Skip to content
Broken UFO Project BROKEN UFO
An MQ-9 Reaper drone in flight, with the sensor turret under the nose
SOURCE PUBLIC EXP-01911 OPEN Sighting SECTOR: SIGNALS AND NOISE

Six Dots That Beat Like Wings: The UFO Science Council’s First Study and the Altitude That Is Still Blacked Out

Six members of the science council that advises the U.S. Government on UAP argue that the six dots in the Pentagon's PR135 video are birds, because their brightness pulses seven or eight times per second. Their math checks out. What doesn't add up is the recorded altitude, and the data point that would settle it is blacked out in the report.

OPERATIONAL SUMMARY

On July 29, 2025, an aircraft of the U.S. Air Force's 65th Special Operations Squadron recorded in infrared six small, spherical objects that the mission report places at FL220 and 480 miles per hour, both estimated. The video (DOW-UAP-PR135) and the report (DOW-UAP-D108) were published on September 18, 2026 in the sixth PURSUE release. A preprint by six members of the science council chaired by Avi Loeb argues they are birds: their brightness pulses at 7.1-7.9 Hz. On October 16 they extended the list to ten videos.

1
What is claimedThat the six objects in PR135 are a flock of birds closer to the aircraft than the crew believed, that the 480 mph speed is a parallax effect, and that nine other PURSUE videos show objects with similar pulses, around 8% of the published sensor videos.
2
What evidence existsReport D108 (identical on war.gov and on Zenodo), video PR135 (identical data on DVIDS and Zenodo), the arXiv 2610.08692 preprint in its four versions with published code and light curves, and the PURSUE portal CSV (downloaded 10/20/2026) for the video count. The frequencies and the parallax calculation were redone separately with the authors' data.
3
What doubts remain openThe aircraft's altitude, blacked out, which decides how high the birds would be flying; whether the crew recorded FL220 and 480 mph correctly; the exact location (Qaruh is a possible identification, not confirmed); the type of aircraft (MQ-9 is deduction); whether the pulse reproduces when decoding the video, which has not been done here; and what peer review will say.
12 min
2,444 words
Middle East, U.S. Central Command area (location blacked out in the report)
V2 · Corroborated Source: Members of the UAP Science Advisory Council (arXiv preprint 2610.08692) on Department of War material (PURSUE)
Start reading
MISSION DEBRIEF EXP-01911

Executive summary of the case file. Everything essential before you dig deeper.

Fact sheet

Code EXP-01911
Source material PUBLIC
Status OPEN
Type Sighting
Incident date 29/07/2025
Location Middle East, U.S. Central Command area (location blacked out in the report)
Opening date 20/10/2026
Last review 20/10/2026

OSINT - Traceability

Verification: V2 · Corroborated Source: Research Entity: Members of the UAP Science Advisory Council (arXiv preprint 2610.08692) on Department of War material (PURSUE)
Opening date 20/10/2026
Incident date 29/07/2025
Status OPEN

The speed is written with an O. In the “Kinetic Velocity” box of mission report DOW-UAP-D108 it says “48OMPH”: a four, an eight, the letter O and the abbreviation for miles per hour. It has been read as 480, which is what the video’s official entry says, and that is how the authors of the study I’m covering here read it too. But they take the trouble to note that 48 miles per hour would, if you had to choose, be a very bird-like speed.

I like papers that make typing mistakes, because they say more than the ones that don’t. This one was published by the Department of War on September 18, in the sixth PURSUE release, alongside the infrared video DOW-UAP-PR135. Since October 6 it has been at the center of an argument that isn’t about saucers but about arithmetic: six members of the science council chaired by Avi Loeb argue that the six dots in the video are a flock of birds. On Friday the 16th they extended the list to ten videos.

