The positive case
What is the best evidence, actually?
Three pages of this archive answer the negative question. No government has ever said it holds a body. No laboratory has ever opened a bag of crash debris and found anything but Earth. No claimed alien material has come back from a spectrometer as anything other than industrial. Those pages are honest and they are also, on their own, a lie by omission — because if you only ever print the failures, a reader is entitled to ask what you are leaving out.
This is what we are leaving out. It is the other half of the ledger: every case in the archive that we think a hostile, competent, well-read sceptic cannot simply dismiss. Not cases that prove anything — there are none of those — but cases where the ordinary explanation has to work for its money.
There are five. Not five hundred. Five cases where a machine and a trained observer recorded the same unexplained object, an official body investigated it with every advantage, and the file is still open.
Every one of the five is set out below with the strongest counter-argument against it stated at full strength, because a list of best evidence that hides its own defeaters is advocacy, not archiving. Underneath them sit the cases with excellent testimony and nothing to check it against, and beneath those, the famous ones that do not hold at all — the ones this field leans on hardest and should not.
From Release 04 · 10 July 2026
The Pentagon just published the experiment nobody asked for
The forty records the Department of War released on 10 July 2026 contain something the previous three tranches did not, and it is easy to miss because it is not a sighting. Three of the files are U.S. Navy Range Fouler Debriefs — the standard form an aviator fills in when something unauthorised strays into a live training range — and each one was released alongside the sensor video of the same event. For the first time in this archive you can read what the observer said and then watch what the machine saw, and hold them against each other.
That is a controlled experiment, and the government ran it by accident. Here is what it returned.
The three videos are, to an unnerving degree, the same video. Eleven seconds, twenty seconds, thirty-two seconds of what AARO's own caption calls an area of contrast — a smudge of infrared, held near the centre of frame while the sensor zooms and pans, exiting the frame when the operator changes mode. No shape. No scale. No measured speed. If you watched all three with the sound off and the paperwork taken away, you could not order them by strangeness, because there is nothing in them to be strange.
Now put the debriefs back. In the Atlantic 2020 file, the operator writes that the object was a darker, maroonish colour, twelve to fifteen feet high, that it travelled with the wind, that it did not manoeuvre or change direction, and that it looked like a large, somewhat deformed balloon. He has solved his own case in the paperwork. It is a balloon, and the Department of War published the footage of it anyway — correctly, because a disclosure programme that only releases the exciting files is not a disclosure programme.
In the Eastern U.S. 2019 file, five military-affiliated personnel watched an object with flight characteristics unlike anything the reporting observer had seen in twenty-eight years of service across the U.S. Air Force and U.S. Navy, and he notes that others with equal or greater experience were also unable to place it. Same form. Same programme. Same smudge on the screen.
The instrument is not adjudicating anything. It never was. Every difference between a balloon and the most unexplained sighting in the tranche lives in the witness statement — and for the hundreds of Navy sensor clips released without one, we have been reading the smudge and supplying the rest ourselves.
That is the most important thing in Release 04, and it cuts against the people who want these videos to be proof and against the people who want them dismissed. It also has a second edge, buried in the Pantex file: pages five and six of that report had already been released, seven weeks earlier, with more of the page blacked out and the incident date withheld entirely. Somebody went back, reconsidered, and let more through. The redactions in this archive are provisional. Treat every black bar as a question rather than an answer.
The instrument and the observer agree, and nobody has closed the file
A machine recorded it. A trained observer watched it. An official body with every incentive to file it away could not. These five are the whole of the serious evidence for unexplained aerial phenomena, and each of them still has a way out.
The Nimitz Tic Tac
The record. AN/SPY-1 radar aboard USS Princeton, tracking objects on and off for a week; ATFLIR infrared video from a second F/A-18.
Four aircrew from two aircraft saw an object with no wings, no exhaust and no rotor, over a patch of disturbed water. The Princeton had been holding radar tracks on something for days before the intercept. The Department of Defense released the video, has never withdrawn it, and has never explained it. It is the only case in the archive with radar, eyewitness and video from a chain of professional observers who were on duty and expected to be believed.
Where it gives way. The three records are not one record. The radar tracks, Commander Fravor’s visual encounter and the FLIR1 video are three separate documents that are *assumed* to concern the same object; the video was shot by a different crew after Fravor’s engagement had ended. The famous performance — the drop from 28,000 feet, the instant acceleration — rests on pilot recollection, not on any measured track that has been published. Twenty years on, no radar data has been released.
Lakenheath–Bentwaters
The record. Three independent ground radars at two RAF stations, plus the airborne radar of a Venom night fighter, over roughly six hours.
The Condon Committee was assembled to close this subject down, and it did. But its own report calls Lakenheath the most puzzling and unusual case in the radar-visual files, and concedes that the probability that at least one genuine UFO was involved appears to be fairly high. That is a hostile witness testifying against interest, which is the strongest kind there is.
