During the night of 11–12 September 1967, radar personnel at Kincheloe Air Force Base in Michigan reported a succession of unexplained returns. Some appeared to change direction or move at very high speed. The surviving Air Force papers record follow-up questions, corrections to the count, checks with other radar units and a proposed explanation involving atmospheric effects on radar. They also preserve important qualifications that disappear when the case is reduced to seventeen objects flying together.
The report from radar approach control
The 25-page official file identifies the reporting location as the base’s radar approach control site and the equipment as an MPN-14. It names Staff Sergeant Michael Burns as the watch supervisor and also identifies airmen Rodney Roe and Robert Carlson as observers of the radar activity. The case rests principally on what appeared on their displays, rather than on visible objects photographed over the base.
The original message gives an interval from 0342Z to 0501Z on 12 September, a span of seventy-nine minutes. This article retains those recorded UTC times while describing the local overnight episode as 11–12 September. It does not silently substitute a modern local-clock conversion for the original message. A later technical review uses a somewhat different local-time interval.
According to the reports, individual returns remained on the scope for about twenty to sixty seconds. Stated speeds ranged from approximately 150 to 2,000 miles per hour, and some courses changed. These were radar-derived estimates. They do not independently establish the speed or manoeuvrability of seventeen physical aircraft.
Seventeen in total, two at once
The administrative correspondence is unusually useful because investigators asked what the numbers meant. A 26 September telephone memorandum records confusion over whether only two objects had appeared rather than seventeen. Information officer Terrance E. Leaman explained that seventeen was the total across the episode; two was the greatest number seen simultaneously.
His 28 September clarification also corrects an earlier count of sixteen. It says Burns reported seventeen and suggests that one had been skipped during the telephone description. A later letter still refers to the original sixteen-item message. Retaining the correction is more accurate than choosing whichever number makes the encounter sound larger or more consistent.
The clarification describes two occasions when two returns appeared together. In one, two differently directed tracks reportedly met west of Kincheloe and then moved east alongside one another at a stated 2,000 miles per hour. In the other, two tracks were present without the same joining behaviour. These distinctions do not support a claim that seventeen objects formed a simultaneous fleet.
The missing visual confirmation
The file repeatedly says the radar targets were not visually observed. Tower personnel were on duty, and a B-52 in the area was asked about unusual observations. A telephone note records that its crew saw nothing. The base’s 3 October reply reports no associated noise and no visual observation, even for returns described as passing overhead.
That lack of visual confirmation does not prove that every return had the same ordinary cause. It does prevent the radar narratives from being turned into a separate eyewitness sighting of lights, shapes or occupants. The only feature tag assigned to this case is radar. No shape, light, landing or water-entry tag is inferred from the courses on the display.
One letter also mentions an earlier report from two children on 10 September. Officials did not forward it as substantiated evidence. It is a separate report, not an eyewitness match for the later radar sequence. This review does not combine it with the radar returns to create apparent corroboration.
What other stations did and did not see
The 14 September enquiries asked which other radars should have detected the targets, whether scope photographs existed and why no interceptors had been launched. A memorandum mentions a single strobe associated with another station, but the follow-up correspondence does not establish independent tracking of the full sequence by the regional air-defence network.
The 29th Air Division’s written response says the SAGE-linked radars did not detect the reported objects. It also explains that radar approach control had a relatively limited range and that a scramble required authority from the division’s senior director. The episode was not reported to that headquarters until the following day. Those recorded procedural reasons are relevant before interpreting the absence of an interception as evidence of secrecy.
The papers retain a complication: low-altitude coverage was limited. The absence of matching SAGE tracks therefore has to be assessed with the coverage problem in mind. It is evidence considered by the investigators, not an absolute demonstration that nothing physical could have been present anywhere in the region.
A proposed radar explanation
The equipment was checked and no malfunction was found, according to the written response. Weather personnel suggested that a temperature inversion could have produced anomalous propagation and false returns. A normally functioning radar can still be affected by the conditions through which its signals travel; a fault check and an atmospheric explanation address different questions.
The packet includes weather material and an original message noting an inversion at roughly 4,000–5,000 feet. It also records consultation about whether a low, supersonic object should have produced an audible shock wave. That consultation was a hypothetical comparison, not an observed sonic boom, and the assumed altitude of the unidentified returns was uncertain.
A Foreign Technology Division assessment dated 15 January 1968, signed by Colonel Jerome J. Jones, is especially careful. It says the radar data provide no positive way to prove or disprove the presence of the reported objects. It nevertheless favours a local atmospheric explanation, citing the changing courses, broad speed range, lack of noise and absence of wider radar detections.
This is stronger than an unsupported dismissal, because it sets out reasons and refers to the available technical context. It is also less conclusive than a demonstrated identification of each return. The letter’s own reservation should remain alongside its preference for temperature-inversion effects.
The Colorado analysis
Gordon O. Thayer’s later Colorado report analysis also favours anomalous propagation. He describes layered atmospheric conditions that could generate false echoes, but notes the absence of target-height information needed to check the calculated apparent altitudes. His analysis adds a physical model; it does not supply missing measurements of the targets.
Assessment and location
UKPX classifies the case as Possibly Explained. The Air Force and Colorado analyses provide a substantive atmospheric explanation, supported by the recorded weather context. The uncertain heights, lack of matching visual observations and the Air Force assessment’s explicit limit prevent a claim that every reported track was independently reconstructed and identified.
The map uses an approximate Kincheloe community reference from the US Census Bureau’s Michigan place gazetteer, rounded from the published internal point. It identifies the locality where the radar observations were reported. It is not the exact radar unit, a plot of alleged flight paths or a statement that every return lay within a particular state or national boundary.
No visitor access to the former military site or current property is implied. This is a historical instrument-report case, not a public investigation venue. The lead image shows the later technical assessment, with its date and qualifications intact. This review read the full official packet and checked the decisive assessment and timing pages against the scans. Source and outcome assessment: 13 November 2026.