A size fit gives 57 percent more mean squared error than a constant wait on six later Vela intervals

We test whether glitch size improves a forecast of the next large glitch in the Vela pulsar. We select fractional frequency steps of at least 10^-6 from a fixed Jodrell Bank catalog snapshot. We fit the first 12 resolved intervals and test the last six. The size fit gives an RMS error of 280 days; a constant wait gives 223 days. Mean squared error rises by 57.4 percent. A 95 percent pair-resampling range for the relative gain is -10.45 to 0.73, so the small sample does not establish a lasting loss or gain. Lower size cuts change the point result but do not give a resampling range above zero. A cut of 10^-7 recovers the published full-sample rank correlation of 0.585, yet gives only a 2.6 percent test gain. Timing error checks retain the main loss. Independent IAR timing supports the last event epoch, not the forecast relation. Catalog selection, unknown errors and shared timing fits remain limits. No new glitch or source is found. Code, selected source rows, exact provenance, controls, figures and offline steps are included.

astronomy · Vela pulsar · glitch forecast · waiting time · chronological validation · null result