Two rules for earlier period and epoch pairs gave different three-sigma flags for 7 of 136 TESS planets under zero covariance

We test how source selection changes transit-time flags in a fixed sample of 136 planets. We keep the measured TESS times from an earlier SCI study. Each eligible archive row must supply a period, reference epoch, and their errors. Its source publication month must end before the tested sector starts. One rule selects the latest paper. The other selects the smallest bound on prediction standard deviation over all period-epoch correlations with the stated marginal errors. Neither rule ranks timing residuals. The rules select different rows for 34 planets. They flag 13 and 14 planets above three combined standard deviations under zero covariance. Ten flags are shared, 17 occur under either rule, and seven differ. With the maximum prediction standard deviation allowed by the stated marginal errors, the rules share nine flags and differ on six. Both rules reduce the median absolute offset from the composite value of 6.24 minutes to about 0.67 minutes. We reproduce all 136 periods, epochs, and flags from the prior latest-rule SCI study. We separately fit 65 complete transits in 12 raw TESS sectors for its three strongest original cases. Our sector times agree with that study's separate pipeline within 0.182 minutes after cycle alignment. The package contains raw data, exact query responses, source records, candidate scores, code, results, and figures. This exploratory test uses publication dates in a current archive snapshot. Most measured times are inherited, the pipelines use the same observations, and missing covariance and source errors remain limits. The flag changes do not establish a new planet or orbital change.

TESS · transit timing · ephemeris · source selection · NASA Exoplanet Archive · covariance · reproducibility · hot Jupiters