NARCAP TR - 6, Part 1







Radar Catalogue: A Review of Twenty One Ground and

Airborne Radar UAP Contact Reports Generally Related to Aviation

Safety for the Period October 15, 1948 to September 19, 1976







Martin Shough

United Kingdom




Copyright  2002










                             NARCAP is greatly indebted to the author for permission to

                             reproduce this useful and important report.  It has been included 

                             on our website as a service to those readers who are interested

                             in electro-magnetic effects that are possibly related to UAP.

                             NARCAP neither endorses nor does not endorse this report -

                              its content is entirely the responsibility of the author.


                             Time given is UTC (first entry) and local (second entry). Some

                             entries cite a six digit reference. The first two digits are the day

                             of the month, the last four are hours (UTC).

                             Class:  R = radar, V = visual.

                             Location(s):  Place names are used whenever possible.

                             Sources:  All references cited are in the open literature.  





Table of Contents


             Case           Date                 Time                     Location                                Page



               1.            Oct. 15, 1948           2305L              Kyushu, Japan                         3

               2.             Jan. 22, 1950          0240/0440L    Kodiak NAS, Kodiak, AK      5

               3.             Mar. 9, 1950          1915L              Selfridge AFB, MI                   12

               4.            Sept. 21, 1950         unkn.                Provincetown, MA                   21

               5.            Feb. 2, 1952            night                 Sea of Japan                           24

               6.            Aug. 5-6, 1952         2330L              Haneda AFB, Japan                27

               7.            Dec. 10, 1952         1915L              Odessa, WA                            43

               8.            Dec. 15, 1952          1915L              Goose AFB, Lab., NF              47

               9.            Feb. 4, 1954            2300L              Carswell AFB, TX                   48

               10.           Dec. 2, 1954            1410L              Ceuta, Morocco                      52

               11.          June 23, 1955            1245L              Utica, NY; Boston, MA         54

               12.          July 17, 1957            0430L              So. Central, U.S.A.                  56

               13.           July 25, 1957            0025L              Niagrara Falls, NY                   98

               14.           Aug. 30, 1957            night                 Chesapeake Bay,  VA             102

               15.           Nov. 4, 1957           2245L              Kirtland AFB, NM                 104

               16.            Nov. 5, 1957           0510L              Gulf of Mexico                      107

               17.            May 5, 1965           0110L              Philippine Sea                          112

               18.            May 4, 1966            0430L              Charleston, WV                       117

               19.            Jan. 13, 1967           2200L              Albuquerque, NM                   120

               20.            Nov. 8, 1975           0053L              Malmstrom AFB, MT              122

               21.            Sept. 19, 1976        0030L              Tehran, Iran                              127


                        Appendix                                                                                            137

















1.  DATE: October 15, 1948      TIME: 2305 local                           CLASS: R/V air radar/air


LOCATION:                            SOURCES: Hynek 1978 135

Kyushu                                              McDonald 1968 (House Symp.) 69

Japan                                                            Ruppelt 65


                                                 RADAR DURATION: 10 minutes overall

                                                        (6 separate episodes)


EVALUATIONS: Blue Book - unknown


PRECIS: At 2305 a 2-seat F-61 Black Widow night-fighter was flying off the NW coast of Kyushu, 50 miles at 330 degrees from Fukuoka, when the radar operator picked up a target, range 5 miles at 12 o'clock & slightly below the aircraft (a/c). The a/c speed was between 200 & 220 mph; that of the target was 200 mph, range slowly closing. The aircrew thought they had a friendly fighter. Then the target showed a "slight" change in azimuth and "rapid" closure, appearing at the same time to dive below the a/c. The pilot attempted to follow in a 3500 fpm dive at 300 mph, but air intercept (AI) radar did not immediately reacquire the target. Shortly the radar operator called a second contact, but the target outdistanced the a/c with "a burst of speed dead ahead". On a third intercept the pilot called a visual at 60 degrees to port; the object was visible in clear silhouette against moonlit cloud and the radar acquired a target crossing ahead of the a/c from 45 degrees to port, range 3000' at -5 degrees elevation. The pilot turned to starboard to head off the object, but the radar target put on a "burst of speed" and was lost at 9-10 miles (maximum radar range was 10 miles). At this time the pilot decided that the object he had seen was unfamiliar and queried his ground control station, who reported that there were no known aircraft in the area. The fourth intercept again began with a pilot visual, the object passing above and from the rear. AI radar again picked up the target slightly above at 12 o'clock, range 5 miles, but again it was lost off the set at 10 miles. The fifth and sixth intercepts were similar: The target was picked up at > 9 miles range at 200 mph, the a/c closing with a speed advantage of 20 mph to a range of 12,000', at which point the target pulled ahead to the maximum radar range of 10 miles in about 15-20 seconds.


  Visuals: an "excellent silhouette" against a reflective moonlit undercast on the 3rd intercept, and a "fleeting" glimpse of the object passing from above and behind on the 4th intercept. The object appeared as a stubby cigar with a tapering, squared-off tail-end, a little like a "rifle bullet" the approximate size of an a/c fuselage; it had a dull or dark finish, with no visible features or control surfaces. There were no other a/c in the area, and there was no ground radar contact with the object. Target altitudes were between 5-6000'.


NOTES:  The fact that no ground radar contact was reported is difficult to interpret; according to the intelligence report the F-61 was detected by ground radar during the incident, but only intermittently. According to McDonald, "The report indicates that this may have been due to 'ground clutter'." However the area of the incident is 50 miles out in the Korea Strait between the Tsu Islands and the west coast of Japan. A more likely reason for intermittent painting of the F-61 is that it was flying at or below 6000' at range 50 miles, a line-of sight elevation on the order of 1 degree and thus close to the likely radar horizon. This might well make the a/c a marginal target whose detectability was critically dependent on aspect. The visual description of the "UFO" - a smoothly moulded appearance without visible canopy, wings, power section or tail assembly - is consistent with a target of very much smaller radar cross-section than the F-61 at any aspect, making it a very much more marginal target even than the "intermittently" detected F-61.


   Of six separate AI contacts two were visually corroborated, one by an "excellent silhouette" which resembled no known controlled aeroform of 1948 and is difficult to equate with birds, clouds, windblown debris or balloons. Perhaps the most plausible explanation is that the pilot saw the shadow of his own a/c on the moonlit undercast. The second "fleeting" glimpse might be dismissed. Nevertheless on each occasion the AI contacts corroborated the visual observations in a natural way, picking up the target as the object would have moved into the forward scanning coverage of the radar - first time moving into the pattern at 45 degrees to port, range 3000' following a visual at 60 degrees; second time being picked up slightly above and at 12 o'clock, range 26,000', moving ahead of the a/c after a visual of the object passing above and from behind. Both acquisitions are consistent with the typical elevation-scan limits of this type of AI radar. The speeds and relative movements of the 6 targets on different headings and at different elevations are individually difficult to interpret in terms of anomalous propagation of ground returns, and collectively impossible to interpret in terms of the same set of AP conditions. At the same time the behavior of the targets exhibits a rational consistency which, supported by two corroborative visual observations, is strongly suggestive of a real radar reflective target.


   On the 5th & 6th intercepts, the target accelerated away from the F-61 at a minimum relative speed of about 1400 mph, which, added to the aircraft speed, yields a true airspeed of well over 1600 mph. Clearly the target could not have been another aircraft. Yet the intelligence report notes that "the object seemed cognisant of the whereabouts of the F-61 at all times" as though it carried "radar warning equipment" and concludes that the airmen were "of excellent character and intelligence".  Both men felt strongly that the object was a controlled vehicle: "In my opinion," offered the radar operator, "we were shown a new type aircraft by some agency unknown to us." The first jets, the Gloster Meteor and the Messerschmidt 262, entered service in 1944, but the first flight to exceed Mach 1 was not achieved until 1947 by the experimental Bell X-1 rocket aircraft, and even if some historically unrecorded prototype development of the X-1 had achieved Mach 2 combined with combat-agility within 12 months one would hardly expect it to be idling around over the Sea of Japan.


   In summary the 6 radar and 2 concurrent visual contacts are not easily interpreted in terms of known propagation anomalies or other natural phenomena, and there is a convincing impression of intelligent evasive flying by a vehicle with a performance greatly in excess of known aircraft capability in 1948.


STATUS: unknown



2.  DATE: January 22, 1950        TIME: 0240/0440 local          CLASS: R/V  air radar/air-
                                                                                                                         ground visual

LOCATION:                              SOURCES: Fawcett & Greenwood 164

Kodiak Naval Air Station

Kodiak Island


                                                    RADAR DURATION: unspecified


EVALUATIONS: No official


PRECIS: On February 10 1950 a detailed report on "unidentified airborne objects . . . in the vicinity of Kodiak" was sent out from the Divisional Intelligence Office, 17th Naval District, Kodiak Naval Air Station. Numerous copies were addressed to the CIA, the Director of Intelligence USAF, the FBI, the State Department and elsewhere. Its conclusion was that the sightings were of "phenomena . . . the exact nature of which could not be determined by this office", and an evaluation of A-2 was assigned to the reliability and priority of the information contained. Of the detailed report and voluminous enclosures listed - including radar scope drawings, aircraft track charts, weather data and witness statements - little remains for public scrutiny (as usual) but a summary and two less-than-helpful brief comments appended by unidentified individuals. The "summary of the information contained" can be separated into three distinct incidents which read as follows:


a) At 220240W January Lt. SMITH, USN, patrol plane commander of P2V3 No. 4 of Patrol Squadron One reported an unidentified radar contact 20 miles north of the Naval Air Station, Kodiak, Alaska. When this contact was first made, Lt. SMITH was flying the Kodiak Security Patrol. At 0248W, 8 minutes later a radar contact was made on an object 10 miles southeast of NAS, Kodiak. Lt. SMITH checked with the control tower to determine known traffic in the area, and was informed that there was none. During this period the radar operator, GASKEY, ALC, USN reported intermittent radar interference of a type he had never before experienced (see enclosure (3) [missing]). Contact was lost at this time, but intermittent interference continued.


b) At some time between 0200 and 0300W, MORGAN was standing watch on board the USS Tillamock (ATA 192), which was anchored in the vicinity of buoy 19 in the main ship channel. MORGAN reported sighting a "very fast moving red glow light, which appeared to be of exhaust nature, seemed to come from the southeast, moved clockwise in a large circle in the direction of, and around Kodiak and returned but in a generally southeast direction." MORGAN called CARVER, also on watch, to observe this object, and they both witnessed the return flight. The object was in sight for an estimated 30 seconds. No odour or sound was detected, and the object was described to have the appearance of ball of fire about one foot in diameter.



The second incident occurred some two hours after the above radar contact:


 c) At 220440, conducting routine Kodiak security patrol, Lt. SMITH reported a visual sighting of an unidentified airborne object at a radar range of 5 miles, on the starboard bow. This object showed indications of great speed on the radar scope. (The trailing edge of the blip gave a tail-like indication.) At this time Lt. SMITH called attention of all crew members to the object. An estimated ten seconds later, the object was directly overhead, indicating a speed of about 1800 MPH. Lt. SMITH climbed to intercept and attempted to circle to keep the object in sight. He was unable to do this, as the object was too highly maneuverable. Subsequently the object appeared to be opening the range, and SMITH attempted to close the range. The object was observed to open out somewhat, then to turn to the left and come up on SMITH's quarter. SMITH considered this to be a highly threatening gesture, and turned out all lights in the aircraft. Four minutes later the object disappeared from view in a southeasterly direction.


The third incident occurred about 24 hours later and was solely visual:


d) At 230435W, the day following Lt. SMITH's sighting, Lt. CAUSER and Lt. BARCO of Patrol Squadron One were conducting the Kodiak Security Patrol and sighted an unidentified object. At the time of the sighting the aircraft in which these officers were embarked was approximately 62 miles south of Kodiak. The object appeared to be on an ascending westerly course, and was in sight for ten minutes. During this period the object was observed by Lts. CAUSER and BARCO, and PAULSON, ADi, plane captain. At no time was radar contact made on the object. Lt. CAUSER was unable to close the object at 170 knots.


