Ukraine.Kyiv · 'Phantoms' and 'Cosmics' of the Main Astronomical Observatory

Between 2018 and 2022, three researchers at the Main Astronomical Observatory (MAO) of the National Academy of Sciences of Ukraine, B. E. Zhilyaev, V. N. Petukhov and V. M. Reshetnyk, video-recorded the daytime sky over Kyiv and its region in search of meteors and what they called 'space intrusions'. They used two meteor stations 120 km apart: one in Kyiv (the observatory is in the Holosiiv district) and one in the village of Vinarivka, in the south of the region. According to their first paper, posted on arXiv on 23 August 2022, the stations carried ZWO ASI 178 MC and ASI 294 Pro astronomical cameras with 6 mm lenses, recorded at 50 frames per second or more with exposures under one millisecond, and pointed at the zenith or the Moon. The fast objects turned up as a by-product of that work, not in a dedicated campaign.

The authors split what they recorded into two classes. 'Cosmics' were luminous objects, brighter than the sky, which they named after birds (swift, falcon, eagle); they described 'squadrons' crossing the field at 3 to 15 degrees per second and some whose brightness varied regularly between 10 and 20 Hz. 'Phantoms' were dark spots, with a contrast from a few percent up to about 50%. To give them a distance they did not use triangulation but colour: they assumed the object is a black body with zero albedo and that its contrast against the sky measures how much scattering atmosphere lies in front of it. With that method, a contrast of 0.3 gave 3.5 km and one of 0.55 gave 6 km, and from there came sizes of 3 to 12 m and speeds of up to 15 km/s. The only distance control they offered was a water tower about 300 m away, for which the method gave '0 ± 1 km'. The paper also includes, undated, a luminous object recorded by both stations at once against the Moon: from a parallax of about 5 degrees they placed it 1,524 km away and 1,174 km high, at 282 km/s, and said the matching light curves at both stations showed it was the same object, flashing about 20 times per second.

The second paper, dated 13 November 2022, added dates. On 24 August 2018, at about 9 a.m., a Canon 5D Mark III in Holosiiv recorded object 9252: visible for 0.17 s, with a computed distance of 8.5 to 13.3 km, a size of about 94 m and a speed of 31.8 km/s. In the same recording, the authors attribute the other flashes to birds and the last one to a plane taking off from the airport; they also note that a military parade for Independence Day took place that day from 11:18 to 11:38. They describe other 'arrivals'. One, undated, supposedly appeared at 11 km, approached to about 4 km and left within 0.72 s, at 79 m and 10 km/s; of it they write that at 4 km it was 'clearly visible to the naked eye' as a black object, though they name no witness. Another, on the same 24 August 2018, supposedly came down from the stratosphere to about 7 km over Kyiv airport at about 30 km/s, dropped from 7 to 1 km in a second and rose again, at about 45 m. A third, on 26 September 2018 (79 m, 0.2 s), supposedly made a 'maneuver' in 0.13 s, and the last was recorded at Vinarivka on 15 October 2022 (20 m at 13.7 km, 2.3 km/s). On 17 October 2022, at 08:58:08 UT, the two stations, synchronised to the millisecond and aimed at the Moon, caught two objects about 0.8 s apart according to their timestamps; by parallax they placed them at 1,130 and 620 km altitude, moving at 78 and 256 km/s, with an angular size that would amount to about a kilometre.

The authors themselves put their doubts in writing: 'Size estimates seem fantastic', and they were troubled that these objects are not detected by radar, asteroid observers or the military. Their way out was to assume an albedo below 1%. They closed the second paper with a quotation about something 'coming from the part of space that lies outside Earth and its atmosphere'. After the first paper, the MAO held an astrophysical seminar on 15 September 2022 at which Zhilyaev presented his arguments, and its Scientific Council concluded that the observations were 'original', but that processing and interpretation had been performed 'at an inappropriate scientific level and with significant errors in determining distances'. It faulted the paper for giving no dates or saying which events were seen simultaneously from both stations, and for not arguing against natural causes or terrestrial objects such as meteors, wind-borne objects or space debris; it called the paper premature and asked that the sentence saying the observatory 'conducts an independent study' of UAP be replaced by one attributing the study to the authors. The second paper, in November, keeps that sentence. On 5 October, Avi Loeb posted on arXiv a comment on the first paper: objects of that size and speed would have produced a visible fireball, at a tenth of the distance they matched artillery shells, and the object at about 1,170 km was probably a satellite.

The case matters for what it shows about method, not for the objects. Every striking number in the papers depends on a single chain: the contrast of a blurred spot becomes a distance, and the distance multiplies the angular size and speed. If the distance is wrong, everything else is wrong in the same proportion. The water-tower control, which the method placed at '0 ± 1 km' when it stood 300 m away, indicates that the technique cannot tell something a few metres away from something a kilometre away, which is exactly the range where insects and birds fly in front of a 6 mm lens. On the Metabunk forum, Mick West pointed to that control, to the fact that the paper did not explain how distance was computed, and to the 'Phantoms' looking very like flies; other participants added that motion blur mixes the object's colour with the sky's, which invalidates the colorimetry. Replying to Loeb, West measured a 'Phantom' from the first paper across three frames taken at 50 per second: the uneven spacing indicates a changing velocity, which would demand about 414 g of a half-metre shell and about 4 g of a half-centimetre fly, and he wrote that it is 'entirely consistent with the curved flight path of an insect'. The authors knew of the risk: they wrote that insects are 'the primary source of interference', that they become invisible beyond 15 m, and proposed telling them apart by curved trajectories, by colour and with two stations; they also computed that a bird crosses the field at 1 or 2 degrees per second and concluded that birds cannot be 'Phantoms'. They did not publish how many objects those criteria discarded.

The observations have to be separated. The single-camera ones, which are almost all the 'Phantoms', have no independent distance measurement at all, and there the reading of nearby insects or birds explains size, speed and darkness with nothing extraordinary. Loeb's artillery-shell explanation was written about the first paper, which does not date its observations; it would fit a wartime sky in 2022, but it does not reach the August and September 2018 recordings that the second paper disclosed later, which predate the full-scale invasion. The two-station detections (the undated object in the first paper and the two of 17 October 2022) are the strongest material, because they do use a parallax. Loeb attributed the first to a satellite, and on Metabunk it was noted that the planned orbit of OneWeb satellites is around 1,170 km; however, no satellite moves at 282, 78 or 256 km/s, so either the parallax matched different objects in each camera or some assumption in the calculation fails, and the papers do not mention checking those positions against a satellite catalogue.

The episode leaves a useful institutional lesson. A national observatory declined to back work by its own staff, said so publicly and explained why; the lead author told the press he would not debate it, and in February 2025 the group posted a third paper on arXiv, 'Characterization of Dark UAPs' (arXiv:2503.05627), which keeps the reading of large, fast black bodies. Both papers came out as arXiv preprints; their comments assign them to Odessa Astronomical Publications and to Kinematics and Physics of Celestial Bodies, publication we have not verified, and we have no data from any other station confirming the figures of hundreds of metres and tens of kilometres per second. The document record ↗ covers the papers and the observatory's correction; this one gathers what was observed and with which instruments.