SS 433 MAXI 4–10 keV data give a smaller 6.29-day term in 2020–2026 than in 2014–2019 with fixed clocks

We measured a known orbital and precession frequency combination in public SS 433 X-ray data. We fit fixed clocks, calendar-year offsets, and orbital and precession Fourier profiles. In MAXI 4–10 keV data, the fitted semi-amplitude is 1.188 × 10^-3 photon cm^-2 s^-1 in 2014–2019 and 0.200 × 10^-3 in 2020–2026. The early-minus-late amplitude is 0.988 × 10^-3. A full-residual wild calculation with 324-day blocks gives a 95% interval of 0.523–1.305 × 10^-3. A direct test of equal sine and cosine coefficients gives p = 0.0007. Both models allow separate orbital and precession profiles in each interval. The contrast stays positive after a camera calibration date cut, equal weights, eight harmonics, and an updated orbital clock. BAT does not clearly establish the same amplitude change. These exploratory tests assume an adequate mean model and sufficiently independent, symmetric noise blocks. They do not identify a unique physical cause or establish a new period. A joint upper and lower combination fit does not clearly reject equal amplitudes. A noiseless control shows how a precession phase cut can create a large false peak if an orbital harmonic is omitted. We preserve the raw light curves, provenance, selections, coefficients, source code, figures, and repeat instructions.

SS 433 · MAXI · Swift BAT · X-ray binaries · nutation · combination frequencies · time series · sampling window · reproducible research