Where it gives way. Nothing primary survives. The case exists as a reconstruction built from teletype summaries and the memories of radar operators interviewed a decade later. Anomalous propagation — a temperature inversion bending radar returns off the ground — can generate exactly this pattern of fast, jumping, unresponsive targets, and Suffolk in August is a good place for one.
The Tehran F-4 Intercept
The record. Ground radar at Mehrabad; airborne intercept radar on the second F-4 Phantom.
Two Imperial Iranian Air Force jets were scrambled at an object over the capital. Both lost instruments and communications as they closed on it and recovered them as they turned away. The second crew acquired it on radar and reported that the weapons panel went dead as the pilot attempted to fire. The Defense Intelligence Agency wrote it up and distributed the report to the White House, the National Security Agency and the CIA, with an appraisal noting that it met all the criteria for a valid study of the phenomenon.
Where it gives way. The entire case is that one DIA summary of pilot interviews. No radar tape, no maintenance record, no independent instrument reading. Electronics failing on a 1976 Phantom under a hard manoeuvre at night is not a rare event, and “the systems failed” told by the man whose systems failed is testimony, not measurement.
Minot Air Force Base
The record. B-52 airborne radar return; UHF interference on the aircraft; ground observation from missile security teams.
Sixteen airmen scattered across a Minuteman missile field watched a large glowing object for hours, and a B-52 on approach picked up a return and lost its UHF at the same moment. Independent groups of observers, on the ground and in the air, none of whom could see each other, converged on the same object over America’s nuclear deterrent.
Where it gives way. Project Blue Book filed it as plasma — an explanation that satisfies almost nobody and reads as a desk clearing a case. But a bad official explanation is not evidence of anything in particular. Ball lightning, a plasma discharge and a very ordinary star seen through a temperature inversion all remain on the table, and the radar return survives only as a crew’s description of it.
Range Fouler Debrief DOW-UAP-D090
The record. Twenty seconds of infrared footage from a sensor aboard a civilian aircraft, released as DOW-UAP-PR112.
This is new, and it is the strongest single item in the July 2026 tranche. Five military-affiliated personnel reported an object with flight characteristics unlike anything the reporting observer had seen in twenty-eight years of service with the U.S. Air Force and U.S. Navy, and he adds that others with equal or more experience were also unsure what it might be. The Department of War released the debrief and the video together — the observer’s words and the machine’s record of the same seconds, side by side. Nothing else in the corpus does that.
Where it gives way. The video is twenty seconds of an area of contrast that leaves the frame when the sensor changes zoom. It has no shape, no scale and no measured speed. Everything a reader takes from this case comes from the debrief; the footage corroborates that a sensor was pointed at something, and nothing else.
The testimony is excellent and there is nothing to check it against
Honest people, often several of them, often trained, describing something none of them can explain — and no radar, no film, no trace that outlived the week. These cases stay interesting forever without ever becoming evidence.
Lonnie Zamora at Socorro
A police officer breaking off a speed chase saw a landed object and two small figures beside it, and when he went back there were four burned, indented landing marks and scorched brush that other officers and Army investigators photographed within hours. Blue Book left it unidentified, and J. Allen Hynek, the programme’s own consultant, thought Zamora’s honesty beyond question.
Where it gives way. One witness. The physical trace is consistent with a craft having landed and equally consistent with a hoax having been staged — a persistent local account holds that it was, and the soil showed nothing unusual under analysis. A single credible man is exactly what a good hoax needs.
The Coyne Helicopter Encounter
Four Army Reserve aircrew on a night flight saw an object close on them head-on, and the helicopter climbed several hundred feet with the collective still held down. The magnetic compass spun. A family on the ground below, who did not know the crew and had no reason to invent anything, described the same object and the helicopter beneath it.
Where it gives way. The climb is the whole case, and it is reconstructed from a crew who were, by their own account, badly frightened at the time. A startled pilot pulling collective without registering it is the mundane reading, and it has never been ruled out. There is no flight-data recorder, no radar and no photograph.
The Levelland Engine Stops
In a few hours, a string of drivers who did not know one another telephoned the police to say the same thing: a glowing egg-shaped object on the road ahead, and the car’s engine and lights dying as it approached and restarting as it left. The sheriff went out and saw a light himself.
Where it gives way. Ball lightning, which does stop engines and does appear in electrical storms, and there was a storm. The Air Force investigation — which sent one man for a few hours — was so cursory that the case has been carried ever since on the weakness of the debunking rather than the strength of the evidence.
The Belgian Triangle Wave
The record. F-16 fire-control radar locks; ground radar at Glons and Semmerzake.
The Belgian Air Force did something no other air force has done: it cooperated openly, published its data, and let civilian investigators argue with it. Thousands of ordinary people reported a slow, silent triangle. Two F-16s achieved radar locks that broke when the target performed accelerations no aircraft could survive.
Where it gives way. The photograph that made the wave famous — the Petit-Rechain triangle — was confessed a hoax in 2011 by the man who took it. Later analysis of the F-16 tapes attributes the locks to interference between the two fighters’ own radars and to ground returns, and no witness on the ground reported the violent accelerations the radar appeared to show, which is difficult to reconcile if both were watching the same thing. What remains is a very large number of people seeing lights.
