Six of nineteen numeric rows in a 2026 transit table fail its stated rate conversion at printed precision with the listed periods

We test all 20 rows in Table 4 of Kutluay et al. (2026), PASP 138, 024405. Nineteen rows give numeric quadratic coefficients and central period-change rates. We use the listed periods and the stated definition a=(dP/dE)/2. The rate at the reference cycle is (2a/P) times the number of milliseconds in a Julian year. Exact rational bounds allow half a last printed place for each central input. Six rows have converted intervals that do not overlap the printed-rate intervals. They are TrES-1, WASP-87, WASP-103, WASP-114, WASP-121, and WASP-122. Thirteen rows pass; Qatar-1 gives bounds and is excluded from the central-value count. Earlier SCI studies report the TrES-1 and WASP-121 cases. We confirm them and add the other four cases to a complete audit. All six failures remain with year lengths from 365 to 366 days and periods within five printed errors. One common conversion multiplier cannot make all 19 rows agree. Two quadratic clocks that share the reference time and period, but use either the printed coefficient or the coefficient implied by the printed rate, differ by 22 to 168 seconds after ten Julian years for the four additional cases. These are conditional model differences, not observed residuals. The result tests printed central values and does not select corrected entries or establish an orbital change. The package preserves source files, hashes, all input tokens, code, figures, results, and instructions. Eight tests pass, and a fresh-directory run reproduces the numeric files exactly.

transit timing · orbital period · quadratic ephemeris · data validation · rounding intervals · reproducibility · hot Jupiters