The summary then amplifies the visual descriptions as follows:


1) To Lt. SMITH and crew it appeared as two orange lights rotating about a common center, "like two jet aircraft making slow rolls in tight formation." It had a wide speed range.


2) To MORGAN and CARVER it appeared as a reddish orange ball of fire about one foot in diameter, travelling at a high rate of speed.


3) To CAUSER, BARCO and PAULSON it appeared to be a pulsating orange yellow projectile shaped flame, with a regular period of pulsation on 3 to 5 seconds, off 3 to 5 seconds. Later, as the object increased the range the pulsation appeared to increase to on 7 to 8 seconds and off 7 to 8 seconds.


Weather and balloon-release information contained in Enclosure 8 (missing) is summarised as follows:


A check with the Navy Weather Center, Kodiak, Alaska revealed that balloons were released at the following times:


      22 January - 0445W and 2200W (approximately)

      23 January - 0400W (approximately)


On 23 January winds aloft at 1000 feet were reported at 0400W as from 310 degrees, at 36 knots, and at 2000 feet, from 240 degrees at 37 knots, while the object was reported to be on an ascending westerly course.


NOTES: As regards the 230435 visual sighting the "pulsation" of the light, considered alone, is quite suggestive of a light below a swaying radiosonde balloon, periodically occluded as observed from the air. However, the winds aloft data quoted indicate that a balloon released from Kodiak at 0400 would, at a typical climb rate of  >1000 fpm, within 2 minutes have encountered winds carrying it NE. If winds above 2000' remained from 240 degrees, then 35 minutes later the balloon would certainly not have been near a position 62 miles south of Kodiak. The upper winds are not specifically stated, however, and it is conceivable that the balloon entered a north-south flow at higher altitude. But for the balloon to reach the observation position within 30 minutes would require a mean wind speed during climb of about 120 mph, and therefore maximum wind speeds of very considerably more than 120 mph. This seems improbable in view of the fact that its maximum altitude even after thirty minutes would be no more than about 35,000'. Also, in view of the fact that wind data are only quoted up to 2000' with specific reference to the possibility of a balloon having been sighted, one may reasonably conclude that the sighting took place at around 2000' (this inference finds independent support later) and thus nowhere near a hypothetical one- or two-candlepower radiosonde lamp at 35,000'. (A leaking balloon at lower altitude would be still less consistent with reasonable wind speeds.) Since even a slow aircraft will rapidly close on any balloon, typical balloon-interceptions involve a close dogfight with a blinking light which appears to make rapid and repeated head-on passes during a circling climb. In this case the aircraft "was unable to close the object at 170 knots" during ten minutes of observation. A radiosonde light, furthermore, would not have been orange.


   The navy intelligence report considered balloons as explanations for this and the other sightings. Its comment on this hypothesis reads:


In view of the fact that no weather balloons were known to have been released within a reasonable time before the sightings, it appears that the object or objects were not balloons. If not balloons, the objects must be regarded as phenomena (possibly meteorites) the exact nature of which could not be determined by this office.


    A balloon was released from Kodiak at approximately 0445 on January 22, just five minutes after the second radar incident reported by Lt. Smith (para. c above). Given that the release time is "approximate" there is presumably a residual possibility that this balloon was in the area at the time of this incident. The report states that Smith "climbed to intercept and attempted to circle to keep the object in sight. He was unable to do this, as the object was too highly maneuverable." This is somewhat consistent with an attempted interception on a balloon. However, the rest of the report is very difficult to interpret as a balloon, in particular the radar-tracked closure at 1800 mph.


   The two radar incidents on January 22 are the core of the sequence and invite more detailed analysis than is possible on the basis of the information available. Nevertheless some observations can be made.


   The incident beginning at 0240 is not strictly a radar-visual, since the sighting reported by the two watch personnel on the U.S.S. Tillamock was "at some time between 0200 and 0300" and thus may not have been concurrent.  However it is reasonable to treat the reports as possibly related. Two separate airborne radar contacts occurred, the first to the north of NAS Kodiak, the second to the southeast 8 minutes later. These locations cannot correspond to the circling of Kodiak and departure southeast of the object observed visually, however, since that entire manoeuvre occupied only some 30 seconds. It is possible that one of the radar contacts related to this object, whilst the other contact involved the same or another object at a different time which was not observed visually.


   The possibility that the second contact may have been a false target generated by RFI is raised by the fact that the radar scope was at that time displaying intermittent interference. However there is insufficient information in the summary to test this hypothesis.


   One of many questions which remain is: if the first radar contact (at least) did correspond to the "very fast moving . . . ball of fire" observed visually from shipboard, why was it not also observed visually from the aircraft? It is possible that both radar targets were caused by propagation anomalies and/or interference from another microwave source. The presence of interference, however, does not exclude the possibility that a real target was also being displayed, and contextually this might be a more attractive explanation given the independent visual report which suggests that there may have been a target (of whatever nature) to detect. An equally plausible hypothesis, therefore, would be that the visual object was the exhaust of an unidentified jet - possibly a Soviet reconnaissance platform - which was briefly painted by the radar but visually aspected such that its exhaust flame was not noticed by the aircrew. The abnormal interference could have been due to signals from the intruder's own radar.


   In the 0440 incident, two hours later, it is made explicit that radar and visual sightings were concurrent. In this case the object seen visually from the aircraft appeared as "two orange lights rotating about a common center", and was compared by the observers to "two jet aircraft making slow rolls in tight formation." The likelihood that this simile is a correct interpretation seems small, given that the lights were observed for several minutes at different bearings from the aircraft. A comment appended to the intelligence summary by an unknown office identified as OP322C2C opines that "the possibility exists that incidents covered by para. 2.a, b & d might be jet aircraft [original emphasis]". No opinion is offered as to the object(s) observed in this case, presumably because the radar-tracked speed is too obviously excessive. The cited 1800 mph, however, is calculated from displayed range-over-time and does not take account of the near head-on closure rate. To correct this figure we need to know the speed of Lt. Smith's patrol aircraft, which is nowhere given. Fortunately this figure can be approximately inferred with reasonable confidence.


   Smith's aircraft is given the designation "P2V3 No.4 of Patrol Squadron One". We know that the aircraft were operated by the US Navy on security patrols around its Kodiak Island facility, and that they carried several crew members. They were clearly not small, high-performance interceptors. The likelihood is, therefore, that these were anti-submarine patrols, and this would be consistent with our inference from the weather data that the patrol by Causer and Barco on January 23 was being flown at the <2000' level. "P2V3" could therefore refer to the Lockheed P-2 Neptune anti-submarine patrol aircraft that was in use by the US Navy up until about 1961. This was a piston-engine aircraft, whose speed and range limitations led to the development of its turboprop successor, the P-3 Orion, in 1958. If Smith was flying a P-2 then he was not flying very fast, an inference consistent with other internal details of the reports.


  In the 230435 incident other members of Smith's squadron were flying the same Security Patrol, presumably in the same type of aircraft, and attempted to close on the object without success at 170 knots - about 195 mph. Presumably this was equal or close to the plane's maximum speed. This can be roughly cross-checked with information in para. a) of the summary: Smith reported a radar contact 20 miles north of Kodiak NAS, then 8 minutes later he reported another 10 miles southeast of the NAS, which, whilst admittedly vague, is consistent with about 25 miles flown at a speed in the region of 180 mph. Adopting a generous 200 mph as the maximum likely speed of Smith's aircraft during the later radar-visual, and neglecting the component of lateral velocity due to the object's closure from a position "on the starboard bow", and assuming the negligible effect of winds to cancel out in the equation, we have a minimum true airspeed for the object of 1600 mph. If further allowances are to be made for possible inaccuracy in timing it might be safe to conclude that the object was travelling somewhere in the region of Mach 2 or greater. This would be pretty remarkable performance for jets in 1950. Even assuming this conservative speed estimate to be out by a factor of 2, one can hardly imagine jets, hostile or friendly, thundering though the Alaskan night at Mach 1 on a head-on pass with a lighted Navy patrol plane whilst performing "slow rolls in tight formation". In summary, it seems highly unlikely that this sighting was caused by jet aircraft.


   The half-hearted suggestion in the intelligence report that the phenomena in all of these incidents were "possibly meteorites" is very much a stab in the dark. None of the features of any of the sightings can be convincingly equated with the characteristics of meteors. Visually the object in the first incident was seen to circle Kodiak and return, and whilst the reported duration of 30 seconds might be questioned as a fallible judgement it is supported by the fact that the first witness watched the object approach and make its wide turn, then had time to call it to the attention of the second witness who joined him to observe its departure. In the two later incidents the visual objects were on rising or turning trajectories and in both cases were in view for minutes rather than seconds.


   The radar contacts reported in the first incident are unable to be evaluated in themselves. The radar contact in the second incident is more circumstantial, and the estimate of speed derived from displayed range and elapsed time is reinforced by the observation that the blip showed a distinct tail on the scope. A target will be swept by a number of radar pulses during each scan of the beam, and if it is moving quite slowly relative to the radar these returned pulses will be integrated into a compact arc which displays as a "spot target" on the tube phosphor. A target moving very rapidly might return pulses showing a changing range /azimuth indication during a single sweep of the beam, and the signals corresponding to these pulses will be displaced on the tube, "smearing" the blip reciprocally to the direction of movement. This smearing will be dependent on several additional factors such as beam width, pulse repetition frequency and scan rate, and no quantitative inferences can be made without detailed specifications of the radar involved; however, the report is qualitatively consistent with a target moving at unusually high relative speed.


    This target is not consistent with returns from meteor-wake ionisation. Multiple trip effects can distort the displayed speed of targets detected beyond the unambiguous range of the set, and in some circumstances a meteor might be displayed at spuriously slow speeds. But typical true speeds of some tens of thousands of mph could not be reduced by this mechanism by an order of magnitude or greater; the target was evidently on an approximate radial heading, and multiple trip effects do not distort radial velocities; meteors on radial headings are generally only observable by ultrasensitive search radars due to the short length of trail scanned at near-grazing incidence, and airborne radars are of low power; and finally the optimum frequency for ionisation returns occurs at about 100 MHz, with cross-sections dropping by a factor of ten-to-the-fifth at about 1000 MHz, and the airborne radar would have operated at a frequency some ten times higher still.


   The reported speed of the target is inconsistent with birds, insects, clear air turbulence, debris, rain, hail, smoke, balloons or other wind borne objects. Partial reflection from headwind-driven waves on an inversion layer just above the aircraft altitude could produce an approaching point target at a displayed speed of twice wind speed (the headwind cancels out the component of Vc due to aircraft airspeed); but winds of 7-800 mph are plainly unrealistic. Several sporadic ground echoes can create the illusion of a fast track on a surveillance PPI when they chance to appear in different locales from scan to scan, but the scan rates of airborne radars are relatively very high indeed an operator is much less likely to be fooled in this way, even disregarding the independent "indications of great speed" given by the smeared target presentation.


   Internal noise sources or radio frequency interference from remote emitters can create false, rapid targets on analogue radars, and the report of interference occurring during and after the first radar incident invites analysis of this possibility. The intelligence summary indicates that two naval offices, in particular, found these reports interesting in the context of research into radar interference.


   Interference "echoes" will generally display as random speckles or patterns of speckles representing small spots of excitation on individual trace radii. Such speckles do not resemble real targets such as aircraft, which display as compact bright arcs due to the several integrated spots of excitation on several adjacent trace radii. To mimic an aircraft-like target on an inbound radial heading, the noise source would have to be a cyclic microwave emission with a duration and sine-wave amplitude comparable to the passage of one beam-width across a target, having a periodicity minutely shorter than the scan rate of the receiving antenna and with a superimposed pulse repetition frequency synchronised to that of the transmitter. This is a very complex set of requirements for any source other than another radar set of the same or closely similar design, whose scan rate would be on the order of a microsecond shorter. Even an interference signal such as this would typically exceed the receiver threshold over a considerable portion of the scan period, displaying as an anomalous broad arc, without some super-added mechanism to filter a discrete train of pulses similar to those returned from a reflective target. The reason for this is the complex lobing which occurs far beyond the narrow angle of the main beam, which is mirrored in the lobed antenna gain of the receiver. Thus signal strength can exceed the receiver threshold unpredictably at various antenna orientations, displaying virtually anywhere around the scope. On a surveillance PPI this effect might be reduced for a relatively weak signal if the two antenna rotations were  synchronised 180 degrees out of phase such that peak output always corresponded with peak gain. With two forward sector-scanning airborne radars it is possible that this filtering might occur in another way: transmitting and receiving antennae could be oriented in such a way that each time the transmitting antenna scanned towards the receiver only a brief train of pulses was detectable at low gain before it scanned away and signal strength dropped below the threshold. With a sector-scanning transmitter it is possible that this filter effect would also work if the signal entered by washing through poorly-shielded amplifier or receiver circuitry rather than by the antenna link.


