Cross-Referenced: Eight Patterns Hidden in Our Own Data

1. The Observable Density Effect — Confirmed, Stronger

Every case file on this site is assessed against the Five Observables — the AATIP-derived framework (Anti-Gravity Lift, Sudden Acceleration, Hypersonic Velocity, Low Observability, Trans-Medium Travel) used to flag performance characteristics that exceed known aerospace technology. Each observable is independently marked present, speculative, or absent for every case, and every case separately carries a 0–100 credibility score built from an entirely different rubric — corroboration quality, chain of custody, witness occupation, instrumentation. Nothing in the site's scoring process ties the two together. At the original 27-case sample, we found a real but statistically thin gap. At 208 cases with full Five Observables data, the same gap is now an order of magnitude more solid.

124 cases — the majority of the archive — have zero present observables, almost entirely eyewitness or debris-based accounts with no instrumented corroboration, and they average 41.0. Cases with three or more present observables average in the high 60s to mid-70s. And there remains exactly one case in the entire, now six-times-larger archive where all five observables are independently marked present: USS Nimitz / Tic Tac Encounter, still the single highest-scored case on the site at 96/100. Adding 175 new case files across every era from 1566 to 2025 gave the archive's staff a real opportunity to independently, if unintentionally, break the pattern — and instead it held: a 55-point gap between the observable-free floor and the fully-instrumented ceiling, with a case count now large enough (124 vs. 1) that the floor number is a real population average, not a handful of anecdotes.

2. ETH and IDH Move Together — and PSH Mirrors ETH Specifically

At the original scale, nine cases carried full theory-alignment data, and we noted that ETH (Extraterrestrial Hypothesis) and IDH (Interdimensional Hypothesis) appeared to move together while PSH (Psycho-Social Hypothesis) moved opposite. That was nine data points. There are now 185, and for the first time we could actually run a real correlation coefficient rather than eyeball a small table.

An r of 0.78 across 185 independently-scored cases is a strong positive correlation by any standard — ETH and IDH really do rise and fall together, confirming the original finding with actual statistical weight behind it for the first time. But the 185-case sample also surfaced a nuance the original nine cases were too small to show: PSH's negative correlation with ETH (−0.40) is more than twice as strong as its negative correlation with IDH (−0.17). PSH isn't simply "the opposite of non-mundane" — it is, specifically, the mirror image of ETH. IDH drifts down when PSH goes up, but far more weakly.

The lowest-ETH cases in the full archive are a recognizable, coherent group in a way the original nine-case sample couldn't reveal: modern, camera- or radar-documented sightings that turned out to have a specific, identifiable mundane cause — a satellite train, a rocket stage, a known space mission. These cases don't just score low on ETH; they score low on both ETH and IDH simultaneously while PSH climbs to 85–92%, exactly the mirror-image relationship the correlation coefficients describe.

3. The Investigator Network Isn't a Triangle Anymore — It's a Hub

The original piece found one closed triangle — three investigators (James McDonald, J. Allen Hynek, Gordon Thayer), each personally tied to exactly two of the same three 1950s radar-visual cases — and treated it as a rare, notable structure worth surfacing. At 212 cases, that same kind of closed triangle is no longer rare. We ran the same search and found 332 of them.

Project Blue Book's scientific consultant, J. Allen Hynek, is now personally connected to 39 of the archive's 212 case files — nearly one in five of every case published on this site, spanning from 1948 radar-visuals through 21st-century military-drone incidents his own career never lived to see cataloged. Between them, Hynek, Vallée, and Ruppelt — the three most-cited names to emerge from the Air Force's own UFO investigation apparatus — personally touch 70 of 212 cases, a full third of the archive. Yet the three men themselves rarely worked the exact same case together: Hynek and Vallée overlap on 8 cases, Hynek and Ruppelt on 9, and Vallée and Ruppelt on precisely zero — two contemporaries of the same UFO-investigation era whose documented case involvements, per this archive's own sourcing, never once coincide.

The original McDonald–Hynek–Thayer triangle still holds — nothing about it was wrong, and it remains a real, historically accurate fact about the 1956–57 radar-visual cases. What changed is its significance: it is no longer a rare structure worth flagging as unusual, because the archive's growth revealed that Hynek alone is such a dominant hub that a closed triangle involving him is now the expected outcome, not the exception. The more accurate, and frankly more interesting, finding at this scale isn't "three investigators form a triangle" — it's that one person's involvement is traceable across nearly a fifth of the entire published archive, a concentration of institutional expertise across seven decades of UAP investigation that a single, still-living-in-memory Air Force consultant's career never remotely intended to produce.

