TESS fits outside eclipses show a lower U Gru pulse strength ratio near 33.86 and 33.04 cycles per day in 2026 than in 2018

We compare two pulse frequency neighborhoods in the first and latest public 120-second TESS visits to U Gru. We fit the same four orbital phase intervals outside eclipses in sectors 1, 68, 95, and 105. Strength is the root mean square of the fitted complex amplitudes across intervals. The strength near 33.86 cycles per day divided by that near 33.04 cycles per day is 0.770 in 2018 and 0.190 in 2026. The endpoint factor is 0.246, or a decrease of about 75 percent. It is 0.203 to 0.364 under the tested flux, phase, baseline, and nuisance-signal choices. The nominal one-day residual-block interval is 0.150 to 0.404. Signal recovery shows upward bias and incomplete interval coverage when the second pulse is weak, so this interval is not a calibrated general confidence bound. Prior work already reported possible variation in 2020. This work adds uniform measurements for the later 2023, 2025, and 2026 visits. The result applies to a fitted statistic in the target aperture flux. It does not establish a physical cause or the identity of a stable mode across eight years. We preserve the four raw light curves, file hashes, code, fitted samples, figures, and recovery trials.

U Gru · TESS · stellar pulsations · eclipsing binaries · amplitude variation · reproducible photometry