   The probability of this hypothesis is difficult to gauge. It has to be presumed from the intelligence report that there was no known air traffic at this time which could have been responsible. The report does not mention a second radar target which would have been the skin paint of the aircraft responsible, although it is true that on the above hypothesis the other aircraft - at least at the time of this initial contact - would have been on the periphery of Lt. Smith's radar scan limit and at unknown range and elevation, and thus possibly undetected. (Radar indications during the later parts of the attempted interception of the visual object are not specified in the report.) The visual object might itself be taken as an indication that a culprit aircraft was in the area, but the rolling configuration of orange lights and general performance observed visually are inconsistent with a type of aircraft which would have carried radar equipment similar to that on a P-2 patrol plane. And if the aircrew did see an unidentified aircraft of some type on a course similar to that being indicated by a concurrent radar target, then the attempt to disassociate the one from the other by improbable RFI mechanisms does become more than a little strained.


   The fact that unfamiliar radar interference was reported by the same aircrew during the first incident two hours before is of ambiguous relevance. It can be taken as strengthening the suspicion that some unknown source of RFI might have been responsible for these and the later targets. At the same time it can be taken as indicating the operator's readiness to interpret unfamiliar radar indications as interference, and as underlining his confidence that the unknown targets appeared to him to resemble "real" radar echoes. Given that the inherent probability of convincing target arcs being generated by RFI must be somewhat low, and given the concurrent visual sighting which is difficult to explain, one might be inclined to give the operator the benefit of the doubt.


   The limited comments of two unidentified, presumably naval, offices are included here for perspective:


The opinion of OP322C2C:


"The possibility exists that incidents covered by para. 2.a, b & d might be jet aircraft; however, there is insufficient intelligence to definitely identify the unidentified objects as aircraft. Several reports of similar radar interference have been received from DIO/17ND. It is possible that this is interference from another radar in the vicinity, malfunctioning of components within the radar set, or both."


The opinion of F2:


"Many of the previous reports of radar interference tend to indicate local interference (generated within the aircraft). This looks more like  external interference from sources outside the aircraft than previous reports, though it is far from conclusive. These reports are always of interest."


   The first reported radar incident can't be evaluated, although broadly concurrent visual sightings are of interest in the context of later events. The third, purely visual, report is likewise unevaluable but again borrows some significance from the context. The core radar-visual incident is not easily explained in terms of the information available, and given the unusual nature of the concurrent visual sighting together with certain quantitative inferences from the radar report this case should be carried as an unknown.


STATUS: Unknown



3.  DATE: March 9, 1950            TIME: 1915 EST (approx)         CLASS: R  ground radar


LOCATION:                               SOURCE: Hynek HUFOR 1977 123/295

Selfridge AFB                                               Shough UFO 47-87 215


                                                     RADAR DURATION: 1 hours (approx)


EVALUATION: Grudge - probable balloon


PRECIS: During radar monitoring of a night flying exercise by F-80s of the 56th Fighter Interceptor Group near Selfridge AFB, an unidentified target was observed by an Air Force controller and three other operators (one of whom was the controller from the previous shift) on the PPI of an L-band CPS-5 surveillance radar and simultaneously on the HRI scope of an S-band CPS-4 height-finder. The presentation of the target, which initially appeared on the CPS-4 at high altitudes, was "definitely that of an aircraft" and comparable to the return from an F-80. The target moved with rapid alterations of height and speed, generally about 20,000' above the highest altitude of the F-80s being monitored. "Several extreme instances of gaining and losing altitude" were observed, the target on the HRI scope rising and falling "up to 20,000' very rapidly" with "erratic" changes of speed observed on both scopes. On at least two occasions the target hovered motionless for a mean period of 2 minutes. Maximum estimated speed was 1500 mph, although the controller noted possible inaccuracy in the timing at this point and, being unable to give the target full attention at this time, was confident only that the target was "very fast". Later the controller was able to concentrate fully on the target, and over a period of 6 minutes during which it was "giving indications on both scopes without fade" he and another operator noted range, height, azimuth and time data in grease pencil on the scope heads. The target became intermittent on the CPS-5 at 2052, range 79 miles, altitude 33,000', but the height-finder carried it to 87 miles before losing it. Subsequent intermittent contacts occurred on the CPS-5 out to 120 miles.


NOTES:  The following is the text of a letter, classified SECRET, from the Air Adjutant General to the Director of Intelligence, HQ USAF, Washington, concerning this incidence:


1. Attached for your information are two narrative reports concerning radar sightings of an unidentified flying object.


2. The fact that the object was sighted on the scopes of two (2) radars is considered worthy of special note.


3. Comment of technical experts, this headquarters [HQ Continental Air Command, Mitchel AFB, N.Y.], was solicited and is quoted in part for your consideration:


a. While it is relatively well known that various ionospheric conditions cause reflections at lower frequencies, it is usually considered that those layers have no effect at the frequencies used by the two radar sets mentioned except when temperature inversions or other atmospheric or tropospheric conditions cause ducting and spurious reflections. Presuming that such idealized conditions existed at the time of these observations, it is conceivable that an actual small change in physical lateral action [sic] in reference to the radar set could cause a seemingly greater change in relative position of the "object" as observed on the radar scope due to the varying path lengths the radar energy takes to and from the "object" as a function of the frequency-sensitive layers and angles of incidence of the propagated wave. However the great difference in frequencies of the L-Band CPS-5 and the S-Band CPS-4 radar sets and the evident correlation of observations between these two sets almost rule out the possibility of anomalous propagation effects. Further, the magnitude of velocity and accelerations of the three-dimensional movements of the "object" reported are beyond the capability of known behavior of lighter than air vehicles in controlled flight.

b. Also substantiating this unlikelihood is the fact that the "object" was reported as remaining stationary in free space for a mean period of two minutes.

c. Further validity is lent to the contention of the reports by statements that first indications, which were at high altitudes, were observed on the CPS-4 height-finder before being observed on the CPS-5 surveillance radar set. This follows logic and field experience, inasmuch as the high-altitude coverage of the CPS-5 is known to be poor and the antenna is not capable of being automatically tilted as in the case of the CPS-4 on which the controller may tilt the antenna within wide limits to observe any high-altitude or high-angle objects. It is to be noted that previous field experience with a CPS-5 surveillance radar set has indicated that targets picked up at ranges and altitudes indicated in subject report would probably have a reflection aspect ratio in the order of magnitude of a B-29 or greater.

d. In the absence of detailed vertical and horizontal coverage charts for the specific radar sites and comprehensive weather reports for the area during the period of time these observations were noted, a more complete study or evaluation at this time is not feasible.

e. In summary, no known electronic phenomena, nor combinations of several electronic phenomena, could conceivably produce all of the observations covered by the attached reports.


4. The frequency of reports of this nature has recently increased; instructions have therefore been directed to all radar installations within this command to report scope sightings of unusual objects.


5. It is recommended that reports of unidentified object sightings be reconsidered for submission from all Zone of Interior Air Force agencies.


S/ Neal J. O'Brien,

Col., USAF, Air Adjutant General,

for the Commanding General.


The narrative report of the incident by the controller, 1st Lt. Francis E. Parker, follows:


  On the night of 9 March, or radar station was in operation monitoring night flying by units of the 56th Fighter Interceptor Group, Selfridge ASFB, Mich. I came on duty at approximately sundown, relieved 1st Lt. Mattson at the PPI scope (of the AN/CPS-5 Radar Sight), and established contact with the F-80s already airborne. Lt. Mattson, Sgt. McCarthy and Cpl. Melton, who made up the rest of our crew for that night, mentioned to me at this time that an aircraft had been picked up intermittently on the HRI scope of the AN/CPS-4 height finder radar at 45,000 feet and over. I knew the highest assigned altitude of the F-80s was 24,000 feet; the target was not at that time visible on either radar scope, so I attributed the report of the high-flying aircraft to interference, crew inexperience, or both. Over the next fifteen minutes the rest of the crew, mentioned above, repeatedly reported this high-flying target at apparently rapidly changing altitudes without my being able to turn around rapidly enough from my monitoring of the F-80s in the area to observe for myself. Finally, however, I saw this target which was a very narrow and clear-cut presentation on the HRI scope. It was at approximately 47,000 feet about seventy (70) miles out, and the indication was definitely not that of a cloud or atmospheric phenomenon. I checked pilots in the area by VHF and was assured by [the] F-80 pilot at the highest assigned altitude that he was at 24,000 feet. The clarity, narrowness and definition of the presentation was definitely that of an aircraft. The target gave a similar presentation to that given by an F-80, and if anything, narrower. It was definitely at this time not presenting a very large reflecting surface toward our station and I could not at this time pick up the target on the CPS-5, ruling out B-36 or other large aircraft. Further indications of this aircraft were picked up intermittently but with increasing regularity for the next 45 minutes or an hour, and entries were made of these occurrences in the controller's log; though relatively fairly correct, [they] are inaccurate, due to the extreme inaccuracy of Sgt. McCarthy's watch. During this period, approximately 1945 to 2030, this target seemed to stay in the area in which our fighters were flying, sometimes approximating their courses, but 20,000 feet above them. During this same 45-minute period, Lt. Matson and other members of the crew reported, both from the HRI scope of the AN/CPS-4 and another PPI scope of the AN/CPS-5, that the target hovered in one position and also that it progressed from a position given as 270 degrees, 78 miles at 45,000 feet to a position at 358 degrees, 53 miles at roughly the same altitude in 4 - 5 minutes. This would give it a speed upwards of 1,500 miles per hour for this run. I cannot substantiate this speed. Coverage of target during this run was reportedly intermittent and the times were not to my knowledge accurately tabulated at actual instances of radar pick up during this run. Subsequent individual questioning I undertook with members of the crew bears out the possibility of inaccuracy in timing during this run. I knew only that the target was very fast. I observed during this period, by momentarily turning around and watching the HRI scope, several extreme instances of gaining altitude and losing altitude. I was not able at this time take down the actual figures, but I observed it losing and gaining up to 20,000 feet very rapidly.


   I was able, at 2046 EST, to identify this aircraft on my PPI scope (AN/CPS-5) and simultaneously on the HRI scope. The only actual timing and figures I took down on this target I did during the six minutes from 2046 to 2052, during which time the aircraft was giving indications on both scopes without fade. I took down the range and azimuth on the minute for this period and Sgt. McCarthy took down the altitudes. (Sgt. McCarthy's times were off as aforementioned but in this case, due to the fact that we were both following the same target, I have reconstructed these times into my own, which were taken in grease pencil directly on the scope head, and later transcribed.) Information recorded is as follows:


          Time          Azimuth        Range in Miles         Altitude in feet



          2046          1,560      45          25,000

          2047          1,510      49          29,000

          2048          1,460      56          35,000

          2049          1,420      60          33,000

          2050          1,390      67          36,000

          2051          1,360      73          38,000

          2052          1,330      79          33,000


   These figures, although not as spectacular as some of the climbs and speeds I observed, show definitely the erratic speed and altitude changes. The differences in speed from one minute to the next were apparent to me as were the climbs and dives. At 2052 the aircraft faded from the PPI scope and was picked up for periods of one and two minutes up to 120 miles. It appeared to hover for two minutes at approximately 110 miles distant. It faded at 120 miles for the last time. The height-finder carried the aircraft past the six-minute period listed above to a 1,230, 87 miles, 31,000 feet where it faded for the night from the CPS-4.