4. The Autumn–Winter Effect — Corrected

The original piece, working from just 23 cleanly-dated cases, claimed a dramatic seasonal inversion: September, November, and December made up nearly half the curated archive (47.8%) against barely a quarter of raw NUFORC volume (25.1%). We owe readers a direct correction. At 194 cleanly single-month-dated cases — more than eight times the original sample, and the result of fixing a real bug in our own extraction script that had been silently missing every date using an abbreviated month name ("Apr" instead of "April") — the effect very substantially shrinks.

Both ratios are now close to 1.0 — the curated, most-scrutinized cases in the archive track the raw NUFORC calendar almost exactly, rather than inverting it. The original 1.9× and 0.8× figures were real numbers honestly computed from the data available at the time, but that data was a small, and it turns out non-representative, slice: the 23 cases that happened to have a cleanly parseable single-month date in the earlier, much smaller archive were disproportionately autumn-and-winter cases for reasons that had nothing to do with any real seasonal signal-to-noise effect. This is exactly the kind of correction our own Methodology section commits to publishing rather than quietly revising away.

Our earlier data exclusive, The July Phenomenon, remains correct on its own core claim — raw NUFORC volume genuinely peaks in summer. What does not hold up is the specific claim that this site's curated case selection meaningfully inverts that pattern. It doesn't, at full scale. January is the one real standout in the corrected table, at nearly double its NUFORC share, but that is a far more modest finding than the one originally published, and we're reporting it as such rather than reaching for a dramatic story the larger dataset doesn't support.

5. The Visibility Signature: Three Independent Proofs Now

This finding asks the sharpest question in the piece: for our best-documented cases, does the raw, independent, uncurated NUFORC civilian dataset — a database this site pulls but had zero editorial hand in populating — show any detectable statistical trace of the event, in the same state and month, against that state's own typical reporting rate? The original piece found one clean proof (Phoenix Lights) and one smaller one (USS Nimitz). Expanding the test set with cases added since then surfaced a third.

Stephenville, Texas, January 2008 now posts a real, independently verifiable 4.6× spike — 46 Texas reports that month against a state baseline of 9.9, a case added to the archive after the original piece was published and one we had not previously tested against NUFORC at all. Chicago O'Hare adds a fourth, more modest data point at 1.9×. Phoenix Lights, recomputed against the current, slightly refined state-baseline calculation, now shows an even starker 8.5× rather than the original 14.2× — the underlying spike is exactly as real, only our averaging method changed, and we're reporting the updated number rather than the more dramatic original one.

The pattern the original piece identified holds up cleanly at the larger scale: an independent statistical footprint in raw public data is a property of mass public visibility, not of high credibility. Malmstrom AFB, one of the best-corroborated military encounters in the entire archive, produces a flat 0.0× — witnessed by a handful of security personnel on a nuclear missile base, not a city. Stephenville and Phoenix, both witnessed by open skies over populated metro areas, produce this piece's two loudest, cleanest signals.

6. Discs Stay Rare-but-Investigated — the Orb Claim Was Wrong

NUFORC's 147,585 raw reports skew toward vague, hard-to-pin-down shapes: LIGHT is the single most common category at 19.48%, followed by CIRCLE (10.15%) and TRIANGLE (9.26%). DISK shapes make up 6.14% of the raw dataset. The original piece found disc-shaped craft dramatically over-represented in the curated archive (3.0×) and, at a sample of just 23 craft-typed cases, found zero instances of SPHERE or ORB — a combined 9.2% of the raw dataset appearing in none of the archive's most-investigated cases. At 205 classified cases, one of those two claims got stronger and the other is now simply false.

Disc-shaped craft are now 6.3 times over-represented in the archive's case files relative to raw sighting share — more than double the original 3.0× finding, and now resting on 79 case files rather than 4. That part of the original finding didn't just survive; it strengthened. What did not survive is the sphere/orb claim. Cases specifically built around a sphere or orb-shaped object — the Cussac Close Encounter's "perfectly smooth metallic sphere," the VFA-11 "Cube-in-Sphere" Encounter, and the AARO Reaper Drone Metallic Sphere Video among them — now make up 13.7% of the classified archive, close to proportional with NUFORC's own 9.61% combined sphere-and-orb share. The "orbs are never investigated" claim was accurate when it was published, for the 23 cases that existed to test it against. It is not accurate now, because the archive itself grew in exactly the direction that falsified it — a real, if unglamorous, example of how a data-journalism piece about a living, growing dataset can be overtaken by the dataset's own growth.

