TESS reference fits put DS Tuc A transits nearer B in 13 of 235 eligible tests

We test transit reference positions in public TESS images of the unresolved DS Tuc binary. Resolved observations identify A as the planet host. We reuse exact pixels, 13 accepted transit intervals and the model grid from a signed SCI source study. Five of seven sectors have more than one reference transit. We remove each of their 11 transit images from the reference and locate it with the other transits. All 11 main-model comparisons put the removed transit nearer A. Across 23 distinct transit settings, 235 of 253 comparisons pass fixed rules for positive amplitude, optimizer success and distance from search bounds. Thirteen eligible comparisons put the known A transit nearer B. They involve three target images and seven model settings. Eighteen comparisons have a negative transit signal. The main offsets across the binary reach 4.00 arcsec although all main nearest-star labels agree with A. We also test each of eight inherited flare images against every separate transit reference for 24 flare settings. The two previously stable B candidates remain nearer B in all 66 and 48 eligible comparisons. Their distances from B range from 0.57 to 3.40 arcsec and from 0.92 to 12.01 arcsec. A stable nearest-star label does not ensure a stable position near that star. These tests share pixels and references and do not give independent trial probabilities. Transit conditions cannot test unknown flare color or PRF errors. The result is a reference audit, not a physical flare source assignment. Exact inputs, all fit checkpoints, code, provenance, figures and offline reproduction steps are included.

DS Tuc · TESS · transit reference · difference image · source position · stellar flares · model sensitivity · reproducible astronomy