      The CPS-5 was very accurate on this particular night which was supported by the F-80 pilots' agreement with many geographical positions given them off the CPS-5. The AN/CPS-4, though a more erratic piece of equipment, could not, through any known or prevalent weakness in its operation, account for this manner of extreme changes in altitude. I went over all possible errors which could be induced by AN/CPS-4 error exhaustively with my technical personnel.

      We are continuing investigations at this station.

       I have been a rated pilot since 12 April 1943, and have been assigned to controller duties for approximately 2 1/2 years.


S/ Francis E. Parker

1st Lt. USAF


   This articulate, careful report gives every indication of being objective and reliable as far as it goes. (The other narrative mentioned in Colonel O'Brien's letter to the Director of Intelligence is presently unavailable.) Project Grudge personnel evidently accepted the evidence from the observers and technical analysts that there was a real radar-reflective target in this case, concluding that it was "unknown". It did not appear as such in the final Blue Book statistics, however, the reason apparently being that Grudge - in defiance of every opinion and every quantitative argument - added the rider that it was a "possible balloon". When the statistics were later reorganized by Blue Book for Air Force PR purposes all "possibles" and "probables" were collapsed into the "explained" category and this case, along with many others effectively disappeared for many years.


   Grudge offered no defense of its "possible balloon" evaluation. Indeed it is difficult to see any grounds for suspecting a balloon given the astonishing wind speeds implied and the fact that the target maintained station at times whilst at others moving at jet speeds on headings 90 degrees apart.


   The heading of the 1500 mph run (whose timing accuracy is uncertain, but which was at least "very fast") is roughly NE, for example, terminating due N of the station at 53 miles; whereas the target was subsequently tracked from a position due E at 45 miles (1,560 mils = 88 degrees). The implied heading between these positions is SW for 70 miles, and the target then departed on a heading roughly NE once more at a mean speed of 370 mph.


   The variations of altitude are also inconsistent with a balloon. The general trend, if there can be said to be one, is of descent from an initial 47,000' to around 30,000', which would imply negative buoyancy; but having dropped to as low as 25,000' at 2046, the target begins to climb, gaining 13,000' in 5 minutes, which is an average rate of climb of 2600 fpm - more than twice that of a fully buoyant weather balloon. Indeed the maximum measured rate of climb during this run was 6000 fpm, with a simultaneous ground speed of about 480 mph, implying inconceivably violent winds and updraughts which would have tossed the F-80s out of the sky - and possibly leveled half of Michigan.


   Birds, insects, weather (rain, hail-cells etc.), clear air turbulence or wind-borne debris cannot account for this behavior. Nor is it likely that internal system faults or mutual/remote radio frequency interference could similarly affect two electronically independent radars with very different receiver bandwidths (on the order of 2gHz apart) even if the presentation and motions of the target were symptomatic of RFI, which they are not. No effect attributable to side lobe returns could account for such a target.


Anaomalous propagation is a poor hypothesis here:

a) because the target behavior is not diagnostic of AP;

b) because the "clarity, narrowness and definition" of the scope presentation is not diagnostic of AP;

c) because the accuracy of CPS-5 positions confirmed by pilot reports does not suggest abnormal propagation;

d) because the super-refractivity due to atmospheric temperature/moisture gradients, and the efficiency of partially reflecting layers, are both sensitive to frequency, rendering it unlikely that correlating returns would be displayed by two instruments;

e) because partial forward scattering from a moving elevated layer would imply (for the 6-minute track alone, means speed 370 mph at a mean altitude of about 30,000') severe hurricane force winds of 185 mph at about 15,000' coexisting with stable stratification, these figures being minima;

f) the mean antenna elevation during the 6-minute plot is about 6 degrees, and the horizontal fan beam of the CPS-4 height-finder (whose vertical beam width would be around 1.0 degree) would probably radiate little energy near zero degrees in comparison with the CPS-5's vertical cosecanted pattern, making ducted ground returns less likely on the height-finder - yet after the target had become intermittent on the CPS-5 at 1330 mils, 79 miles, it continued to give consistent paints on the CPS-4 out to 1230 miles, 87 miles (a 12-mile track, or a further 115 seconds at the extrapolated mean speed of this run).


   No purpose is served by invoking multiple-trip returns from beyond the unambiguous range. The pulse repetition frequency of each set must be identical in order for such echoes to correlate in range, and even then the question of the source of such echoes remains unanswered. The probability of ground returns due to anomalous propagation seems negligible for the reasons discussed, which leaves aerial targets.


   Aircraft beyond the unambiguous range of both sets could be displayed at spurious speeds and on distorted courses by multiple-trip echoes, but displayed speed will always be equal to or less than true speeds - never greater - and the echo could not appear to hover on both PPI and RHI scopes unless the target did, so that the required performance envelope for aircraft detected by multiple-trip echoes would be even more remarkable than it is already. Further, the echoes would be confined to the true azimuths, whereas in this case a somewhat con-sistent pattern of behavior was observed at scope azimuths differing by 180 degrees, which would imply chance detections of different multiple-trip aircraft at remote ranges which presented the illusion of a sequence of connected movements by a single object. This in itself seems highly improbable.


Satellites may sometimes be displayed in this way, but aside from objections already discussed none were orbiting in 1950. Speeds and courses are inappropriate for multiple-trip detection of meteor-wake ionization, even if the frequencies and peak powers of the sets had been appropriate. In general, a target which could give a scope presentation comparable to an F-80 even at second-trip ranges (maximum operating range of the set plus the displayed range of the echo) would have to be a much more efficient reflector than an F-80, due to the inverse 4th power signal attenuation of point targets, which implies a much larger aircraft of inferior performance. (Note that in the opinion of technical experts at Continental Air Command the reported echo presentation was comparable to that of a B-29 painted by first-trip echoes.) Finally, it should be remembered that the speeds and altitudes indicated would be performance minima for a multiple-trip target and the actual displayed performance is already difficult enough to account for.


   No fixed-wing or rotor aircraft known to be flying in 1950 could approach the performance envelope of the target, with a transition from extended hover at an altitude of about 45,000' (reportedly observed on both RHI and PPI scopes, ruling out the improbable hypothesis of a steep tangential climb or dive at constant slant range) to ground speeds of over 450 mph in a 6000 fpm climb. Notably, the highest ground speed coincides with the highest rate of climb. The earlier plotted speed of 1500 mph over a 93-mile track may have been inaccurate, as the controller suspected. But again note that even a factor-2 error here leaves a speed which is 25% better than the maximum clean speed in dive of an F-94 (602 mph), the fastest operational jet flying in 1950.


    The only hypothesis which, prima facie, seems at all attractive in this case is that of a "ghost" - a secondary echo from an aircraft due either to a ground reflector or another aircraft. Details seeming to hint at this possibility are found in the earlier phase of the event:

1) During the roughly 45-minute period from about 1945 to 2030 the target "seemed to stay in he area where our fighters were flying, sometimes approximating their courses"

2) At this time its altitude was about 20,000' above the F-80s, that is, at about twice their altitude of 24,000'

3) Although the echo was good it was at this time observed only intermittently

4) The target presentation was comparable to an F-80 but "if anything, narrower. It was definitely at this time not presenting a very large reflecting surface . . ."


Arguments against a ghost are rather strong, however. Firstly, all of this behavior occurred on the HRI scope of the AN/CPS-4 height-finder, which introduces special circumstances. To understand this, consider how a ghost is generated on the more familiar PPI of a surveillance scope such as the CPS-5.


 If an aircraft at 20,000' were so oriented as to scatter radar energy to an efficient ground reflector, the signal could return by the same path, being scattered back off the aircraft to the antenna. If this signal were strong enough to be displayed it could appear on a surveillance PPI as a weak echo on the same azimuth as the F-80 blip but at a greater slant range proportional to the total out-and-back path length - in this scenario the range would be that of the F-80 plus a minum of 20,000', being the minimum distance to a reflector on the ground. Generally such a ghost would be brief and weak, one or two blips appearing as the reflection geometry between aircraft and reflector approached optimum the  vanishing as the aircraft flew by and the geometry was quickly lost. In special circumstances involving an efficient corner-reflector, however, such as an empty metal truck or the corner formed between metal fences and wet ground, the reflection geometry could be maintained over some angular distance and the ghost could persist long enough to behave like a solid moving target. Since it would always be displayed on the aircraft azimuth at greater slant range, its movements would appear to relate generally to those of the aircraft, approaching and receding at higher speeds.


On such a PPI display no information about altitude or elevation is available. A height-finder operates differently, however: the antenna radiates a horizontal fan-shaped beam which is broad in azimuth but very narrow in elevation. The target altitude is derived from the slant range and the known elevation of the antenna at the time the reflected signal is received. This allows some simple calculation. With an hypothetical F-80 illuminated by the main beam so as to generate a ghost by secondary reflection the HRI antenna elevation corresponds to that of the F-80, and the apparent position of the ghost will be in the same "line of sight". In the present case the ghost is known to have been displayed at 47,000', range 70 miles, corresponding to an elevation of approximately 7 degrees. The maximum altitude of the highest F-80 being monitored at the time (checked by radio) was 24,000', which at 7 degrees elevation represents a slant range of about 35 miles. Therefore the minimum geometrically possible ray path to the ground reflector, via any of the F-80s at this time, is about twice the range from the antenna to the F-80 which (ex hypothesi) is concurrently being displayed on the scope.


Now a reflector of the kind required behaves as a specularly reflecting point target, and because returned signal intensity from point targets varies inversely with the 4th power of the distance we can show, assuming that the scattering efficiency of the F-80 is isotropic, that the maximum theoretical strength of the ghost echo due to a perfectly-oriented 100%-efficient ground reflector would be 1/16 or a little over 6% of the strength of the direct return from the F-80. In practice one would expect the aircraft aspect and the overall geometry to vary in respect to an imperfectly  oriented <100%-efficient reflector, making it likely that the mean signal would be even weaker, and the scope presentation fluctuating.


It should also be noted that this hypothetical ghost is being displayed at twice the range of the aircraft. A target at 70 miles might not be thought to be "staying in the area in which our fighters were flying . . . but 20,000 feet above them" if the range to the fighters is only 35 miles.


The consistency of the "ghost" model with the controller's report of a clear, well-defined target which was almost identical to the F-80s and in the same [implicitly geographical] area is rather poor.


(Note: The geometry underpinning the above argument assumes scattering of the main beam by an F-80, not scattering of side lobe radiation. In the latter case the indicated elevation and resultant displayed elevation of the ghost would not be that of the aircraft. However the attenuated signal reflected from a target in any lobe will be on the order of hundreds or thousands of times weaker than that from the same target illuminated near peak gain in the main beam. Moreover, an aircraft so placed as to generate a ghost due to any vertical lobing of the height-finder fan would still generate a far stronger ghost at the boresight position. Nothing is to be gained, therefore, by considering side lobe ghosts; indeed these worsen the theoretical fit, inasmuch as one would expect multiple ghosts as the antenna scanned in elevation.)


A further counterindication is the fact that, during this same period, the target appeared to hover in one spot. If accurate, such a circumstance cannot possibly occur with a moving aircraft and a ground reflector (whichever reflector is primary). Reflections between two aircraft could in principle achieve this, but it is highly improbable that the reflection geometry could be maintained more than very fleetingly. It is also highly improbable that an aircraft-to-aircraft ghost could persist as an extended coherent track for many minutes. The compound probability of these effects is vanishingly small. Any operator would be surprised to observe an air-to-air ghost reflection occur at any time. Let us assume that one such incident on a given shift is unlikely. Then how likely is it that an operator will observe such effects "intermittently but with increasing regularity for the next 45 minutes or an hour"? A long sequence of improbable ghosts which gave the impression of the track described, involving stationary episodes and culminating in a time-flagged 6-minute plot on HRI and PPI scopes "without fade"over a distance of some 50 miles with a final 2-minute "hover" is frankly so improbable as to be for all practical purposes impossible.


