errors in weak Kepler transits. It uses the public DR25 INJ1 pixel injection table. A local plan was recorded before the first summaries. The primary sample has 32,221 detected, on-target signals with supplemental fits. It requires period agreement within 0.1%, source N_Transit at least 4, and positive radius ratios. Input periods are 0.5–500 days. Input durations are 1.5–15 hours. Input impact parameter b is from 0 to 1. Error is the fitted radius ratio divided by its input value, minus one. At measured multiple event statistic (MES) from 7.1 to below 10, the 15,692 selected signals have a median error of +0.96%. The median is +4.08% for input b<0.5 and −10.16% for b≥0.8. Errors below −20% occur in 10/8,399 signals (0.12%) in the first group and 361/2,070 (17.44%) in the second group. Input b measures transit distance from the star center in stellar radii. Of 933 selected recoveries with expected MES below 7.1, 234 have errors above +20%. None have errors below −20%. These results depend on detection and sample selection. They do not establish a new pipeline fault or a general correction for planet sizes. The main provenance cut excludes the known original fit fault. The package preserves source bytes, all input rows, an error map, code, figures, audits, and repeat-run instructions. The contribution tests conditional errors in supplemental point estimates. It does not test KOI posterior samples or report a new planet.

Kepler DR25 · transit injection · planet radius · impact parameter · selection effects · reproducible research