DECLASSIFIEDREF: DOW-UAP-D108 / PR135
VideoDOW-UAP-PR135, “Unresolved UAP Report, Middle East, 2025”, DVIDS 1023397, 1,798 frames (60 s)
ReportDOW-UAP-D108, MISREP from the 65th Special Operations Squadron, Operation Inherent Resolve, 8 pages
ContactJuly 29, 2025, 20:22:32 Z
DeclassifiedJuly 10, 2026 (scheduled date: July 29, 2050)
PublishedSeptember 18, 2026, sixth PURSUE release
StudyarXiv 2610.08692, version 1 of October 6 and version 4 of October 16, 2026, not peer reviewed
Data and codeZenodo, DOI 10.5281/zenodo.23021686
CheckedTuesday, October 20, 2026

I’m not here to say they are birds. I’m here to tell you what you have to believe for them to be birds, what you have to believe for them not to be, and which data point would settle it. That data point exists, it’s in the report, and it’s blacked out.

What can be verified first: eight pages of MISREP

A MISREP is the form in which a military unit reports how a mission went. This one has eight pages. I downloaded it on October 20 from the Department of War server, which this time let itself be fetched with browser headers, and it is byte-for-byte identical to the copy the authors left on Zenodo. And it carries its text inside: it is not a photo of a sheet of paper.

It bore the SECRET//NOFORN marking. It was declassified on July 10, 2026 by Major General Richard A. Harrison, chief of staff of U.S. Central Command, according to the stamp on each page. The original declassification box said something else: it was meant to sleep until July 29, 2050.

It was issued by the 65th Special Operations Squadron, of Air Force Special Operations Command. Operation: Inherent Resolve. A single aircraft took off at 17:34 Z on July 29, 2025 and landed at 14:11 Z the next day. At 20:22:32 Z, in transit and before reaching its work area, the first contact took place. The report, in capitals and with what the marker left behind:

  • Description: “6X SMALL SPHERICAL OBJECTS GROUPED TOGETHER”. The box for the number of objects, “Number of UAP Sighted”, is blank.
  • Behavior: “SEEMED TO BE AGILE, FREQUENTLY CHNAGING DIRECTION”, typo included, and “ABLE TO CHANGE DIRECTIONS QUICKLY AND SIMULTANEOUSLY”.
  • Object altitude: FL220, about 22,000 feet. Speed: “48OMPH”. Heading: southeast. All three marked “Estimated”.
  • The aircraft: heading northwest, at 120 knots indicated. Its altitude, position, model and tail number, blacked out.
  • Was the object interrogated with any sensor? “NO”. Advanced capabilities? “NO”. Under intelligent control? “UNKNOWN”. The observer’s own assessment, blacked out.

I keep coming back to the last line. What the observer believed they were seeing sits under the (b)(1)(B) mark, the FOIA exemption (FOIA is the U.S. freedom of information law) for what remains classified; of the informant, only the rank is left, sergeant (SSGT/E5). The black covers. But what it doesn’t cover speaks too: the same paper that describes agile objects ticks “NO” in the advanced capabilities box.

EV-01Reliability: HIGH
Report DOW-UAP-D108, 8 pages, downloaded from war.gov on 10/20/2026. SHA-256 9ffec164725f15d4a1d75c680fceae367b6813119d183f52df33a91c23ee116c, identical to the file the authors archived on Zenodo. According to its metadata, created 09/16/2026 and modified 09/17/2026.
EV-02Reliability: HIGH
Video DOW-UAP-PR135, DVIDS file DOD_111985782 (video 1023397). The DVIDS copy is 22,771,452 bytes and the Zenodo one 22,766,393: the first 22,766,393 bytes of the DVIDS file are the Zenodo file, byte-for-byte identical, and the remaining 5,059 bytes are XMP metadata that DVIDS adds at the end. 1,798 frames in both.

What the study measures: a brightness that pulses

Now, the study. It is signed by Jacob Haqq-Misra (Blue Marble Space, Seattle), Ravi Kopparapu (NASA Goddard), Richard Cloete and Omer Eldadi (Galileo Project, Harvard), Michael Shermer (The Skeptics Society) and Loeb himself. A professional skeptic and the astrophysicist who has spent years asking out loud whether some object in the sky could be alien technology, on the same author list. For that alone it deserved a read.