Finally, one must consider the failure of the CPS-5 to at first detect the target indicated by the CPS-4 at high altitude. This is not easily understood in terms of a ghost reflection from an F-80 since the F-80s were simultaneously being monitored at 24,000' and below on the CPS-5 by the controller, and the reflection geometry would be the same for both radars, as it obviously must be (ex hypothesi) to account for the subsequent simultaneous paints. On the other hand, this does, as the report of technical personnel, HQ, Continental Air Command, points out, "follow logic and field experience" with the CPS-5's known poorer coverage of real targets at high altitude. Again, at the end of the incident the target gave consistent paints on the CPS-4 after it had become intermittent on the CPS-5 at 79 miles, 33,000'. This behavior, too, can be rather easily understood in terms of null zones in the vertical diagram of the CPS-5 due to lobing, a particularly prominent effect at the short-centimetric wavelength of the CPS-5. Vertical lobing is caused by reflected ground-incident energy modifying the free-space pattern, a ubiquitous problem because of the need to fill the surveillance drum with radar energy down to the lowest possible elevations. The horizontal nodding fan of the CPS-4 radiates very little energy at negative elevation angles compared to the CPS-5 (the final boresight elevation of the target was approximately 6 degrees) and so would be relatively free of this sort of defect.


In conclusion there appears to be no satisfactory explanation of this incident. The simplest and most natural interpretation is of a radar-reflective aerial object capable of high subsonic and probably supersonic speeds and extended hover (or near-hover) at high altitude. An object of some size or unusual reflection efficiency is indicated by a radar cross-section comparable (according to cognizant technical specialists) to that of a medium-sized bomber. Aeronautical history does not record any such high-performance aircraft, and clearly none was known to USAF Continental Air Command in 1950. A currently unrecognized natural phenomenon is possible; however, it is fair to say that the behavior of the object in relation to the F-80s, staying in their area and "sometimes approximating their courses" but far above them, could also be interpreted as rational - or at least animate, possibly inquisitive - intelligence.


POSTSCRIPT: It is worth noting that the Blue Book final statistics carry only two "unknown" reports originating from Selfridge AFB for the entire period June 1947 to January 1969 (the present case being carried as "probable balloon"). One of these, Case # 650 (#203 in Brad Sparks' re-evaluated catalogue), was a nocturnal visual sighting of a vertically descending yellowish light which disappeared in level flight at high speed. The witness in this case was none other than 1st Lt. Frank Mattson, one of Controller Lt. Parker's radar crew on the night of March 9, 1950. That sighting took place on March 3, 1950. It isn't easy to conclude, however, that this fact is material to an interpretation of the radar observations just six days later.



4.  DATE: September 21, 1950        TIME: unknown                    CLASS: R  ground radar


LOCATION:                                    SOURCES: Hynek (1978) 139



                                                          RADAR DURATION: unspecified


EVALUATIONS: Blue Book - unknown


PRECIS: An MIT radar observer and 2 colleagues conducting a weather radar project under contract to the US Signal Corps were tracking Air Force F-86s from a radar site at Provincetown. The following is from a report made to Major Tuttle, Staff Weather Officers, 33rd Fighter Wing, Otis AFB, Mass.:


  An exceedingly puzzling event occurred during the 3rd run when the planes were heading northeast at 30,000 feet. We picked up another plane in the radar beam travelling about due north on a converging course towards the F-86s. It was moving very rapidly and I told the pilots about it, its range and direction from them. The echo caught up with, passed, and then crossed the course of the 86s, suddenly went into a very tight (for the speed) turn to the right, headed back toward Boston and passed directly over our flight. (Perhaps went under.) The sketch [unavailable] represents, as closely as we can remember, the relative positions of the two planes. Two other observers were with me at the time and we have checked over the facts rather closely. The pilots will undoubtedly recall the incident. They said they didn't see anything which is not too surprising considering the speed of the object and the fact that it may have passed several thousand feet above or below them and still looked like coincidence to the radar. Figuring conservatively, the speed of the object was approximately 1200 MPH, and the centrifugal force exerted on the ship during the turn amounted to something more than five g's. It gave an excellent radar echo which could not be mistaken for anything else and in all respects except for the velocity seemed a normal radar target. It passed out of the beam while we continued to track our flight, but we focussed on it again for a few seconds shortly after it was rapidly approaching Boston . . . . It was very evidently an interception of some sort on our flight, but what? The turn was utterly fantastic . . . . A few rough calculations concerning control surfaces, angles, etc., only adds to the puzzle that this object must have been entirely unconventional in many and basic respects. Perhaps the thing that bothers me the most is that it gave a very good radar echo, which implies irregular surfaces and comparatively large size, large enough so the pilots might have had a good chance to see it. . . It seems highly probable that I may be poking into something that is none of my business, but on the other hand, it may be something that the Air Force would like to know about if it doesn't already. . . .


 NOTES: The description seems to be of a PPI display with no height finder. It appears that the aircraft altitude cited came from pilot reports.


   This kind of target behavior - a single discrete target, presentation identical to that of an aircraft, making a continuous track at very high speed with a midcourse turn onto a markedly different azimuth - is not at all typical of anomalous propagation. Partial reflection from moving waves on an elevated inversion could generate fairly good spot targets, but the speed of such tracks is 2 x wind speed with some consistent relationship to wind direction: in this case the target was too fast by an order of magnitude and the rough geography of the account is sufficient to establish a change of heading too great to comfortably equate with winds. Sporadic ground echoes due to superrefractivity can present the illusion of fast targets on any heading, but the probability of a random mechanism generating a coherent track of any length is very low, as is the probability that a number of sporadic returns from very different ground reflectors over a very wide area, each with a very different propagation history, could present a) consistently and b) as "very good" spot targets. Neither CAT nor any other possible atmospheric inhomogeneity seems appropriate to the target presentation, speed or behavior.


   Radio frequency interference or spurious internal signals are equally improbable explanations of a target which a) "in all respects" except speed had the presentation of an aircraft target (requiring discrete pulse trains with a systematic reference to the PRF and scan rate of the receiver, such as signals from another similar radar), and b) made a complex PPI track with no obvious geometric relationship to the scope center (implying a source with no systemmatic reference to the scan-rate of the receiver). A so-called spot target such as an aircraft is displayed on a PPI as a short arc composed of the integrated spot returns from a number of radar pulses on adjacent scope traces, whose non-radial movements bear a very complex relationship to a changing set of trace radii; and a repeating noise pulse or pulse-train (even neglecting the complexity of the target arc) which displayed as a track comprising at least one straight sector with a significant vector tangential to the scope radius, plus a "very tight turn", would in itself imply a very complex and fortuitous pattern of changing pulse-repetition frequencies. Such varying cyclicity is not typical of any pulsed emitter in the radar band, and the likelihood of random noise generating such a target is obviously vanishingly small.


   Birds, insects, balloons and other wind borne objects are ruled out for reasons including speeds, headings and presentation. Multiple trip returns from aircraft beyond the unambiguous range are not helpful in this case, since moving targets so detected will be displayed at spuriously slow speed proportional to their tangential vector, and the displayed speed is already a problem; and AP ground returns detected in this way are subject to the same objections already discussed.


   The date of the incident rules out the possibility of multiple-trip returns from an artificial satellite. But one possibility which deserves to be considered is multiple-trip returns from meteor-wake ionisation: a large meteor on a tangential heading to the receiver at low elevation might be detected by a sensitive search radar circa 1950, although neither the peak output nor the frequency are known. The track of such a meteor might be considerably distorted and appear to execute an approach towards the site and a hyperbolic turn onto a receding heading. Due to the rate of ion recombination the trail will be scanned  essentially as a spot target. But there are objections to this hypothesis: 1) the geography and headings cited cannot be made to fit such a trajectory; 2) a meteor trail is unlikely to yield an "excellent echo" indistinguishable from that of an aircraft even within the unambiguous range of a radar operating at optimum frequency, still less at ranges displayed by multiple trip; 3) displayed speed of 1200 mph is far too low even for this mechanism; 4) the duration of the track (sufficient to allow the operator time to contact the F-86s and supply range and heading of the target before it even approached them) is clearly too great for a meteor which would only be within the earth's atmosphere for a few seconds.


   A "ghost" reflection is on the face of it the most plausible explanation. Qualitatively speaking, such apparent "interceptions" by high-speed targets are quite typical of ghosts due to radar energy being scattered from the aircraft to an efficient secondary reflector and back again. The most efficient secondary reflectors are large ground structures, such as metal roofs, and corner-reflectors such as empty trucks and metal fencing. The displayed range of a ghost will exceed that of the primary reflector by an amount equal to the distance between the two reflectors, and the ghost will always appear on the azimuth of the primary reflector. Since the angular rate of the ghost equals that of the primary reflector but at greater displayed range it can appear to move at much higher speeds. In this way, an aircraft flying over such a reflector might be "intercepted" by a ghost which approaches at high speed, merges with or paces the aircraft target briefly and then accelerates away again. In the present case, however, the aircraft were flying at 30,000' so that any ghost due to a ground reflector could not be displayed closer than about 5.7 miles from the F-86s.


   For aircraft and ghost to coincide on the PPI the secondary reflector would have to be another nearby aircraft. Since there were apparently two F-86s the possibility arises of reflection from one to the other, but the target was initially painted at considerable range from the two F-86s (far enough that it could be seen closing at 1200 mph whilst the radar operator talked to the pilots about it) whilst the aircraft were evidently flying together on the same heading and both at 30,000. Later the target "passed out of the beam while we continued to track our flight." Finally, it can be shown that if the F-86s were on a NE heading and the "intercepting" ghost was on a N heading then the reflection geometry requires the F-86s to be SE of the radar site, in which case a ghost could not possibly be displayed turning "back toward Boston" (NW of Provincetown) as this would place it at closer displayed range than the primary reflector (an F-86).


   Another unreported aircraft would be required as a secondary reflector, but the coincidence of ghost and F-86 echoes would still demand that it flew by in close physical proximity, which raises the questions of why it was not seen and, more importantly, why it was not itself painted by the radar, being (ex hypothesi) an efficient enough reflector to generate a "very good" ghost echo by secondary scattering.


   The only possible explanation would seem to be a ghost due in no way the F86s but entirely to another unreported aircraft and a secondary ground reflector. The collocation of the ghost and the F-86s on the PPI would thus be completely coincidental, and a remote primary reflector would explain why it was not seen visually by the pilots. The aircraft would probably have been a fast jet for its ghost to achieve 1200 mph and would be required to maintain a very favourable aspect with an efficient ground reflector for a period evidently of some minutes, initially producing a ghost track on an essentially straight N heading at 1200 mph with a sufficient duration to permit confirmation of its speed and heading over a number of sweeps followed by radio contact with the F-86 pilots, then turning abruptly as the culpable aircraft flew by the ground reflector. This trajectory simply must represent a ground track of some tens of miles, all the while presenting an "extremely good" ghost return. This is inherently somewhat improbable, and at no time is the presence of the responsible aircraft mentioned despite the fact that when its ghost track turned abruptly onto a heading for Boston it would itself have been displayed on-scope very close to it, and thus very close to the F-86s, its relationship to its ghost then, if at no other time, becoming at least noteworthy - if not actually understood by the operators.


   In summary, although there is no visual corroboration of the target in this case and only one radar instrument was involved, the target as described is not easily explained. However, in the absence of scope photographs a ghost echo remains a possible, albeit a not very satisfactory, interpretation.


STATUS: Insufficient information



5.  DATE: February 2, 1952                TIME: night                     CLASS: R/V   shipboard
                                                                                                                    radar/deck visual

LOCATION:                                       SOURCES: Hynek (1978) 126

Sea of Japan

Near S. Korean coast

                                                            RADAR DURATION: approx. 10 minutes


EVALUATIONS: Blue Book - unknown


PRECIS: The aircraft carrier USS Philippine Sea was making 13 knots on a 180-degree (S) heading off the E coast of Korea. A single target was detected on radar at 0 degrees (N) range 25 miles approaching the ship. At range 20 miles the target began a wide turn to the E, radius 11-12 miles. The radar operator queried the aircraft controller for an ID of the target. At this point three signal observers on deck independently reported visual sightings to the bridge of an object or objects resembling three "exhaust flames" at 30 degrees elevation and turning away from the ship at the same azimuth as the radar target, which at this time  was shown at an  altitude  of  52,000',  range  17  miles and increasing. The aircraft controller determined that the target was not a friendly aircraft, and the general alarm was sounded. The target continued to turn away onto a heading approximately NW until it returned to a 0-degree bearing from the carrier,  range  now  55  miles,  where it split and opened as  2 contacts several miles apart.  At this point the radar operator began to  make "careful plots, checking constantly" for 3 minutes while the 2 targets headed N abreast on a slightly  zigzag  course,  fading  from  the  scope  at  range 110  miles.  During  this period the targets accelerated:


"Measured speed 10 miles per minute (600 mph) for first minute, 15 miles per minute (900 mph) for second minute, 30 miles per minute (1800 mph) for third minute."


