A 22-interval NICER test does not establish adjacent cancellation beyond the PSR J0537-6910 size fit

We test adjacent errors around a frozen RXTE glitch-size forecast for PSR J0537-6910. The 22 NICER interval errors give Pearson correlation -0.370 and lower-tail order-permutation probability 0.0645. This does not establish cancellation at a 5 percent level. A conditional model with shared event-date errors gives probability 0.0838 for independent intrinsic errors. The 44 historical RXTE residuals give correlation -0.129, outside the main range of a specific independent Gaussian threshold-difference model; its conditional upper-tail probability is 0.00206. NICER does not resolve the tested residual laws. This diagnostic does not exclude broader threshold models. Shared date errors, missing glitches, fit covariance, and short series limit inference. No new glitch, planet, or physical mechanism is found. The package includes code, selected rows, source checks, exact hashes, sensitivity tests, a figure, and offline steps.

astronomy · PSR J0537-6910 · pulsar glitches · residual correlation · stress threshold · date uncertainty · null result