Chiles–Whitted
Two Eastern Air Lines pilots reported a torpedo-shaped object with rows of lighted windows passing within a few hundred feet of their DC-3 in clear night air. It was the case that convinced the staff of Project Sign to draft the Estimate of the Situation — the classified document that concluded the objects were interplanetary, and which the Air Force Chief of Staff ordered burned.
Where it gives way. A bolide. The 1948 Delta Aquariid stream was active, a bright fireball was reported over the same region that night, and a long-duration meteor breaking up in the atmosphere produces exactly the strung-out row of glowing fragments that two startled pilots would read as windows. This is now the mainstream explanation and it fits without strain.
The famous ones, which do not hold
These are the cases the field leans on hardest. Each has been taken apart, and in three of the five it was taken apart using its own primary evidence. A list of the best evidence that quietly omits the celebrated failures is not a list, it is a sales pitch.
GO FAST
The record. ATFLIR infrared video, released by the Department of Defense.
A U.S. Navy sensor video, officially declassified, showing an object skimming the ocean at extraordinary speed while the pilots whoop at what they are seeing.
Why it fails. It is not going fast. The ATFLIR display puts the aircraft’s own altitude, the target range and the sensor angles on screen, and working the trigonometry from the numbers in the frame gives an object at around 13,000 feet moving at perhaps 40 miles per hour — a balloon, drifting. The apparent speed is parallax from a jet moving quickly past a slow object against a fixed sea. The video refutes itself, using its own on-screen data.
GIMBAL
The record. ATFLIR infrared video, released by the Department of Defense.
The object appears to rotate on its axis in mid-air while holding station against a high wind, and the aircrew can be heard reacting to a whole fleet of them.
Why it fails. The rotation is in the camera. The ATFLIR is mounted on a gimbal, and when the sensor tracks a target across the aircraft’s flight path the field of view rolls to keep the target centred — which is precisely where in the clip the apparent rotation happens. The glare-shaped signature is characteristic of the sensor’s own optical artefact, not of an object.
Aguadilla
The record. Thermal video from a Customs and Border Protection DHC-8 surveillance aircraft.
Nearly three minutes of continuous instrument footage — far more than any Navy clip — of an object crossing an airport, entering the sea, emerging, and splitting in two.
Why it fails. Balloons. The object’s track matches the wind, its speed matches the wind, the “split” is two balloons on a common tether separating, and the “entry into the sea” is the object passing behind the coastline from the aircraft’s viewing angle — it never touches water. The reconstruction has been done frame by frame against the sensor’s own metadata and it fits. The single longest piece of instrument footage in the field turns out to be the strongest argument for how easily instrument footage misleads.
Rendlesham Forest
The record. Lieutenant Colonel Charles Halt’s real-time audio recording, made in the forest.
Trained U.S. Air Force security personnel guarding a base that stored nuclear weapons walked into the trees after a light and came back describing a landed craft. The deputy base commander went out the second night with a tape running.
Why it fails. The bearing Halt gives on his own tape points at the Orford Ness lighthouse, which flashes on that line every five seconds, and the tape captures the men reacting to a light doing exactly that. The landing marks were rabbit diggings. The radiation readings were within background for the area on the instrument used. The case has been taken apart using its own primary evidence.
Range Fouler Debrief DOW-UAP-D091
The record. Thirty-two seconds of infrared footage from a U.S. military platform, released as DOW-UAP-PR116.
A U.S. Navy sensor video of an unidentified object, published by the Department of War in July 2026 under a programme created by presidential order to disclose UAP encounters.
Why it fails. The operator solves it in the debrief. He describes a darker, maroonish object twelve to fifteen feet high, travelling with the wind, that did not manoeuvre or change direction and looked like a large, somewhat deformed balloon. It is a balloon. The government released the footage anyway, correctly, because the programme releases what it holds — and this file is the most useful document in the tranche precisely because it shows what a UAP video looks like when the witness statement is attached to it.
What would put a sixth case in Tier I
Not a better video. The Aguadilla footage runs for nearly three minutes of continuous calibrated thermal imagery and it is balloons; length and resolution are not the bottleneck and never were. What Tier I is short of is independent instruments — two systems, on different physical principles, at different locations, recording the same object at the same timestamp, with the raw data released rather than described. Radar and infrared together, with the tapes. Nothing in five thousand cases has that. The Nimitz encounter comes closest and its radar data has been withheld for twenty-two years.
The July 2026 tranche did not deliver it either, and that is worth saying plainly rather than dressing up. What it delivered instead was a lesson about how little the videos carry on their own, which is less thrilling and considerably more useful.
The scoring method behind every number on this page is set out in the Alien Proof Index, which is deterministic and disagrees with the ranking above in places — deliberately, and the disagreements are the point. The negative case is in alien recovery, crash recovery and recovered technology. The primary documents are in the disclosure archive, and you should read them instead of taking our word for any of this.