NOTES: According to Hynek's commentary on the intelligence report: "During the time it was in view, the coast of Korea and the island of Ullung Do were visible at a distance of twenty miles, and an escorting destroyer was visible on the scope 2,000 yards from the carrier." Presumably this means that the coast and island were both seen on radar, rather than visually, although this is unclear. But it should be noted that Ullung Do is some 80 miles from the nearest point on the coast of South Korea at Ulchin, so that at no time could the ship have been 20 miles from both. However this inconsistency is probably an introduced error,  and  should  not  be  taken  to  signify  any  extraordinary  propagation conditions.


The radarscope map of the incident is consistent with all speeds, ranges and azimuths cited. It should also be emphasised that independent visual observers estimated the elevation of the "exhaust flames" as 30 degrees when turning past the ship, whilst radar at this point showed the target at 17 miles (slant) range, altitude 52,000', or almost exactly 30 degrees. It is worth noting that at this point it would be irrational to conclude that the target was anything other than aircraft, and if the incident had ended there no "UFO" report would have been submitted. Even the separation of the target would not of itself challenge this conclusion, since more than one aircraft flying within the resolution cell could be displayed as  a single echo, and the three  "exhaust flames" could be said to confirm this interpretation.


However, the rather careful plotting of speeds up to 1800 mph instantly changes the complexion of the incident, and once it has become a "UFO report" it  invites  us  to  attempt  all  manner  of contorted  interpretations.  The  question  in a case such as this is whether the a priori improbability of a "UFO" (whatever that might mean) outweighs the prima facie probability that a target consistently tracked on radar and confirmed visually was physically present. This question is not answerable in practice. Suffice it to say that the radar incident has no easy explanation in terms of super-refractive AP of sea clutter or surface ships, partial inversion reflection,  CAT,  birds,  insects, balloons  or other  wind borne  objects, side lobe  returns,  multiple-trip  returns  from  targets  beyond  the  unambiguous range, spurious internal signals or RFI. A "ghost" echo from an efficient surface reflector (say, the angle made by the side of a ship with the sea - a destroyer was in the vicinity) received via an aircraft as primary reflector might achieve the speeds reported, but: a) no such aircraft, which must ex hypothesi have been within the  radiation pattern,  was separately detected,  and no known aircraft were reported to be in the area; and b) a total radar duration on the order of 10 minutes and a track painted continuously to a range of 110 miles makes any ghost reflection involving a ship only ?  mile away extremely improbable. (A ghost reflection with the ship as primary reflector could not achieve the >45 degrees change in azimuth reported, being displayed always on the ship's true azimuth.)


The target behavior appears to have been rational and consistent, without erratic jumps, disappearances or course-reversals, executing a smooth turn with a radius of some 12 miles in the manner of an aircraft, though at somewhat high altitude.  Together with multiple  independent  visual  sightings  at  a  position consistent with the concurrent radar target, it is reasonable to conclude that the incident was most probably caused by a real radar-reflective object or objects emitting light resembling that from an aircraft exhaust. However no aircraft in 1952 was remotely capable of speeds up to 1800 mph, and this is the crux of the case: if the displayed speed can be explained then the object(s)  can be explained as probable aircraft.


The intelligence report has this to say, pertinent to the possibility of operator error:


A thorough  debriefing  was made  of the  radar operator. Personnel stated  that   the  operator   was  very   intelligent,  efficient   and cooperative. Operator was cognizant of capabilities and limitations of the radar equipment . . . . The three minutes of careful plotting were made after the object had turned and was heading away from the station. Operator was sure of the accuracy of the plots for the three minutes, and was adamant that the speeds shown were approximately correct.


The likelihood of gross malfunction in the radar display(s) would appear to be minimised by  the very  close  match  with the  visual azimuth and elevation reported as the target turned past the ship. It is conceivable, however, that there was a range-scale error on the PPI. If, for example, a selected range-scale of 100 miles was actually giving a 30-mile scale due to electromechanical fault, and if the operator failed to notice the error, then there would be a factor-three error in plotted ranges, leading to a true maximum speed on the order of 600mph during the final 60 seconds of the track - roughly the maximum speed of,  say,  an  F-94  in  a  powered  descent.  Angular position  and  motion  would  of course be preserved.


This hypothesis is highly speculative and, it must be said, not particularly plausible. It is first of all unlikely that an operator with the stated experience would neglect to notice so basic a fault during or after the ten minutes or so of the event whilst making "careful plots, checking constantly", evidently cognisant of the abnormal nature of the track, and the (presumably known) position of the escort destroyer - concurrently displayed at 2000 yards from the carrier - would have been an unmissable and constant reference datum on-scope long before and long after the appearance of the target, not to mention the coasts of Korea and Ullong Do island. (The aforementioned ambiguity in the stated range of these land echoes is unfortunately not relevant to this hypothesis, since: a) if there is an original error here it is in the wrong direction - that is, the displayed range would be exaggerated by 3 to 120 miles, not reduced to 20 miles; and b) it is anyway much more likely that this is a constructional error in Hynek's case summary, the figure of 20 miles referring either to the mainland or to Ullong Do, clearly not to both.)


In conclusion there appears to be no reasonable explanation of the object(s) observed.


STATUS: Unknown



6.  DATE: August 5/6, 1952         TIME: 2330 local                      CLASS: R/V  ground-air
                                                                                                             radar/ ground visual

LOCATION:                               SOURCE:  Thayer, Condon 1970 123

Haneda AFB                                                  McDonald, Sagan/Page 1972 90

Nr. Tokyo                                                      Shough UFO 47-87 217


                                                     RADAR DURATION: 40 mins. approx.


EVALUATION:  Blue Book - unknown       

                            Thayer - AP/Capella

                            McDonald - unknown


PRECIS:  At about 2330 local time on the night of August 5, two control tower operators were walking across the ramp towards the tower to begin the midnight shift at Haneda AFB, adjacent to Tokyo International Airport on the W side of Tokyo Bay. The night was generally clear with bright moonlight (full moon near the meridian), some scattered thin clouds and excellent visibility. The airmen observed an unfamiliar brilliant light in the NE sky which, as they looked at it, disappeared for about 15 seconds and then "returned to approximately the same spot". Puzzled, they went straight to the tower where they indicated the light to the on-duty controllers, who had not noticed it because "the operating load had been keeping their attention elsewhere" according to Air Intelligence Information Report IR-35-52. The four controllers began to observe the light, which was variously described as "exceptionally bright", "intense", "blinding" and "brilliant blue-white". Naked eye comparison was made with a bright celestial body low in the E which the observers believed to be Venus (Venus had set 3 hours earlier - it was in fact Jupiter, which rose almost due E about 2300 with an apparent magnitude of -2.0): the object was significantly brighter and had a naked-eye angular subtense estimated (on the basis of later comparison with a pilot balloon at known range) of about 3 arc minutes. Observed through 7 x 50 tower binoculars the light appeared as an extended disc of uniform brilliance, specifically "not due to a point source of light", inside a "round dark shape" along the curved bottom edge of which were four distinct lights, with "a glare around the whole thing". IR-35-52 states:


The light was described as circular in shape, with brilliance appearing to be constant across the face. The light appeared to be a portion of a large round dark shape which was about four times the diameter of the light. When the object was close enough for details to be seen [the changes in apparent size, magnitude and position are discussed later], a smaller, less brilliant light could be seen at the lower left hand edge, with two or three more dim lights running in a curved line along the rest of the lower edge of the dark shape. Only the lower portion of the darker shape could be determined, due to the lighter sky which was believed to have blended with the upper side of the object. No rotation was noticed. No sound was heard.


   After a few minutes the controllers called the 528th Aircraft Control & Warning Group's Ground Controlled Intercept (GCI) radar station at Shiroi, about 20 miles NE of Haneda near Shimofusa, above the N end of Tokyo Bay, requesting a search for a low altitude target somewhere to the NE of Haneda. (Visual estimates of the altitude of the object varied. The intelligence report states 1000'-1500' at one point, but includes an estimate of 5000' elsewhere. It is possible that this discrepancy is due either to the inevitable subjectivity of  such judgements or to the fact that the source was reported to have ascended somewhat in elevation by the end of the sighting. The 5000' estimate may relate to a later phase of the observation.)


   At about 2335 the Shiroi duty controller began a search of the indicated area towards the N end of the bay on the PPI of the CPS-1 10cm surveillance radar. There was nothing on the high beam; but when he tried the low beam he found "three or four blips" which were "at a position 050 degrees bearing from Haneda, as reported by the tower", although the exact location and disposition of these targets is not stated. "No definite movement could be ascertained." The controller attempted to acquire these targets on the CPS-4 heightfinder, but was unable to do so, apparently due to excessive ground-clutter on the CPS-4 at so low an elevation. At 2345 the relief controller arrived for duty, and the two 1st Lieutenants discussed the desirability of scrambling an interceptor.


   Meanwhile back at Haneda the tower operators were still watching the light when, a few minutes later, a call came in from Tachikawa AFB, 21 miles WNW of Haneda, independently reporting a light over the bay. According to Air Intelligence Information Report IR-35-52: "The control tower at Tachikawa Air Force Base called Haneda tower at approximately 2350 to bring their attention to a brilliant white light over Tokyo Bay. The tower replied that it had been in view for some time and that it was being checked." McDonald has pointed out that a bearing "over Tokyo Bay" from Tachikawa would be significantly S of E, whereas the bearing of the light as seen from Haneda was NE, and these two lines of sight would indeed intersect over the N part of Tokyo Bay.


   By this time the Shiroi GCI controllers had decided to scramble an aircraft, and at 2355 called in their request to the ADCC flight controller at Johnson AFB, 35 miles W of Shiroi. Five minutes later, at about 0000-01, they had acquired a definite contact on the 50-mile high beam of the CPS-1. According to the statement of the relief controller:


At the time of the scramble, I had what was believed to be the object in radar contact. The radar sighting indicated the object to be due south of this station over Tokyo Bay and approximately eight (8) miles northeast of Haneda. The target was in a right orbit moving at varying speeds. It was impossible to [accurately] estimate speed due to the short distance and times involved.


   A number of visual observers at the radar site attempted to locate the object in the sky to the S of Shiroi but reported seeing nothing. At Johnson AFB, meanwhile, a two-seat F-94B of the 339th Fighter-Interceptor Squadron was experiencing delay due to fuel-system trouble. When it finally got airborne the time was 0003-04, and it reached the Tokyo Bay area several minutes later. By this time the CPS-1 target had been tracked through one approximately circular right orbit of radius 4 miles, about half over the sea and half over land. It was a distinct target over the sea, but was lost for part of the track due to the ground-clutter pattern near the coast. When the F-94 first approached the area the target was still lost and GCI could not supply effective assistance, so the aircraft headed down over the bay in a blind search.


   At this time, about 0009-10, the Shiroi CPS-1 displayed a target to the starboard (W) of the F-94 and the controller alerted the pilot, who was easily able to visually identify it as a C-54 in a landing pattern near Haneda. With no  unknown target showing on the CPS-1, the controller recalled the F-94 to the N area of the bay and instructed a visual search at an altitude of 5000'.


   By about 0012 what appeared to be the same unidentified target had reemerged from the clutter and begun a second similar manoeuvre, described by the controller as "a starboard orbit in the same area as before," but accelerating now to a higher speed. These manoeuvres are described in the intelligence summary as follows:


An F-94 was scrambled to investigate. The object at this time had left the ground clutter and could be tracked (on the CPS-1) at varying speeds in a right orbit. Although impossible to accurately estimate speed, Lt. ----- gave a rough estimate of 100-150 knots, stopping, and hovering occasionally, and a maximum speed during the second orbit (just before the F-94 was vectored in) of possibly 250-300 knots.


