Seven Fireballs Clear a First Interstellar Screen—None Clears the Error Bars Yet

SCIENCE WATCH // SEPTEMBER 8, 2026

Avi Loeb and Volkan Duran have used independent meteor observations to calibrate the public NASA/JPL CNEOS fireball catalogue. Their analysis identifies seven potentially interstellar events among 354 records with complete velocity information. The result is a candidate list, not seven discoveries.

Why speed alone is not enough

An object is classified as interstellar when its reconstructed speed relative to the Sun exceeds the local solar escape speed. For many marginal candidates, the excess is only a few kilometers per second. That is also the scale of plausible measurement error, so a small velocity bias can move an event from an unbound to a bound orbit.

The CNEOS catalogue is unusually valuable because government sensors provide near-global coverage of bright atmospheric entries. Its public records include time, location, energy and velocity components for many events. They generally do not include event-by-event velocity uncertainties or the sensor details needed to reconstruct those uncertainties from first principles.

The calibration approach

The authors compare 19 CNEOS events with independent reference trajectories drawn from optical, satellite, radar and infrasound observations. In the high-quality subset, the comparison gives an empirical measure of typical residuals. That permits a hierarchy: strong candidates for follow-up, marginal cases whose classifications depend on assumed errors, and events consistent with ordinary solar-system origins.

The April 1, 2026 Polar-IM event illustrates the discipline. It is a strong post-2018 anomaly in the CNEOS list, yet independent observations still allow a marginally bound orbit. The scientifically useful conclusion is that a combined uncertainty-bearing reconstruction is needed.

Why this matters to UAP research

The study demonstrates a practical transparency model. A government can publish scientifically valuable event parameters without disclosing the full capabilities of classified sensors. Independent observations can then test the public measurements. UAP releases would benefit from the same structure: time, geometry, uncertainty, calibrated observables and sanitized metadata rather than isolated clips.

Assessment

Promising candidate ranking, no single-event confirmation. The seven events justify targeted cross-matching with independent networks and preservation of raw observations. They do not support claims of alien technology. The advance is methodological: uncertainty changes a dramatic label into a testable research program.