Seven 2024 BF Draconis eclipse triplets lie 48 seconds below a steady apsidal forecast fitted to 2019–2022 data at e = 0.3865 and i = 88.42 degrees

We test whether steady apsidal motion can explain the later eclipse phase of BF Draconis. We reuse published times from 38 public TESS light curves. Each local phase uses one secondary eclipse and the two primary eclipses that enclose it. The ratio of the two time intervals removes any affine clock map shared by these three times. There are 16 reference triplets in 2019–2022 and seven later triplets in 2024. The mean later phase lies 47.8 seconds below an equal-weight early line forecast in corrected flux. A bootstrap of connected observing groups gives a conditional 95 percent interval of [-60.3, -39.5] seconds. Only three later groups are available. Six timing methods and two flux products give later means from -49.0 to -43.6 seconds. We fix eccentricity at 0.3865 and inclination at 88.42 degrees. An exact conjunction model uses a constant apsidal rate that matches the early slope. It predicts a later departure of +0.0074 second. A finite grid of 1,620 fixed orbit models gives no departure larger than 0.042 second. Adding fitted common quadratic or cubic time delays changes the local phase by less than 0.0025 second. These tests do not explain the measured change if eclipse centers trace conjunctions up to a constant offset. Changes in eclipse shape, other clock errors, and changing orbit elements remain possible. The result does not identify a planet or a unique cause. The package preserves native data columns, provenance, source times, code, controls, figures, and an offline reproduction check.

BF Draconis · TIC 229776076 · TESS · eclipsing binaries · apsidal motion · eclipse timing · local eclipse phase · reproducible astronomy