A 3.46-hour test transit has a 2.0-hour best grid fit in selected Kepler-88 TESS data

Can selected TESS light curves test a predicted transit duration change for Kepler-88 b? We use public two-minute SPOC products from sectors 40, 41, 54 and 55 as the main sample. Later data from sectors 74, 75, 81 and 82 form an extension. We compare finite transit-template grids, remove sectors, vary fit assumptions and add test signals to real data. In one of two sets of off-transit windows, a known 3.4595-hour signal has a 2.0-hour best fit. The same input has a 3.2-hour best fit in the other set. The injected shape parameter is held at its true value. The fit gain over the true long duration reaches 10.94 in the failed recovery. Removing sector 40 moves the primary coarse-grid minimum from 2.4 hours to the 4.6-hour grid limit. Conditional residual block tests assess the actual duration profile. These tests do not establish a physical duration change or a constant duration. They show a measurement limit for this data sample and this flexible fit family. The injections occur after mission processing and do not test signal loss in that processing. Four real-noise trials do not give a general rate of false duration changes. The package has all 26 mission FITS files, source hashes, the prior prediction table, code, fit grids, simulation draws and figures.

Kepler-88 · KOI-142 · TESS · Kepler · transit duration · signal recovery · public astronomical data · correlated noise