The record, on one timeline
Every documented case in the corpus — 197 of them, from the 1942 Battle of Los Angeles to the present — placed at its year and colored by where it stands epistemically. The 191 cases with a ring have a deep dive. Hover any dot for the case; click to open it.
What the distribution actually says
As you investigate any single case harder, it tends to collapse toward a prosaic explanation — not harden into an anomaly.
That asymmetry is the central finding of the whole record. The strongest-sounding cases deflate on scrutiny: the Belgian wave's iconic photo was a confessed styrofoam model, Ubatuba's "impossible" metal was ordinary magnesium, the Trinity crash artifact was ordinary aluminum and its key witness was on another continent. A 2026 adversarial re-audit of the strongest open cases bore this out directly: of 16 marquee cases pressure-tested against the best current prosaic analysis, most were downgraded — even the famous instrumented ones (RB-47, Nimitz, Lakenheath, Tehran, JAL-1628) fell to under-corroborated once you notice their decisive data was recalled decades later, or was never preserved or released. "Open" is not "evidence for" — only 10% of the 197 cases stay open, and the handful that survive the audit (Levelland's independent car-stallings, the Coyne helicopter, Socorro's same-day traces) share one feature: a correlated physical effect across independent observers, not a single ambiguous light — and even those mostly hinge on data no one can now re-examine. The robustly real thing this record documents is human: how people perceive ambiguous skies, and how belief is built, sold, and institutionalized. The discipline of separating documented fact from claim is the point; read the colors honestly.