TESS eclipse times confirm the timing drifts of most of 27 apsidal-motion circumbinary-planet candidates, but measured radii make tides sufficient for 9 and leave a clear excess in 4

Thornton et al. (2026, MNRAS) report 27 circumbinary-planet candidates. The evidence is apsidal precession of TESS eclipsing binaries above the general-relativity and classical (tidal plus rotational) rates. Their classical rate uses main-sequence radii. We measured 1,871 eclipse times in 256 QLP and 79 SPOC light curves (TESS sectors 1-104) for all 36 binaries in their table and for the known planet host TIC 172900988. We used the time D from the primary to the secondary eclipse per sector. Light-travel-time and period errors do not change D. Before the paper cutoff, 21 of 32 binaries have a D drift above 3 sigma. All 21 agree in sign with the paper, with a median amplitude ratio of 0.99. In the other 11 binaries, the drift is below 3 sigma. Eleven binaries have sectors after the cutoff. Over 20 sector values, chi-square is 18.3 if the drift continues and 82.7 if it stops. Two binaries give most of this result. Prograde apsidal motion requires a fixed relation between the drift sign and the eclipse-width ratio. The relation holds in 14 of 15 testable binaries and in the control. It fails for TIC 293225466. Radius sums from eclipse durations agree with TIC (Gaia) radii. In 25 of 35 binaries, they are larger than the values implied by the paper's classical term (many hosts are subgiants or giants). The classical rate scales as (R/a)^5 and becomes a median of 3.8 times larger. With these radii, GR plus classical precession explains the rate of 7 candidates. In 2 near-circular candidates, it can supply the minimum required rate. Three candidates keep an excess only if the apsidal constant is 0.003. Four candidates (TIC 123951716, 146204045, 292357653, 82893635) keep a 3.5-5 sigma excess. TIC 146204045 also has a light-travel-time signal, so its companion is probably a star. The control keeps a 4.2 sigma excess. Seven candidates have no significant drift in our data. Limitations: e sin w comes from duration ratios with an assumed 0.02 systematic error. The classical model uses equal stars and an assumed apsidal constant. The paper cutoffs were read from its figures. Most post-cutoff baselines are 1-2 years.

eclipsing binaries · apsidal motion · circumbinary planets · eclipse timing · TESS · QLP · tidal precession · forecast test · replication