Three spline node sets give KELT-19 A b width changes from -2.25 to +0.46 minutes in TESS sectors 7 and 87

We test whether two TESS visits resolve a change in the transit brightness profile of KELT-19 A b. Three shared symmetric spline fits give width changes of -2.252, -0.064 and +0.464 minutes for sector 87 minus sector 7. Their residual root mean squares are 853.5, 848.9 and 840.1 parts per million. Thus spline resolution changes the estimate by more than the conditional random error scale. We fit four sector 7 events and five sector 87 events, with mean times 5.947 years apart. The model has exact exposure integrals, a relative sector time stretch, and free event centers, depths and local baselines. Width is the full width at half of the fitted central light loss, before exposure averaging. For the primary node set, the central 95% of 30-minute residual block draws is -1.496 to +1.613 minutes. Ten- and sixty-minute block ranges also include zero. These ranges assume the fitted profile and measured residual process. Synthetic three-minute changes are recovered for both the fitted input profile and a different profile with cosine edges. An earlier SCI study used identical primary FITS files and also found no clear KELT-19 change. This contribution adds a fitted shared-profile estimator, residual block refits and explicit profile-resolution checks. The results are not independent evidence. They do not set a general physical width bound or measure orbital precession. The package preserves raw data, source hashes, code, all statistical draws, figures and reproduction instructions.

KELT-19 A b · TESS · transit width · transit profile · cubic spline · residual blocks · exposure integration · reproducible astronomy