PURSUE did not prove aliens. It did something more citable: it moved a set of claims from suspicion, civilian research, and secondhand testimony into the government's own declassified files. Here is the before and after, claim by claim, with the primary source for every “after” — and an honest ledger of what still cannot be said.
Through early 2026, the defensible public position was narrow. UAP were officially acknowledged as real sensor phenomena (the 2021 ODNI assessment; the 2004 and 2015 Navy videos), officially unexplained in a minority of cases, and officially disconnected from anything exotic.
AARO's 2024 historical review found no evidence of reverse-engineering programs or non-human intelligence, and attributed much historical excitement to misidentified secret aircraft such as the U-2 and OXCART. The nuclear-proximity claim lived in civilian research without primary-source confirmation. The Grusch allegations were sworn but secondhand. The green fireballs were a footnote known mostly through secondary literature.
PURSUE did not overturn any of that. What it did was move specific claims across the line that matters most for a journalist or a court: from “asserted” to “documented in the government's own files.” The ledger below tracks exactly that shift, and stops precisely where the documents stop.
Seven claims, before and after. Every “after” cell links to the primary record that moved it.
A claim from civilian researchers, most prominently Robert Hastings' compilation of veteran testimony. No primary-source government confirmation, and skeptics reasonably noted the pattern could reflect where cleared witnesses and good sensors happen to be.
The government's own files record unexplained aerial activity over its most sensitive nuclear installations across seven decades: the 1949 Los Alamos conference on the green fireballs, the 2015 Pantex incident tracked by radar and pursued by the Protective Force, and, at the Soviet equivalent, a CIA report of an unidentified object over the Sary Shagan weapons range in 1973 (which US reporting bias cannot explain). Tranche 5 carried the line forward again: two witnesses with no known connection, interviewed nine months apart by the FBI, describing triangular objects over Colorado Springs, home to NORAD, Space Command and Cheyenne Mountain, in the same month of 2023, a year after the Cheyenne Mountain object of 2022. Seventy-four years, the same address.
The public evidence was a handful of Navy clips (2004 Nimitz, 2015 Gimbal and GoFast) and the 2021 ODNI assessment that most UAP were real but unexplained. Striking, but a small and contested sample.
The corpus holds standardized military mission reports logging objects with repeatable anomalous signatures, cold-infrared, shortwave-only detection, right-angle turns, parent-child deployment, filed through official channels. And the government's own analytic arm leaves its flagship case roughly 40% unresolved, in writing, signed.
That the government ran perception-management and monitored civilian researchers was alleged, most vividly in the Bennewitz affair, but rested largely on testimony and inference.
It is now primary-source fact: the 1953 Robertson Panel's debunk-and-monitor recommendation, the Blue Book closure pipeline from the O'Brien committee to Condon, and compartmented multi-agency routing are all in the released files. The management machine is the single best-evidenced thing in the record.
The idea that reported shapes drift by era, discs giving way to tic-tacs and orbs, was a pattern noticed in the UFO literature: anecdotal, and disputed.
It is now measurable in the government's own dated files. The classic disc takes 71% of the 1940s record and 45% of the 1950s, then collapses to two documents in the whole of the 2020s; the luminous orb runs the entire span, from the 1949 green fireballs to 35% of the 2020s record, where every structured modern form joins it. The tic-tac has its own era in between, 65% of the 1990s–2010s, and recedes in turn.
The 1948–49 New Mexico green fireballs were a historical footnote, known mostly through secondary accounts and Lincoln LaPaz's later writing.
The verbatim, classified transcript of the 16 February 1949 Los Alamos conference is public: Teller, Bradbury, Reines, and the meteoriticist LaPaz on the record, LaPaz arguing the lights “could not be conventional meteorite falls,” the men who built the bomb unable to explain the sky over their own labs.
The working assumption, held as firmly by believers as by skeptics, was that the government had never actually measured one of these things. The historical record was eyewitness description and grainy film, and the famous Great Falls (1950) and Tremonton (1952) movies were argued over for seventy years on the strength of what people thought they could see in them.
The Naval Photographic Interpretation Center did the arithmetic in 1953 and wrote the numbers down. Frame by frame, across 322 measurements, it computed velocities from 378 to 2,457 mph and accelerations reaching 21,168 mph per second, on the order of 965 g, and it used those very figures to rule out birds, aircraft and balloons before concluding the objects were “of a nature which we are not able to identify in terms of natural phenomena or commonly known man-made objects.” Eleven years later the destroyer USS Gyatt held a contact above 3,800 knots for twenty-one minutes, filmed its own radar scope, and watched an F-8 Crusader fail to catch it. The measurements existed all along. They were classified.
The field’s oldest accusation, formal enough that a French committee of retired generals and scientists put it to a head of state in 1999. But the hardest evidence was the 1953 Robertson Panel’s debunk-and-monitor recommendation, and a recommendation from an advisory panel is not an instruction.
Tranche 5 produced the instruction, and it is five years older than the Panel. Air Materiel Command’s own 1947–48 study found a residue of reports with no everyday explanation, recorded that no physical evidence had been obtained, and stated that the discs, cigar-shaped objects and balls of light were not of domestic origin. The same compilation carries the guidance: “It is not considered advisable to present to the press information on those objects which we cannot yet identify or about which we cannot present any reasonable conclusions.” The private finding and the public posture sit in one file, signed and dated November 1948.
The other half of the ledger, and the reason the rest can be trusted. None of the following is established by the record, and this site does not assert them.
Origin, nature, or intent. Nothing in the corpus speaks to where the objects come from, what they are, or whether anything intends anything. Every document stops at observation.
Craft, bodies, or materials. No released document claims a recovered vehicle, biological remains, or exotic material. The record is reports and sensor data, not hardware.
A single phenomenon. “UAP” is a residue category, what's left after the mundane is removed. The record does not establish that the strong cases are the same thing as one another.
Nuclear clustering, corrected for base rate. The proximity pattern is documented, but not shown to survive correction for observation density, nuclear sites are also where the cleared observers and best sensors are.
An unbreakable case. The flagship case, the Western U.S. Event, still rests on trained eyes rather than instruments. Where instruments do exist, they are old and their operators said so: the 1953 photogrammetric analysis notes in its own text that “because of the lack of proper equipment, money and personnel, the investigation was limited.” Any individual strong case could yet fall to better data.
Put the two halves together and the honest headline writes itself. The management apparatus, the seventy-nine-year documented pattern, the standardized modern reporting, the era-drift of form, is real and now primary-source. The cargo, what the objects are, remains undetermined. Five PURSUE releases did not break that tie. What they did instead was push both ends of the documented record further apart in time and closer to primary sources: Tranche 4 supplied the 1949 Los Alamos transcript and the 2015 Pantex report; Tranche 5 pushed the origin back again, to the 1946 Swedish ghost rockets and the 1948 order not to tell the press, and forward to a 2026 event still being reported. That is a stronger and more defensible story than any claim of proof.
The machine is real. The cargo is undetermined.
The one-sentence verdict of the whole investigation