Triangle shapes moved the other direction and are now genuinely under-represented (0.4×) rather than roughly proportional, a comparison the original piece's smaller sample didn't have enough triangle cases to make with any confidence. The likely mechanism for disc over-representation is unchanged from the original analysis: a disc is a bounded, describable object that survives cross-examination and produces a detailed case file, while featureless lights and orbs rarely generate a second detail worth investigating — except, evidently, when the orb in question is a metallic sphere tracked on military FLIR, which is exactly the kind of case this archive has added more of since the original piece ran.

7. NEW: The Wright-Patterson Gravity Well

The original piece never checked whether case files cluster around institutional locations rather than people. At full scale, one location stands out sharply above every other: Wright-Patterson Air Force Base, Ohio — home to the Air Force's Foreign Technology Division and, per persistent but never officially confirmed lore, the alleged storage site of recovered UAP material — is directly referenced in 15 separate case files, more than double the next-closest location.

The 15 cases citing Wright-Patterson span nearly the entire history of American ufology: Kenneth Arnold's 1947 sighting that coined the term "flying saucer," Roswell, the Thomas Mantell crash, Levelland, the Tremonton UFO film, Kecksburg, and Langley AFB's 2023 drone incursions. No other single institution in the archive connects a 1947 contact case to a 2023 one. Unlike the Investigator Network finding, which reflects one man's extraordinarily broad personal caseload, this is a finding about an institution rather than an individual — every officer, analyst, and researcher who ever passed through Wright-Patterson's Foreign Technology Division changed, but the base itself remained the recurring node the historical record keeps routing back through, from the earliest post-Arnold flap through the most recent drone-incursion wave.

8. NEW: Credibility Scores Peaked in the 1950s–90s, Then Fell

Breaking every scored case out by the decade it occurred in surfaces a pattern the original 37-case sample was too small and too concentrated in a handful of eras to show at all: average credibility scores are not flat across time, and they are not simply rising as instrumentation improves. They rose through the mid-20th century, held steady for four decades, and then dropped.

The 1950s and 1990s are statistical tied for the archive's highest-scoring decades, at 53.3 and 53.4 respectively, and every decade from the 1950s through the 1990s sits in a tight 50.3–53.4 band. The 2010s are the clear outlier at the bottom — 37.8, the lowest average of any decade with a double-digit case count, a full 15 points below the mid-century plateau. This almost certainly is not evidence that 2010s UAP encounters were less real; it is evidence about what kinds of 2010s cases this particular archive has chosen to catalog. Several of the decade's added cases are exactly the low-ETH, high-PSH, camera-documented-but-mundanely-explained incidents described in Finding 2 — drone swarms, misidentified satellites, viral videos with a specific identified cause — included precisely because the site's own editorial standard treats a well-documented misidentification as worth a case file, not because the era produced weaker raw evidence than the 1950s did.

Methodology, Sample Size & What We Ruled Out

This piece draws on a direct script-based cross-reference of case-files/index.html's structured card data, the site's internal knowledge-base export of parsed score/date/figure/location fields for all 212 cases, every individual case file's own Five Observables and Theoretical Alignment markup, key-figures.html's 1,238 registered figures, locations/index.html's 424 registered locations, and the live 147,585-record NUFORC dataset that also powers the Sightings DB. No finding here was hand-picked before the data was pulled — the analysis script ran first, against the full current dataset, and this article was written from what it actually returned, including the two corrections above.

Sample size, stated plainly: all 212 live case files carry a complete, parseable numeric credibility score, up from 27 of 37 at the original writing — a real data-completeness improvement in itself. 208 cases carry complete Five Observables data, 205 carry a parseable craft-type field, and 185 carry full ETH/IDH/PSH alignment percentages. 194 of 212 cases reduce to a single, unambiguous month for the seasonal analysis in Finding 4; the remaining 18 either span a date range crossing multiple named months (like the Foo Fighters' 1944–45 wave) or carry only a year.

A bug we found in our own script, and fixed: the original month-extraction logic in Finding 4 matched only full month names ("April") and missed every date using an abbreviated form ("Apr"), which turned out to be the majority format used across the archive. That bug alone shrank our usable sample from a true 194 down to 34, and the 34 that survived were not seasonally representative of the other 160 — which is exactly how the original, now-corrected 47.8%-vs-25.1% figure came about. We're documenting this in detail rather than quietly fixing it, because it's a direct illustration of why every finding in this piece needs to be re-verified as the underlying dataset changes, not treated as permanent once published.

