The 2023–2024 TESS eclipse times of BF Draconis do not follow linear clocks fitted to 2019–2022 data

We test whether later TESS eclipse times of BF Draconis follow separate linear clocks fitted to sectors 14–53. We measure 133 complete eclipses in 38 standard 120-second SPOC light curves. The reference set contains 74 eclipses from 2019–2022. The test set contains 59 later eclipses. Across this set, the half-sum of the two mean forecast residuals is -21.3 seconds, with a sector-cluster standard error of 4.2 seconds. We also give an exploratory summary for the final observing block. In this block, from December 2023 to December 2024, 19 primary and 19 secondary eclipses have mean forecast residuals of -7.6 and -54.1 seconds. Their half-sum is -30.9 seconds, with a sector-cluster standard error of 3.3 seconds. A bootstrap of complete sectors gives a 95 percent interval of [-37.4, -24.7] seconds. Simple aperture flux gives -32.0 seconds. Independent symmetric profiles give -30.1 seconds in corrected flux. Clocks fitted to all 80 published early times give -32.1 seconds after constant zero-point alignment. Removing reference observing blocks preserves the sign. One 2022 secondary time differs by 84.6 seconds between the flux products; we retain this event. The result extends earlier reports of timing drift. The half-sum is a descriptive statistic and does not isolate light travel time in an eccentric binary. The cause remains open; these data do not establish a planet. The package preserves the source columns, hashes, all candidate fits, uncertainty checks, figures, and code for offline reproduction.

BF Draconis · TIC 229776076 · eclipsing binaries · TESS · eclipse timing · forecast test · public data · reproducible research