Kepler third-body timing orbits predict TESS 2019-2024 eclipse times for 11 of 13 testable detached binaries but for only 5 of 25 contact and close binaries

Borkovits et al. (2016) fitted light-travel-time third-body orbits to 4 years of Kepler eclipse times. We tested these orbits with TESS full-frame images from 2019 to 2024, 6 to 12 years later. We measured eclipse times in Kepler and TESS with one template method, fitted each published orbit to our Kepler times only, and extrapolated it. Our Kepler times reproduce 117 of 160 published light-travel-time solutions. For 38 of them, TESS can separate the published orbit from no third body. The forecast passes for 11 of 13 detached (EA) binaries (median TESS residual 42 s, skill 0.96) and fails for none. It passes for only 5 of 25 contact and close (EW, ELV, EB) binaries and fails for 15 (Fisher p = 0.0003). In all 105 testable systems, TESS departs from the published orbit by more than 3 times the forecast uncertainty for 38 of 65 close binaries and 2 of 40 detached binaries. These departures grow with time and have both signs, as expected from period changes or spot effects. Quadratic ephemeris terms make forecasts worse. Gaia DR3 orbital periods agree with the published outer period for 7 of 8 systems. Limitations: 67 forecasts are too uncertain for the TESS FFI precision, TESS formal timing errors needed a factor of about 3 calibration, and dynamical solutions were tested only with a model-free continuation.

eclipse timing variations · hierarchical triple stars · light-travel-time effect · Kepler · TESS · contact binaries · forecast verification · Gaia DR3