The idea is simple, which is why I like it. A bird flapping its wings changes the surface it shows the camera, and in infrared that looks like a brightness that rises and falls at the rhythm of the wingbeat, or at double that. The frequency of that pulse does not depend on distance. It can be measured without knowing how many meters away the object is, which, according to the authors, is exactly what no PURSUE sensor video provides.

They tracked the six dots frame by frame between seconds 41.3 and 46.3 of the video. All six pulse between 7.11 and 7.85 times per second, with oscillations of 12 to 26% around their mean. And they do not pulse together: each goes at its own rhythm, with phase offsets of between 30 and 170 degrees. If the flicker were manufactured by the video, it would hit all six the same way and at the same time.

The obvious catch is one they see before anyone else. The video’s compression repeats its pattern every four frames: a cadence of 7.49 hertz, right next to what they measure. So they planted constant-brightness points in the video, encoded them with the same settings and measured them the same way: they oscillate between 0.6 and 1.7% in that band, against 12 to 26% for the objects. Ten defective sensor pixels peak between 1.5 and 3.3 hertz.

When I covered the Galileo Project here, I wrote that a project that publishes how many birds it has mistaken for something strange deserves to be listened to. Well, here are the birds. The authors applied the same method to the project’s infrared cameras near Las Vegas and found flying animals pulsing between 3.4 and 10.0 times per second. The literature they cite puts cruising wingbeat at about 2.5 hertz in large birds and above 10 in small songbirds.

“What limits such analysis is not the phenomena but the data.”

Haqq-Misra et al., arXiv 2610.08692, version 4
EV-03Reliability: MEDIUM
My own test. With the light curves the authors publish on Zenodo (136 samples per object) I redid the spectrum using the method the paper describes: peaks of 7.851, 7.375, 7.851, 7.107, 7.644 and 7.603 Hz for objects A to F, the same as in their table. This checks their math, not their measurement: I have not decoded the video, because this server has no FFmpeg.

The parallax math, redone at home

The second pillar is geometric, and here at Broken UFO it has come up twice already, with Go Fast and with Aguadilla. Without a measured distance, the targeting system can only assign the object the distance to the point on the ground where the line of sight intersects it. The speed that comes out is that point’s, not the object’s. And a nearly still object between the aircraft and the sea, seen from a moving aircraft, projects onto the water a point that runs backward.

It fits the report itself: the aircraft heads northwest and the objects, according to the informant, southeast. While the sensor tracks the group, the north indicator on the screen turns only 2.7 degrees: the aircraft was flying almost straight toward them. The changing of direction “simultaneously” also has a reading, because with the camera following the group, any turn of its own moves all six at once. Meanwhile the group rearranges itself and is seen, one after another, as a line, a cluster, pairs and a loose ring.

With the 120 knots indicated and the standard atmosphere, a still object would give 480 miles per hour over the ground if it were at between 66 and 77% of the aircraft’s height. I have redone the calculation with their formula and I get their numbers: the three figures below are from my calculation, rounded, and match those the paper gives.

AIRCRAFT AT 4 KM

3,0km

height of the stationary object that gives 480 mph

AIRCRAFT AT 6.7 KM (FL220)

4,8km

above typical nocturnal migration

STATIONARY OBJECT AT FL220

10,0km

altitude the aircraft would need to have

Choosing the right column is not possible. The aircraft’s altitude is the “Friendly Aircraft Altitude/Depth” box of the report, and it is blacked out.

The catch they raise themselves

What I liked most about the paper is that it doesn’t hide its weak point. They write it like this: “The reported altitude of the objects provides the greatest challenge to the bird hypothesis”. They reason that, since the image has no horizon, only the island and the sea, the objects were below the aircraft. If the crew put them at FL220 knowing how high they themselves were, the aircraft was higher still.

