Signal Analysis
What the Data Actually Says
214,480 reports run through the instruments — the full timeline, the shapes, the hotspots, the hour of night they happen, and a detector for the years the signal spiked. No interpretation the numbers don't support.
214,480
Total sightings
1947–2023
Modern record
+31,328 fringe reports pre-1947, earliest 1400
31,328
Unique locations
51 states · 23 shape categories
3
Reporting sources
Sightings over time
1947–2023 · per year- 1952 D.C. flap · Blue Book opens — 102 reports
- 1965 The 1965 wave (Exeter) — 403 reports
- 1973 1973 humanoid wave (Pascagoula) — 499 reports
- 1997 Phoenix Lights — 2,514 reports
- 2008 Stephenville & Tic-Tac era — 9,463 reports
- 2012 Reporting-era peak — 14,692 reports
- 2013 Reporting-era peak — 14,085 reports
- 2014 Reporting-era peak — 10,414 reports
Shape distribution
all-time · rankedTriangles as a share of all reports
by decadeTriangles climb from under 1% in the 1940s to a 1980s peak of 12.8% — the "black triangle" era — then ease back as lights and circles reclaim share. Growth, but not the straight line you'd expect.
Geographic hotspots
where the signal concentratesTop 20 states
raw countPer capita
per 100k · 2020 censusNormalizing by population flips the map: dark-sky, low-density states (WA, VT, MT, AK, ME) rise to the top while the big-population leaders fall back.
Top 20 cities
31,328 distinct locations on recordTemporal patterns
when sightings happenHour of night
peak 21:00 · 57% fall in the 8pm–1am windowBy month
summer = 31.8% (base 25%)By day of week
peak SatWeekend nights win — Saturday and Sunday evenings carry the most reports, consistent with more people outdoors after dark, not more craft.
The flap detector
years > 2× their 5-yr trailing averageA "flap" is a statistical anomaly: a year whose volume exceeds twice the average of the five years before it (off a non-trivial baseline). 6 years clear the bar.
Explained vs. unexplained
characteristics proxyNUFORC carries no official disposition flag — there is no "debunked" column to sum. What the corpus does record is witness-reported characteristics. We can split those into prosaic signals (an aircraft was nearby) and high-strangeness signals (missing time, landing traces) — a proxy, not an adjudication.