   During this second orbit at about 0012 the target broke into three, and the F-94 was directed towards the strongest target on a vector of 320 degrees. The intelligence summary continues:


At 0012 the object reportedly broke into three smaller contacts, maintaining an interval of about mile, with one contact remaining somewhat brighter. The F-94 was vectored on this object, reporting weak contact at 0015 and loss of contact at 0018. Within a few seconds, both the F-94 and the object entered the ground clutter and were not seen again.


The times of AI contact here cited are evidently in error. This paragraph was prepared on the basis of the Shiroi controller's report which, according to the preparing officer Captain Charles Malven, contained erroneous times at this point due to "typographical error". Malven here attempts to reconstruct the times into those consistently reported by other participants, but the time of acquisition of the AI radar contact, and probably the reported 3 minute duration of contact, are still in error by a minute or so. The error is not substantive, however. The signed report of the F-94B pilot states:


The object was reported [by Shiroi GCI] to be in the Tokyo Bay area in an orbit to the starboard at an estimated altitude of 5,000 feet. I observed [visually] nothing of an unusual nature in this area; however, at 0016 when vectored by Hi-Jinx [GCI] on a heading of 320 degrees, and directed to look for a bogie at 1100 o'clock, 4 miles, Lt. ---- made radar contact at 10 degrees port, 6000 yards. The point moved rapidly from port to starboard and disappeared from the scope. I had no visual contact with the target.


The radar operator's statement confirms and amplifies this account:


At 0015 Hi-Jinx gave us a vector of 320 degrees. Hi-Jinx had a definite radar echo and gave us the vector to intercept the unidentified target. Hi-Jinx estimated the target to be at 11 o'clock to us at a range of 4 miles. At 0016 I picked up the radar contact at 10 degrees port, 10 degrees below at 6000 yards. The target was rapidly moving from port to starboard and a "lock on" could not be accomplished. A turn to the starboard was instigated to intercept target which disappeared on scope in approximately 90 seconds. No visual contact was made with the unidentified target. We continued our search over Tokyo Bay under Hi-Jinx control. At 0033 Hi-Jinx released us from scrambled mission.


   Whilst this activity was going on, the luminous object was still under observation from the tower at Haneda (no information is available about further visual observations from Tachikawa AFB during this time). No exact times are given, but two episodes were observed of apparent dimming, apparent coincident diminution in angular diameter and apparent coincident lateral motion. IR-35-52 states:


The object faded twice to the East, then returned. Observers were uncertain whether disappearance was due to a dimming of the lights, rotation of object, or to the object moving away at terrific speed, since at times of fading the object was difficult to follow closely, except as a small light. Observers did agree that when close, the object did appear to move horizontally, varying apparent position and speed slightly.


One of the controllers stated:


I watched it disappear twice through the glasses. It seemed to travel to the East and gaining altitude at a very fast speed, much faster than any jet. Every time it disappeared it returned again, except for the last time when the jets were around. It seemed to know they were there.


   The time of the object's final visual disappearance is not reported exactly, but appears to have been no more than a few minutes after the ground- and airborne-radar target(s) had been lost. Thayer's reading of the case file indicates about 5 minutes, but since he reports a total duration of "about an hour" this may be uncertain. McDonald gives 0020 as the time of visual loss after a duration of "about 50 minutes", which may be equally uncertain. 0020 would be quite close to the time of AI radar signal loss at about 0018. The fact that visual disappearance was coincident with the presence of the jets is indicated by the Haneda controller's statement above, but it cannot be established that radar and visual disappearances were simultaneous.



NOTES: A lighted pilot balloon released at 2400 from the weather station, 2000' from the Haneda tower, allowed the observers to scale both the brightness and naked-eye angular subtense of the source against the known 1.5 candle luminosity of the balloon lamp and the balloon's 24" diameter. The angular size of the object "when closest to the tower" was approximately the same as the balloon, about 3 minutes of arc, and "the balloon's light was described as extremely dim and yellow, when compared to the brilliant blue-white light of the object." Another real-time comparison was made between the appearance of the unidentified light when at its faintest and the planet Jupiter: "At the greatest distance, the size of the light appeared slightly larger than Venus [Jupiter], approximately due East of Haneda, and slightly brighter." The implication here is of a very bright source if its minimum visual magnitude exceeded Jupiter's -2.0, and McDonald points out that the balloon light would have an apparent visual magnitude of approximately -0.5 at 2000' (about twice as bright as Capella, at  magnitude +0.21 the brightest star visible in the NE quadrant) yet appeared "extremely dim" in comparison with the object. There is a rough but persuasive consistency in these estimates which lends credence to the subjective impressions described by the observers, of an "intense", "brilliant" or "blinding" light, and the quasi-physiological evidence reported in the intelligence summary to the effect that "observers stated that their eyes would fatigue rapidly when they attempted to concentrate their vision on the object." McDonald concludes that the object must have had a visual magnitude probably well in excess of -3.0 when at its brightest/closest, and there seems no good reason not to accept this as a reasonable approximation. (For comparison, the planet Venus at maximum brilliance attains magnitude -4.28.) In any event it is fair to conclude that it was at least bright enough to appear highly unusual.


   The approximate azimuth of the object when first seen appears to be in no doubt. It was clearly stated by observers, and by the preparing intelligence officer, to have been NE or, in one estimate, NNE. The initial radar targets detected at Shiroi were "on a bearing 50 degrees from Haneda, as reported by the tower." NE appears to be a fair approximation. However there were no significantly bright celestial bodies in that part of the sky, and indeed none exceeding or even approaching the magnitude of Jupiter anywhere in the sky (Venus having long set). Thayer suggests that the relatively faint first-magnitude star Capella (magnitude 0.21, 37 degrees azimuth, 8 degrees elevation at midnight) was "the most likely light source".


   Given the conjectural atmospheric mechanism by which he proposes that Capella produced the phenomena observed, one can reasonably quarrel with the assertion that this star is "the most likely" source, but there are certainly some arguments in favour of an astronomical explanation. Notably, one of the Haneda observers reported that the object appeared to be higher in the sky at the end of the event, and further that the gain in elevation appeared to be roughly proportional to that of the moon; also the pilot of a C-54 in a landing pattern at Haneda was requested to observe the object, and replied that it looked like "a brilliant star", which was what he assumed it was. Given that the C-54 pilot was observing the same object, and not Jupiter (how precisely he was directed to the right bearing and what his heading was at the time are not known), then its more ordinary appearance when seen from an aircraft at higher altitude would be circumstantial evidence for a propagation anomaly local to, or restricted to the line of sight of, observers on the ground at Haneda. The failure of observers at Shiroi to see anything in the S sky from their position is also consistent with a source at much greater - possibly astronomical - distance from Haneda, as is the failure of the intercepting aircrew to see anything (other than the inbound C-54) in the bay area. However, these persuasive indications are counterbalanced by other arguments, as we will see, and make no more than a circumstantial case without a plausible explanation for the observational details carefully recorded and sketched by experienced tower observers over a period approaching one hour.


   Thayer's attempt to account for the reported visual display proposes either or both of two hypothetical mechanisms: 1) a sharp temperature inversion at the boundary between moist air over the bay and an overlying drier airflow, with collected patches of mist or fog at the boundary generating an annular diffraction pattern or corona around an image of Capella; and/or 2) effects due to (a) the focussing of wavefronts along the upper boundary of the refracting layer, producing a region of locally enhanced brightness (so-called "Raman  brightening" after the wave-optical theory of mirage advanced by Sir C. V. Raman in 1959) and/or (b) interference of wavefronts, causing the appearance of dark and bright bands as also found (theoretically and experimentally) by Raman. Thayer calculates that a diffraction corona with a dark aureole of about the size observed might be produced by a mist of 200-micron droplets.


   Thayer admits that this explanation is conjectural. Weather data are lacking to substantiate the existence of an inversion, and the additional mechanisms proposed combine to represent "a phenomenon which must be quite rare." Indeed there must be considerable doubt that phenomena due to focussing and, especially, interference of wavefronts have ever been observed at all in the free atmosphere (see: Viezee, "Optical Mirage" in Condon 1970 637, 650, 653) and there is further doubt that, if they can be observed, they could create the luminous display reported from Haneda. Considering these effects first: the predicted Raman brightening arises when plane-parallel rays from a very distant source pass into a relatively thick inversion and are incident on the upper boundary near the critical angle (0.5 degrees) for total reflection. The result is a layer near the top of the inversion, narrow with respect to the diameter of the incident beam, within which there is continual crossing of incident and reflected rays. Because this layer is narrower than the beam, whilst the luminous flux per unit time (neglecting absorption) is preserved, the energy density within the interference layer is greater and hence the brightness of the beam is enhanced for an observer viewing the layer close to the critical angle for total reflection. The crucial phrase here is "the critical angle", since the light must be both incident and viewed at close to 0.5 degrees, which means that the observer must actually be in the inversion layer near its upper boundary, viewing a source at a real relative elevation very near zero. This is why observations of this effect in nature, if they occur, must be extremely rare, and the conditions are only likely to be met briefly by an observer viewing an astronomical source from an aircraft.


  In the present case the "source" - Capella - was at a mean elevation of about 8 degrees (2400) during the 50-60 minutes of observation and the observers were (to all intents and purposes) on the ground. Raman brightening could conceivably occur in ground-based observations of a source on the horizon, i.e. rising. But Capella would traverse 15 degrees of Right Ascension during 1 hour, only a proportion of which would be gain in terrestrial elevation (due to the inclination of the celestial equator); that is, the minimum elevation of Capella at 2330, when the object was first seen, must already have been greater than 0.5 degrees. Therefore, Raman brightening would require a highly stable surface inversion initially canted upwards to the NE at somewhat more than 0.5 degrees, gradually increasing its inclination in synchrony with the rotation of the earth over the next 50-60 minutes to an eventual angle of somewhat less than 15 degrees, its upper boundary anchored at no more than a few tens of feet at Haneda and all the while maintaining alignment with the line of sight from the tower to Capella (with only a couple of brief decouplings to allow the transient fadings of the source) to within an accuracy of about 30 arc minutes. This is so unlikely a phenomenon as to be meteorologically preposterous. Furthermore, Raman brightening would not produce enlargement of the stellar image to a disc with perceptible angular extension; rather it would "focus" the available light from what is already effectively a point-source and compress the image.


   Banding effects due to constructive and destructive interference of wavefronts within a refracting layer are not known to have been observed in nature, but  would be ephemeral and rapidly changing phenomena unlike the stable display observed at Haneda. In particular, a circular annulus could not be seen along a lateral refracting layer. Raman's experiment involved a collimated beam incident on a heated plate to simulate an inversion layer, the result being a dark zone of total reflection adjacent to the plate followed by a layer of enhanced brightness and then further alternating light and dark bands diminishing with distance from the plate. It is evident that an annular pattern of banding would require variations of refractive index to be symmetrical about the line of sight, a phenomenon that is difficult to conceive. Furthermore such interference effects are only observable at the critical angle within the layer and are therefore subject to the same objections raised above.


   The mist/diffraction-corona hypothesis seems less implausible. However, McDonald has criticised Thayer's model on a quantitative basis. He points out that the proposed droplet size of 200 microns is never found in mist but only in raining or drizzling clouds. Maximum droplet diameters of 10-20 microns would occur in hypothetical mist patches under the prevailing conditions, he argues, which would lead to coronal diameters of between 30 and 60 milliradians (100- 200 arc minutes) - several times the size of the full moon and more than an order of magnitude larger than the annular display observed. (It can be added that if rain coud had been present in the line of sight - they were not - it is somewhat improbable that Capella's modest magnitude of 0.21 would be visible at all, let alone as the brilliant centre of a diffraction corona, even viewed with binoculars.) McDonald also points out that a corona with an inner dark annulus several times the diameter of the central luminary equates to nothing in the literature of meteorological optics, and further that the smaller lights visible along the lower periphery of the annulus remain unexplained by any optical model. (It might be noted here that there are three fainter 4th magnitude stars known as The Kids lying near Capella, which could just appear within the field of typical night glasses at the same time as Capella; but these are fully 5 degrees away, clustered to the lower right of Capella, and form no sort of arc. Moreover there is no star conforming to the somewhat brighter light seen lowerleft: Beta and Theta Aurigae are both more than 10 degrees away. There are no stars at all brighter than about magnitude +7.0 - on the order of a thousand times fainter than Capella - within several degrees in the sector below Capella, and thus nothing that corresponds to an arc of "distinct" lights with a radius on the order of 10 arc minutes.)


