Fixed 20.058-day fits to IGR J16493−4348 differ between the 2005–2017 and 2017–2026 Swift/BAT data

We test whether the first sine and cosine coefficients of the published 20.058-day X-ray pattern in IGR J16493−4348 stay the same in later public Swift/BAT data. We use 84,168 accepted single-pointing measurements from MJD 53416 to 61333, with a division at MJD 57924, just after the earlier study's time range. Each interval has its own baseline and orbital model. At the fixed period, the coefficient pairs differ with a two-parameter statistic of 14.292 and an approximate probability of 0.00079 under a 60-day block covariance model. A 100,000-draw wild score test gives 0.00049. Daily averages and changes to weights, orbital terms, and block length give similar results. Dithered-only data have lower precision and give 0.0277. The later fixed-period signal is not detected on its own, with probability 0.380 against zero. Its conservative 95% amplitude bound is 0.277 in units of 10^-3 count cm^-2 s^-1, above the earlier fitted amplitude of 0.229. Thus, the data give evidence against equal fixed-period coefficients but do not establish a decline in amplitude alone, loss of the physical cycle, or a cause for the change. The probabilities depend on the period, nuisance model, and block independence assumptions. The package preserves frozen source tables, provenance, code, figures, results, and an independent full-matrix check.

IGR J16493-4348 · Swift BAT · X-ray binary · superorbital modulation · fixed-period test · public astronomical data · block covariance · reproducible analysis