And then, to give 480 miles, nearly still birds would have to be at about 4.8 kilometers or more (between 4.3 and 5.2 if they move somewhat). That is above typical nocturnal migration, which according to the radar studies they cite goes mostly through the first kilometers, and within reach of few birds, such as herons and gulls. Their way out is that the crew estimated or aimed the speed, the altitude or both badly. It could be. The paper, certainly, admits typos.

Terns at eleven at night

The location in the report is also blacked out, but in the first ten seconds of the video a low islet appears with a lattice tower and a building. The authors compare it with the island of Qaruh, off the coast of Kuwait, and call it, in so many words, a “possible but unconfirmed” identification. Terns nest there: between 5,000 and 6,000 on Kubbar, Umm al-Maradim and Qaruh, from May to July, according to the source they cite.

The problem is that terns flap more slowly: 3.3 and 4.5 hertz in the two species measured by Bruderer’s team. They would fit if the brightness rose twice per wingbeat, with an actual wingbeat of 3.6 to 4.0 hertz. That is neither ruled out nor demonstrated. The direct reading, seven or eight wingbeats per second, fits other birds in the same study: ducks, grebes and waders such as plovers and godwits, which flap at 6.6-8.5 hertz. The site, they conclude, admits both.

That leaves the hour. The contact is at 23:22 local time (UTC+3), almost five hours after sunset, and a flock of terns at that hour is, in their phrase, “possible but not typical”.

With the aircraft there is no certainty either. The type is blacked out and the authors deduce it was an MQ-9 Reaper, a remotely piloted drone, from three data points in the report: the 65th Squadron operates MQ-9s, a single aircraft flew 20.6 hours straight, and 120 knots is normal cruise for that model. The first is backed by the Air Force news article on its activation as an MQ-9 squadron, in 2018. The rest is deduction, and they themselves write that no result depends on it.

⚠
WHAT THIS STUDY IS NOT
It is not an official conclusion: in the PURSUE catalog the video is still titled “Unresolved UAP Report”, and as of October 20 I have found no public reaction from AARO or the Department of War. It is a preprint without peer review, submitted to Scientific Reports. Its six authors are unpaid members of the science council that, according to its website, ODNI commissioned Loeb to create and lead, and which reviews only unclassified information. In version 1 they declared “no competing interests”; since version 2, that membership appears as a declared interest. The paper also states that the code, the figures and drafts of the text were developed with Claude Opus, a language model from Anthropic.

From one video to ten in ten days

On October 6, the first version of the paper dealt with a single video. On the 13th, version 3 added four (PR041, PR044, PR088 and PR099). On the 16th, version 4 added another five (PR077, PR079, PR085, PR097 and PR141): in all, nineteen objects with pulses of between 2.3 and 11.0 hertz spread across nine videos. That Friday Loeb reported it on his blog with a percentage headline: “In More Than 8% of the Sensor Videos Declassified by PURSUE, the UAP Might be Birds”.

Inside the text it drops to “about 8%”, and the count is “10 PURSUE videos containing a total of 25 bird candidates”. The 25 match the paper’s table: six in PR135 and nineteen in the other nine. Nobody gives the denominator, so I worked it out from the official catalog.

EV-04Reliability: HIGH
My own test on the CSV of the PURSUE portal (downloaded from war.gov on 10/20/2026, byte-for-byte identical to the 10/12/2026 copy; the portal still ends at the sixth release): 450 entries and 134 videos. Of these, 122 are entries for Department of War military clips (112 in the first five releases, the same figure given in the August paper, and 10 in the sixth; two, PR057a and PR057b, share a file on DVIDS), 7 are from the FBI, 4 from local police and one is the Tremonton film. Ten of 122 is 8.2%; ten of 134, 7.5%.

So, “more than 8%” if you count only the 122 military clips, and “about 8%” if you count everything. Both are true. What cannot be read is “8% of the Pentagon’s UFOs are birds”. They are candidates with a pulse compatible with wings. The paper itself warns that in those nine videos there is no distance, no sizes or speeds are measured, and a body spinning on itself could also pulse like this.

