Published BJD-TDB ephemerides reduced the median absolute timing offset from 6.63 to 0.62 minutes for 129 TESS planets

Can complete published ephemerides remove large timing offsets in a composite-table test? We reuse the measured times from a prior SCI study of 153 TESS planets. We select 129 planets with a complete period and reference-time pair in one NASA Exoplanet Archive PS row. The row must state BJD-TDB and have errors for both values. Its publication month must end before the tested sector starts. We select the latest eligible month without use of the measured offset. For the same 129 measured times, the median absolute offset falls from 6.63 to 0.62 minutes. The number of offsets above three stated standard deviations falls from 29 to 10. The absolute offset falls for 92 planets. We also fit all 12 public SPOC 120-second light curves for the three largest original flags: HAT-P-16 b, KELT-18 b and WASP-20 b. For the main fit, their latest tested sector times agree with those pairs within 1.5 combined standard deviations when covariance is zero. For HAT-P-16 b and WASP-20 b, the old and selected periods are equal. The complete reference pair alone changes the prediction by 13.5 and 13.7 minutes. These offsets do not require a period change. The main sample reuses prior measurements and omits period-time covariance. Its ten remaining flags need separate checks. We do not report a new planet or exclude real timing changes. The package preserves the code, source data, selection records, new time measurements, controls, figures, and reproduction instructions.

TESS · transit timing · ephemeris · hot Jupiters · NASA Exoplanet Archive · data provenance · reproducibility