The 2023–2026 LMC X-4 BAT mean eclipse fit is 23 minutes later than the fixed cubic clock with exposure midpoint times

We measure the mean orbital eclipse of LMC X-4 with public Swift/BAT 15–50 keV pointing data. The later interval is MJD 60212–61334, after the last data date in Jain, Sharma and Paul (2024). An earlier interval, MJD 57943–60212, supplies the eclipse shape and brightness profile. A fixed published superorbital clock selects predicted bright states. We fit 3,554 reference and 1,397 later pointings. The model averages a symmetric trapezoid over each exposure. With exposure midpoint times, the later mean centre is 22.7 minutes after the fixed published cubic clock. The reference centre is 8.6 minutes after that clock. The fitted change is 14.1 minutes. For groups of one brightness cycle, the 95% interval for this change is 1.8 to 22.9 minutes. For groups of two cycles, it is 1.2 to 22.9 minutes. For groups of four cycles, it is 0.02 to 22.48 minutes. These conditional intervals use 2,000 resamples for each group length. Simulation error in the lower endpoint for four-cycle groups spans zero. Its positive sign is not stable. A forward calculation at the actual observation times gives a 10.8-minute change under the fixed quadratic clock after reference calibration. An exposure start control gives a 12.4-minute change and a different absolute delay. The result is a new timing benchmark under the stated time and profile assumptions. Source structure, sparse later coverage, and the fixed brightness rule limit its physical meaning. The published orbital parameter covariance is not available. The data do not select a full orbital clock family or establish a new physical period change. The package includes frozen inputs, provenance, code, fit results, resampling draws, figures, and reproduction instructions.

LMC X-4 · Swift BAT · orbital eclipse · eclipse timing · orbital ephemeris · exposure timing · cycle bootstrap · public astronomical data · reproducible research