Pattern relationships
212 records · shape × time, reporter, observer position, colourCross-tabulations of the extracted incident fields. The corpus is bimodal in time and reporter, so these relationships are confounded — era, reporting agency, and observation mode covary. Read them as descriptive structure, not causal claims.
The corpus as a field
199 dated records · year × observer positionEach point is one incident, placed by year (horizontal) and observer position (vertical lanes), encoded by attribute (recolour below). Groupings that emerge — e.g. mid-century ground discs vs. modern airborne orbs vs. the spacecraft cluster — are the distinct populations in the record, not predefined categories.
Object shape over time
composition per decade, records with a stated shapeDisc/saucer descriptions dominate the 1940s–50s; orb/sphere descriptions dominate the 2000s+. The shift tracks period vocabulary and the change in who is reporting and how, as much as any change in the objects.
Shape by reporting agency
FBI records (mid-century) are disc-dominated; DoD records (modern) are orb-dominated — largely the same temporal split on another axis.
Shape by observer position
Observer situation: ground, aircraft, ship, or spacecraft. Ground observers account for most records; instrumented platforms report a narrower shape range.
Shape × colour
co-occurrence count across the corpus| metallic | silver | white | black | red | orange | green | blue | yellow | ||
|---|---|---|---|---|---|---|---|---|---|---|
| disc | 9 | 6 | 1 | 3 | 2 | 1 | 1 | |||
| orb | 3 | 8 | 6 | 8 | 10 | 3 | 3 | |||
| ellipsoid | 6 | 2 | 3 | 2 | 5 | 2 | 2 | 1 | ||
| cylinder | 5 | 2 | ||||||||
| triangular | 1 | 1 | 1 | |||||||
| football | 2 | |||||||||
| irregular | 3 | 1 | 1 | 1 | 1 |
Notable cells: metallic/silver with disc (mid-century structured craft), green with orb (the green-fireball cases near nuclear sites), and orange with orb (modern orange-orb reports). Colour and shape associate by era.
Reading across the variables
what the cross-tabs converge onOne split, three axes. Shape over time, by reporting agency, and by observer position are not three findings but one, viewed three ways. Disc/saucer descriptions fall from 47% of pre-1960 reports to 0% after 2000, while orbs rise from 22% to 53%. The agency and platform splits track this exactly, because all three are proxies for a single shift — from mid-century ground eyewitnesses (FBI case files) to modern instrumented sensors (DoD targeting pods). The dominant structure in this dataset is how it was observed, not necessarily what was.
Shape detail falls as measurement rigour rises. Ground eyewitnesses produce the richest shape vocabulary (disc-led, 46% discs, seven distinct forms); aircraft sensors collapse to orbs (48%, discs nearly absent); spacecraft footage resolves no defined shape at all. And the most-instrumented era is the least resolved — post-2000 records are overwhelmingly left open (47 unresolved vs 2 explained). This cuts both ways: the dramatic “saucer” shapes cluster exactly where the evidence is weakest, yet the strongest modern footage — though featureless — is genuinely unresolved, not explained away. It is the same lesson as the testimony-vs-camera concordance: rich shape is a property of the account, not the measurement.
The residue that isn’t era. One colour–shape pairing recurs across the whole timeline and clusters geographically: green orbs/fireballs run from the 1944 Foo Fighters through the 1948–49 New Mexico green fireballs, a 1973 Soviet weapons range, and a 1976 French encounter — repeatedly near nuclear, weapons, and naval sites (10 green-orb records). It is the one pairing with both temporal persistence and site specificity. Flagged for scrutiny, not asserted as a cause.
Caveat: these are descriptive associations in a small, bimodal corpus with strong selection effects — the modern set is a curated military UAP collection, biased toward the unresolved. The structure is real; the explanation is mostly us — who looked, when, and with what instrument.