Much public UAP discussion treats every craft as one thing, “the UAP,” and argues about whether the thing is real. Skywatcher, a private group of former military and contractor personnel led publicly by James Fowler, took a different approach: they pointed a multi-sensor stack at a fixed patch of sky for years and sorted what came back into nine distinct classes. The method is instrument-first, turning “I saw something” into a record of what was seen across six sensors. It is a single team’s framework, published by Skywatcher, with no outside replication to date.
The sensor stack
Every claim in the taxonomy is anchored to a specific instrument suite. Skywatcher’s stated reasoning is that one sensor can be fooled or evaded, but a craft that registers on some bands and not others is doing something a balloon, a drone or a bird does not. The suite is electro-optical (high-resolution daylight imaging), shortwave and midwave infrared (heat, propulsion signatures, low-observable detection), radar (altitude, speed, cross-section), a laser range finder (a hard physical return at range), and radio-frequency analysis. Each class is defined partly by which of these bands it shows up on and what it does while being watched.
“Class one tend to fly in groups. We’ve seen one or two as a singleton, but usually they’re in groups of three to 20. That group tends to fly through in some kind of formation.” James Fowler, in Jesse Michels’ Skywatcher clip, 2025
The nine classes
I · Tetra. A black tetrahedron, 1.5 to 2 m, with white vapour bursts; strobes on electro-optical, tumbles multi-axis while holding a stable path, descends in coordinated groups of 3 to 20+. Strong infrared, but no reflective return when lased.
II · Tic Tac. A bright white cylinder, 9 to 12 m, that shifts through purple and pink in transitions. Hovers, drops from 60,000+ ft at Mach 1.5+, stops dead, and exerts an area effect on local electronics. Seen in groups of 1 to 3, often near Manta Rays.
III · Blob. A red, cloud-like orb, 3 to 5 m, with a pulsing core that flares in size. Makes near-instant jumps between positions with no visible transition. Always a singleton.
IV · Beam. An infrared-only orb, invisible to the eye, strong heat signature. Hovers motionless for hours, vibrates laterally, moves slowly when it moves. Usually in pairs. Size unestablished.
V · Manta Ray. A black manta-ray shape, 2 to 3 m, venting coloured plumage (pink, green, teal, white). Tumbles on multiple axes, moves with and against the wind. Rare, and seen near Tic Tacs.
VI · Bright Star. A reflective tetrahedron, 1.5 to 3 m, flashing multiple colours. Oscillates horizontally fast enough to throw a radar artefact like a spinning propeller. Some encounters involved apparent directed-energy sensor disruption.
VII · Jellyfish. A white form with a bulbous ~2 m head and 3 to 5 m trailing tendrils that move independently of the wind. Winds along serpentine paths with a “seeking” head motion, and throws a localised area effect on electronics.
VIII · Hornet. A multi-coloured dual-body craft, 2 to 4 m, with hanging tendrils and a tail that folds into a ball or extends like a stinger. Its upper and lower components rotate asynchronously while it manoeuvres.
IX · Egg. An off-white metallic ovoid, and explicitly provisional: one low-quality recording, no motion data, no spectral coverage, no verified size. Skywatcher lists it with a flag attached.
On the record
The nine classes function as a vocabulary. “A tetrahedron that tumbles multi-axis while descending from high altitude” is a more falsifiable description than “an unidentified flying object.” Three facts sit with the framework.
It is single-source. Skywatcher is the only group publishing it; the Galileo Project, AARO, NASA’s UAP study and the peer-reviewed literature do not use these categories.
Skywatcher does not describe it as settled. Its own pages flag it as a living taxonomy, describe the hypotheses as exploratory, and mark Class IX as provisional on a single recording.
A “class” implies discrete types, but the same observations could reflect one or two phenomena expressing differently under different sensor conditions, with the nine-way split following from how the data is binned. Skywatcher’s stated answer is that the differences are too consistent across encounters to fold together.
What to watch
The framework predicts that independent observation with a separate sensor stack would reproduce the same nine kinds of signature. Such a replication needs no leak and no confession, only another team pointing comparable instruments at the same sky. As of writing, no independent replication has been published.
Sources for this page
- Skywatcher project, Research / UAP Classification Guide (the canonical, continuously updated taxonomy), skywatcher.ai/research, plus the individual class pages.
- James Fowler interviewed by Jesse Michels, 2025 (used as a secondary source for context and quotes).
- American Alchemy long-form interview with James Fowler, 2025.