A Newtonian fit to 2009–2013 data predicts 16 later Kepler-16 eclipses with 43-second RMS error at fixed eccentricities

We fit a Newtonian model to 63 public Kepler eclipse times from 2009 to 2013. It predicts 16 TESS times from 2019 to 2024 with a root mean square (RMS) error of 42.8 seconds. The largest error is 96.0 seconds. Separate straight-line eclipse clocks give an RMS error of 67.9 seconds on the same later events. This test is retrospective. It fixes stellar masses, orbital eccentricities, inclinations, and nodes. A fit to all 79 times has a residual RMS of 20.7 seconds. With fixed eccentricities, the fitted planet mass is 0.2970 Jupiter masses. Its conditional local standard error is 0.0034 Jupiter masses. When the binary eccentricity and planet eccentricity vector can vary, the mass is 0.3060 with an error of 0.0276. The published mass is then within one local standard error. These errors hold fitted scatter and other orbit assumptions fixed. An independent integrator with exact light propagation agrees with the main timing model to less than 0.0041 seconds. The result gives a reproducible prediction check. It also shows that a narrow mass error can depend strongly on fixed eccentricities. It does not establish a new planet or a complete physical orbit. The package has flux extracts, source hashes, input times, code, fitted results, figures, and reproduction instructions.

Kepler-16 · circumbinary planets · eclipse timing · Kepler · TESS · Newtonian dynamics · orbital eccentricity · prediction errors · reproducible research