Two of 3651 matched Kepler candidates keep a duration change above a factor of 1.5 across three-error boxes

We test duration stability between the Kepler DR24 and DR25 catalogs. We select 3651 matched signals with pipeline candidate status in both releases, the same host ID, and periods that agree to 0.1 percent. At nominal values, 39 duration pairs differ by more than a factor of 1.5. Positive three-error boxes exist for 3627 signals. Of these, 67 boxes are disjoint and two retain a factor above 1.5: K01064.01 and K00456.02. The first is a known eclipsing binary; the second is known Kepler-160 b with DR25 score 1.000. A prior campaign report already flags both. Our added result is the error-box test. Only one signal retains the large difference at five errors. Missing and broad errors prevent strong claims for several larger nominal differences. These are fit-stability checks, not measured physical changes or confidence tests. Both releases use overlapping Kepler data and can inherit fits. We do not fit flux, test pixels, or obtain independent observations. We report no new planet. The package supplies a full quality-control table, frozen raw inputs, provenance, code, error sensitivity tests, a figure and offline reproduction instructions.

Kepler · transit duration · catalog stability · DR24 · DR25 · null discovery result