What we checked and discarded, again this round: we tested whether the word count of a case file predicts its credibility score, on the theory that more thoroughly researched cases might score higher simply because they're longer. The correlation is real but weak (Pearson r=0.26, a modest climb from a 40.8 average score in the shortest quartile of cases to 56.0 in the longest). That's not nothing, but it's far too weak to explain the far stronger patterns in Findings 1, 2, and 8, and we're not presenting it as a standalone finding — a case file being long is a symptom of it having real evidentiary substance to report, not an independent cause of a higher score.

Related Theories & Cases

This is one of three original Omni Ledger data-journalism pieces built directly from the site's own aggregate case data, and the case files that anchor the most findings above.

Sources

This piece is a direct, script-based cross-reference of the site's own already-published, already-sourced structured data — not a new set of primary-source claims about any individual case. The sources below are the actual datasets and pages the analysis script read, plus the external, independently-published citations behind the specific historical claims made in Finding III (the McDonald–Hynek–Thayer investigator triangle).

• The Omni Ledger. Case Files Index & Knowledge Base — structured credibility scores, craft-type classifications, Five Observables data, and Theoretical Alignment percentages for the site's 212 live case files. /case-files/
• The Omni Ledger. Key Figures Archive — 1,238 registered key-figure profiles and their case cross-links, the source of the Investigator Network in Finding III. /key-figures.html
• The Omni Ledger. Locations Archive — 424 registered location records, the source of the Wright-Patterson AFB finding in Finding VII. /locations/
• National UFO Reporting Center (NUFORC). Sightings Database, 147,585 records — the independent, uncurated civilian dataset queried directly for Findings IV, V, and VI. nuforc.org
• The Omni Ledger. USS Nimitz / Tic Tac Encounter case file — source of the archive's only 5-of-5 Five Observables case and its 96/100 credibility score (Finding I). /case-files/uss-nimitz.html
• The Omni Ledger. The Roswell Incident case file — source of the ETH/IDH/PSH theory-alignment figures cited in Finding II. /case-files/roswell.html
• The Omni Ledger. Stephenville Lights, Chicago O'Hare, and Phoenix Lights case files — the three independent NUFORC-verified spikes in Finding V. /case-files/stephenville-lights-2008.html
• The Omni Ledger. The Cussac Close Encounter, VFA-11 "Cube-in-Sphere" Encounter, and AARO Reaper Drone Metallic Sphere Video case files — the three sphere/orb cases correcting the original Finding VI claim. /case-files/the-cussac-close-encounter.html
• The Omni Ledger. Lakenheath-Bentwaters case file — source of Gordon Thayer's radar-physics analysis cited in Finding III. /case-files/lakenheath-bentwaters.html
• The Omni Ledger. RB-47 Bomber Chase case file — source of James E. McDonald's and Gordon Thayer's technical review, cited in Finding III. /case-files/rb-47-bomber-chase.html
• The Omni Ledger. Levelland case file — the case where Hynek's and McDonald's investigation footprints overlap, cited in Finding III. /case-files/levelland.html
• The Omni Ledger. Kenneth Arnold Sighting, Thomas Mantell P-51 Crash, and Langley AFB UAP Incursions case files — three of the 15 cases citing Wright-Patterson AFB across the archive's full historical span, per Finding VII. /case-files/kenneth-arnold.html
• McDonald, James E. "Science in Default: 22 Years of Inadequate UFO Investigations." Paper delivered to the American Association for the Advancement of Science annual meeting, Boston, December 1969. kirkmcd.princeton.edu
• University of Colorado UFO Project. Scientific Study of Unidentified Flying Objects (the Condon Report), 1968 — the source for Gordon Thayer's role as the Condon Committee's radar-physics specialist. archive.org
• National Archives and Records Administration (NARA). "Project Blue Book Records Research Guide" — the source for J. Allen Hynek's role as Project Blue Book's scientific consultant. archives.gov
• The Omni Ledger. Cash-Landrum and Malmstrom AFB Missile Shutdowns case files — two of the no-detectable-spike comparison cases in Finding V. /case-files/cash-landrum.html
• The Omni Ledger. Seasonal Sightings Analysis ("The July Phenomenon") — the site's own earlier NUFORC data exclusive, revisited and partially corrected against this piece's Finding IV. /theories/seasonal-sightings-analysis/

Last Updated: August 6, 2026