For some there isn’t even an account to dismantle. In PR041 and PR044 the official entry says the informant “did not provide any oral or written description”. The one for PR099, incidentally, puts 25SEP19 in the title, 23SEP19 in the description and 2023 in the incident date. Three dates for one video. I’m filing that in the folder of things that get fixed with one keystroke.

1952-07-02

Tremonton

Delbert C. Newhouse films between 12 and 14 bright objects in Utah. The Air Force will end up leaning toward seabirds.

Source ↗
2025-07-29

Contact at 20:22:32 Z

An aircraft of the 65th Special Operations Squadron, in transit, records six objects. The report puts them at FL220 and 480 mph, estimated.

Source ↗
2026-07-10

Declassification

The chief of staff of Central Command declassifies the report, which had a scheduled date of 2050.

2026-08-12

112 videos

Haqq-Misra and Kopparapu analyze the PURSUE sensor videos and conclude that none carries all the data needed to go from pixels to speed.

Source ↗
2026-09-18

Sixth PURSUE release

PR135 and its report D108 come out, and the Tremonton film too.

Source ↗
2026-10-06

First version

arXiv 2610.08692 and data and code package on Zenodo. PR135 only.

Source ↗
2026-10-13

Version 3

Four more videos with pulsing objects.

Source ↗
2026-10-16

Version 4 and the 8%

Nine more videos, 19 objects. Loeb headlines it as “more than 8%”.

Source ↗
2026-10-20

Check at Broken UFO

Report identical on war.gov and on Zenodo, calculations reproduced and 122 military clip entries in the catalog, which is still at the sixth release.

Tremonton, in the same release

A digression, because the archive has a sense of humor. In the same sixth release that brought PR135 came the Tremonton (Utah) film, shot on July 2, 1952 by Navy petty officer Delbert C. Newhouse with a Bell & Howell Auto Master camera: between 12 and 14 bright objects in the sky. According to the official entry for the Blue Book file (D102), the Air Force’s assessments ended up leaning toward seabirds reflecting the sun.

The entry for the video (PR159) adds, with a straight face, that from second 55 the film goes on to show what looks like “a demonstration of sewing equipment”. Sewing equipment, being demonstrated. Between the two recordings lie seventy-three years, and both end in the same question: how far away was what you see. In 1952 there was no way to know. In 2025, the data point that would narrow it, the aircraft’s altitude, was in the report, and someone blacked it out.

And it’s not the first time the word has come up in PURSUE. In report D010, from May 2022, the operator himself wrote that four of the five objects he saw “FIT CLOSER TO THE PROFILE OF POSSIBLE BIRDS”. The official entry sums it up in two words, “possible birds”, and there it stayed.

No need for Martians (or for birds)

REPORT D108 (07/29/2025)
Six small, spherical objects, agile, changing direction “simultaneously”. FL220 and 480 mph, both estimated and with no distance measured. Advanced capabilities: NO. The observer’s assessment, blacked out.
VS
THE STUDY (VERSION 4, 10/16/2026)
Six objects pulsing at 7.1-7.9 Hz, each at its own rhythm: compatible with wingbeat. The speed may be parallax from something nearly still at two thirds to three quarters of the aircraft’s height. Weak point, acknowledged: if the aircraft was above FL220, birds at about 4.8 km at night.

The hypotheses still standing, from simplest to least simple, with what they have for and against them:

  • Birds near the aircraft. For: the pulse, which is a published measurement with its code. Against: the recorded altitude and the hour.
  • A crew error in estimating or recording. The authors themselves propose it, and the O in “48OMPH” shows that the form admits typos. A speed of 48 would be, as they point out, more typical of a bird.
  • Something tumbling or spinning, like a balloon or debris. That also flickers, but six at once, each at its own rhythm and moving together, is for the paper a less simple explanation.
  • Something fast and distant. The image does not exclude it: parallax only shows that the speed does not require anything fast, not that it wasn’t.

