For 110 selected Kepler stars with unflagged timings, an independent common median model bounds shared 372.5-day amplitude at 2.79 seconds with 95 percent coverage

We tested for a shared 372.5-day term in public Kepler transit times. Fixed rules selected one planet at each of 110 stars. The fits used 35,213 accepted transits from Holczer et al. (2016). Each fit included a quadratic slow trend and a fixed annual sine and cosine. The coordinate medians were 0.330 and 0.520 seconds. Exact binomial intervals gave a 2.79-second upper amplitude bound with at least 95 percent joint coverage. This coverage requires independent star coefficients with a common median after selection. Keeping source outliers and imputed errors gave a bound of 18.46 seconds in 112 stars. All fixed joint intervals included zero. The early and late halves gave upper bounds of 5.91 and 5.58 seconds. The package reports every fixed sensitivity check. Independent solver checks, source reconstruction, signal injections, and Gaussian controls test the calculation. The result measures a shared term in selected catalog records. It does not establish a physical clock limit or rule out signals that differ between stars. The package preserves the inputs, source hashes, selection decisions, fitted coefficients, figures, code, and repeat instructions.

Kepler · transit timing · annual systematic errors · robust statistics · common median · reproducible astronomy