Two speed-order checks retain over 95 percent of formal unbound flags in selected 2024 and 2025 GMN samples

We test two checks of average and initial meteor speed in public Global Meteor Network annual summaries. The main quality samples contain 153,078 unassigned events in 2024 and 208,444 in 2025, with initial speed below 50 km/s. A local solar-escape test selects 351 and 495 formal five-error flags. After a gravity allowance, average speed exceeds initial speed by more than 0.1 km/s for 17 and 15 flags. A second rule uses five times the sum of the quoted speed errors and marks 8 and 7. Each rule thus retains over 95 percent of flags in each main sample. Both rules retain all 54 flags in the strict 2025 sample. A controlled constant-speed example shows that a two percent shared speed-scale error can make a bound object pass the formal escape test while both reversal checks pass it. This example does not diagnose a real event. We confirm no interstellar meteor or new source. Two annual samples from one network do not give independent instrument confirmation. The package includes original code, attributed compact inputs, source hashes, aggregate tables, day resampling, figures, independent numerical checks and offline reproduction steps.

astronomy · Global Meteor Network · interstellar meteors · speed estimates · instrument artifacts · catalog diagnostics · null result · reproducibility