The measured midpoint of a DI Her eclipse in TESS sector 81 changes with depth

Eclipse times can change when the eclipse profile changes. We measure profile times at three fractional depths for DI Herculis, TIC 358517121. The inputs are three public TESS two-minute light curves from sectors 26, 80, and 81. Six complete events pass the stated coverage rules. Four events are from 2024 and two are from 2020. For the sector 81 primary eclipse near BJD_TDB 2460527.34, the 75% midpoint follows the 25% midpoint by 191.8 seconds. A block bootstrap gives a short-scale noise standard deviation of 1.5 seconds. The difference stays positive in all 96 polynomial reduction choices, from 176.2 to 274.7 seconds. Separate reference-depth and sinusoidal baseline tests also give positive differences. Synthetic controls use the actual observing times and quality flags. Their largest difference error is 5.2 seconds without added noise. The measured profile does not give one midpoint that is independent of depth. Earlier work already found variable eclipse structure in DI Her. This contribution adds the specified 2024 profiles, three-depth times, a complete event audit, and reproducible method tests. The analysis does not identify the physical cause or measure an orbital delay. The package contains the original data, input hashes, code, joint bootstrap samples, results, and figures.

DI Herculis · TESS · eclipsing binaries · eclipse timing · eclipse shape · public data · reproducible research