What would settle this

My reading, and it is mine: the method is good, it is published with data and code, and this first paper from the council, which is how Loeb presents it, can be checked by anyone at home. I have checked it as far as I can without decoding the video, and the math works. But “compelling evidence”, which is what they write, seems to me more than a pulse without distance can give, with a recorded altitude that forces birds to almost five kilometers at night.

The data point that settles it doesn’t have to be sought in any hangar. It is the aircraft’s altitude at 20:22 Z on July 29, 2025, and it sits in a box of D108 that someone covered. If it comes out low, the birds lose their main objection. If it comes out high, they would have to fly at almost five kilometers at night and the hypothesis gets much more complicated. Either way we will know something we don’t know today, which is more than a UFO usually gives.

OSINT – Source traceability

File completeness63/70
View breakdown
  • +20 Primary source linked
  • +8 Type: investigacion
  • +10 Identified entity
  • +10 Source date
  • +5 Editorial notes
  • +5 Verification level assigned
  • +5 Documented corroboration
Source type Research
Entity / Origin Members of the UAP Science Advisory Council (arXiv preprint 2610.08692) on Department of War material (PURSUE)
Verification level V2 · Corroborated
Timeline
Published source
Published on BUFO
Primary source arxiv.org
OSINT editorial notes

V2 and not V1: the official document is V1, but the thesis under examination comes from a preprint without peer review, by authors who belong to the advisory council chaired by one of them. The video has not been decoded (there is no FFmpeg on the server), so the pulse measurement has not been reproduced from the pixels: the calculations were redone from the published light curves. Own reading, marked in the body: that the blacked-out observer assessment and the “NO” in advanced capabilities say something; that “compelling evidence” is more than the pulse without distance gives; the three heights in the figures block (own calculation with the paper's formula, which matches its figures). The two counts of the 8% (122 and 134 videos) are our own. No public reaction from AARO or the Department of War found as of 10/20/2026.

Sources
[12] arxivblog.substack.com

FREQUENTLY ASKED QUESTIONS

Has the Pentagon said the objects in the PR135 video are birds?
No. As of October 20, 2026 the video is still cataloged in PURSUE as “Unresolved UAP Report”. The bird hypothesis comes from a preprint by six members of the science council chaired by Avi Loeb, without peer review and submitted to Scientific Reports.
What exactly do they measure?
The brightness of each of the six dots, frame by frame. All six rise and fall between 7.11 and 7.85 times per second, each at its own rhythm and out of phase with one another, something that according to their controls neither the video compression nor the sensor produces. It is the signature a wingbeat can leave on an infrared camera.
How can a bird go 480 miles per hour?
It doesn't. Without a measured distance, the speed the system gives is that of the point on the ground where the line of sight intersects it. A still object at between 66 and 77% of the aircraft's height projects onto the sea a point that moves at that speed. It is parallax, the same effect that explained the Go Fast video.
Is it true that 8% of the PURSUE videos are birds?
Not exactly. They are 10 videos with 25 objects whose brightness pulses at a rate compatible with a wingbeat. Out of the 122 military clips in the catalog it is 8.2%; out of the 134 videos, 7.5%. In nine of them there is no distance or speed to test against, and an object spinning on itself could also pulse.
What is missing to close it?
The aircraft's altitude at the moment of contact, which is in report D108 and is blacked out. With it we would know how high the birds would have to be flying, and whether that is reasonable at night in that area, whose exact location is also blacked out.
Channel active
Newsletter
SEC:BUFO
Email newsletter

Get a briefing when something moves in the dark.

New case files, declassified documents, recommended reading and signs of anomalous activity. One email when it matters. Zero noise. Zero spam.

No mass mailings
Unsubscribe anytime
OSINT-friendly
Activate subscription
Alternate channel
Follow via RSS
Unit note

We don't sell your email, we don't run campaigns, we don't chase you with cookies. We only send what deserves a place in your archive.

Stable signal No ads One-click unsubscribe