   The vexed issue of the reported movements of the object is interesting, and McDonald calls this "the single most important ambiguity in the case file." He concedes that the object was probably reported to have moved in azimuth but (a little inconsistently) concludes that this movement can have had no relation to the movements of the radar target, and that there was never at any time radar-visual simultaneity. Yet the tower operators statements describe the light initially stationary in the NE, then travelling "to the east" with an apparent gain in range so that "it became difficult to follow closely, except as a small light", then repeating this movement to the E and back before disappearing "when the jets were around". One controller states: "I watched it disappear twice through the glasses. It seemed to travel to the East and gaining altitude at a very fast speed". When the object "returned" from each of these excursions and was "close", the observers each agreed that it appeared "to move horizontally, varying apparent position and speed slightly." Granted that these statements are not as clear as they might be, nevertheless phrases such as "travel to the East" from a position NE and "difficult to follow closely" at least arguably imply non radial motion; and the fact that this motion was observed twice is rather easily related (at least qualitatively) to the two orbits made by the radar target. McDonald argues that no movements were observed large enough to equate to the 4-mile-diameter radar orbits over the bay, but motion from the NE "to the E" could imply anything up to 45 degrees and is not necessarily inconsistent with the 30-degree subtense of those orbits at Haneda. Further, it has to be remembered that the radar's PPI displays not ground range, but slant range, so that the ground track of a "roughly circular" inclined orbit, or a more complex figure with changing altitudes whose varying slant range approximated a roughly circular PPI track, could be an ellipse or other extended figure with a significantly smaller azimuth change as viewed from Haneda (bearing about 130 degrees W of Shiroi as measured from the rough centre of the radar track). This would reduce the apparent lateral motion, and it is noteworthy that when the object was seen to move from Haneda it did so by "travelling to the East and gaining altitude".


   Departures in elevation and azimuth of more than a fraction of a degree would be quite inconsistent with any mirage image of Capella, and the independent line of sight to the similar object reported from Tachikawa would, as McDonald emphasises, be more than 45 degrees away from Capella. It was, however, only a few degrees away from Jupiter which was bright in the east, and the possibility has to be considered that this sighting was of Jupiter. In view of the scant information from Tachikawa no real analysis is possible, but one can only follow McDonald in feeling that probability does not favour this hypothesis, especially given that Jupiter had been rising in the east for many days. What indeed is the likelihood of two independent simultaneous misinterpretations of two different astronomical bodies, when combined with subsequent multiple radar contact with a target appearing to correlate with motions observed visually and detected in the very area triangulated by these two lines of sight?


   The radar contacts are clearly central to the case. Thayer's diagnosis of anomalous propagation is inferred from four central symptoms: 1) a "tendency" for targets to disappear and reappear; 2) a "tendency" for targets to break up into smaller targets; 3) a lack of correlation between ground and airborne radar targets; and 4) the fact that nothing was seen visually from Shiroi or from the interceptor despite visibility which the aircrew rated as "exceptionally good". The first three of these points are all arguable, and again one must admit the weight of McDonald's objections.


   Firstly he objects that there was no "tendency" for targets to appear intermittently or to break up - i.e., the echoes were in no way described as being unstable or fugitive. IR-35-52 states:


Lt. ----, GCI Controller at the Shiroi GCI site, has had considerable experience under all conditions and thoroughly understands the capabilities of the CPS-1 radar. His statement was that the object was a bonafide moving target, though somewhat weaker than that normally obtained from a single jet fighter.


This target, though small, was reported as sharply defined. Its appearances and disappearances were evidently due to its passage through the CPS-1 ground clutter pattern, as charted on maps of the radar plot and indicated in several written passages. When away from the clutter it was distinct and had no tendency to disappear or fluctuate in the manner of an unstable AP echo.


     Similarly, it is unreasonable to describe the single episode of fission as a "tendency for the target to break up into smaller targets": this occurred once on the second orbit, and the three fission products were described as small and relatively weak, but still sharply defined, "maintaining an interval of about mile, with one contact remaining somewhat brighter. The F-94 was vectored on this [brighter] object." And the nub of the issue is in this vectoring and its consequences: Thayer regards the timing discrepancy in the Shiroi controller's report as a basis for doubting that "the GCI radar ever tracked the fast-moving target described by the F-94 crew", despite the specific statement in the intelligence summary that this was due to a "typographical error", and despite the clear contextual evidence. The controller stated (times omitted to avoid confusion):


Sun Dial 20 [the F-94's codename] was vectored to the target. He reported contact at . . . and reported losing contact at . . . . Sun Dial 20 followed the target into our radar ground clutter area and we were unable to give Sun Dial 20 any further assistance in reestablishing contact.



The F-94 radar operator stated:


. . . Hi-Jinx gave us a vector of 320 degrees. Hi-Jinx had a definite radar echo and gave us the vector to intercept the unidentified target.


That Shiroi tracked both the target and the pursuing F-94 into their groundclutter whilst the F-94 had radar contact for at least 90 seconds with the same target seems beyond reasonable doubt, if only because their was no other attempted vectoring/intercept episode with which either party could have become confused. This conclusion is surely proven by the very close match in target range and bearing as supplied by Shiroi and almost immediately found by the F94 radar operator: Shiroi sent the F-94 towards the target on 320 degrees, stating that the target would be at 11 o'clock (30 degrees to port), range 4 miles (7000 yards) from the aircraft; a matter of seconds later the APG-33 picked up a target at 10 degrees to port, range 6000 yards, "rapidly moving from port to starboard", and the aircraft followed the target in a starboard turn, unable to close to within the APG-33's lock-on range of 2500 yards but maintaining contact for 90 seconds at 375 knots (about 11 miles of travel), whilst ground radar simultaneously tracked both F-94 and target into its groundclutter. According to IR-35-52:


Lt. ----, F-94 radar operator, has had about seven years' experience with airborne radar equipment. He states that the object was a bonafide target, and that to his knowledge, there was nothing within an area of 15-20 miles that could give the radar echo.




The APG-33 radar is checked before and after every mission and appeared to be working normally.


   The last of Thayer's four symptoms of AP is the invisibility of the object from Shiroi and the interceptor, despite the fact that the target range from these observers was at times as low as about 7 and 3.5 miles respectively. Personnel at Shiroi were specifically sent out to look for it, and the aircrew  were obviously alert: "Sun Dial 20 was ordered to search the Tokyo Bay area, keeping a sharp lookout for any unusual occurrences." There is no wholly satisfactory interpretation of these facts. Leaving aside the radar indications for the moment, these negative visual searches do lend prima facie support to Thayer's hypothesis that the object being concurrently seen from Haneda was at much greater, possibly astronomical, distance. If the primary visual sighting were a misinterpretation, of course, then the probability of subsequent radar detection of a completely unrelated "UFO" which happened to occur by chance in the same area would become implausibly small, and a conventional cause such as anomalous propagation would have to be seriously considered despite the apparent counterindications.


   The general "exceptionally good" weather conditions were certainly conducive to ducting: a calm, clear August night, temperature 78 degrees F, visibility excellent (Mt. Fuji, 12,390', "clearly discernable" from the air at 69 miles SW), and the likelihood of moist air over the bay overlain by advected drier air from land to the SW. Combined nocturnal radiative cooling of the land and evaporation from the sea might be expected to cause temperature and humidity inversions and thus the kind of surface ducting common to such conditions. It would certainly not have been surprising had unusual AP ground echoes been detected, although it is also true that the degree of ground-clutter on the CPS-1 is nowhere indicated to have been abnormal (the CPS-1 was not fitted with MTI so that there was always clutter on-scope). In the absence of detailed information it is possible that the "three or four blips" with "no definite movement" initially picked up on the low beam of the CPS-1 after Haneda called Shiroi were AP targets, perhaps ships in Tokyo bay or beyond detected in a surface duct.


   Even during the later phase of the incident, when a clear moving target was detected, no confirmatory contact appears to have been made on the CPS-4 height-finder. (The F-94 was instructed by Shiroi to go to 5000', but it appears that this was only a tactical best-estimate based on the Haneda visual report.) This certainly could be considered as possible circumstantial evidence of a propagation or electronic anomaly, inasmuch as a concurrent HRI track on an electronically independent set with a different operating frequency would typically be considered a counterindication of AP, system noise or radio frequency interference.


   This is an interesting point. In the present case the precise wavelengths of the CPS-1 and CPS-4 sets are unstated, but both instruments were in fact Sband, operating in a similar EM region around 10 cms; and to the extent that their respective bandwidths were relatively close they would be proportionately likely to respond to AP conditions with similar false echoes - unlike a situation where a CPS-4 is operated in tandem with, say, an L-band CPS-5 surveillance set. This raises the question of why, given what one might call classic conditions for ducting to occur and the presence (ex hypothesi) of unusually sharp, stable, moving AP targets on the CPS-1, no targets of any kind were reported on the CPS-4 heightfinder. A possible answer is that the CPS-4 had already been discounted as ineffective (at the time of the earlier, AP-like contacts) due to excessive ground-clutter at low elevation angles, and the controller may have continued to encounter this problem or even given up altogether the attempt to acquire HRI contact with the target. This interpretation implies a differential susceptibility of surveillance and heightfinder sets to AP clutter at minimal elevations, and of course also requires the target being sought to be at a  minimal elevation. If these conditions are fulfilled, then the absence of heightfinder plots could no longer in itself be considered diagnostic of AP.


   Firstly, the approximate mean elevation at Shiroi of the CPS-1 target (assuming a real target) can be inferred: the F-94 at an altitude of about 5000' acquired radar contact with it at range 6000 yards and at a depression angle of 10 degrees, which places the true target altitude at this time at about 3500'; and therefore, during the times when the controller had a clear PPI target on which to azicate the heightfinder (that is, when the target's orbit took it out of the CPS-1 clutter and over the sea where it was intercepted) the mean elevation of this target at ranges of about 13-15 miles would be only about 2 or 3 degrees. The question then arises of the the likely differential performance of CPS-1 and CPS-4 scopes, possibly in the sort of AP conditions inferred, with targets at such low elevations.


   Now a typical surveillance beam with an approximate cosecant-squared vertical profile has a horizontal beamwidth on the order of 1 degree, whereas a typical nodding-fan height-finder like the CPS-4 has a vertical beamwidth also of about 1 degree but a horizontal beamwidth on the order of 4 degrees or so; and it can be seen that when the two sets have matched peak powers and pulse repetition frequencies (for matching range performance) then the total energy radiated per unit time through any 1 degree of elevation by the CPS-4 cannot be less than four times that radiated by the CPS-1, and will in fact be much greater because the CPS-1 output has to fill a much larger solid angle many degrees in elevation. (Even if the powers of the two sets are not matched - only that of the CPS-1 is known: 1 megawatt - and if the power of the CPS-4 is a more modest 500 kW, then the ratio of energy density per degree of elevation must still be on the order of 10:1.) A height-finder beam of 4 x 1 degrees, therefore, will radiate more energy into the solid angle defined by the top edge of the beam and the earth than will the 1-degree surveillance beam if that solid angle is less than about 4 degrees of elevation; and given that the heightfinder's antenna elevation is close to a critical grazing incidence for anomalous propagation of less than about 4 degrees then the onset of ground returns due to AP will be sooner for the height-finder. Put another way, in trapping conditions the 4-degree height-finder beam will have a ground-incident "footprint" four times as large per unit range as the surveillance beam, so that the ratio of the clutter signal to a point-target signal at a given range on the height-finder can be several times larger.


   What all this means in terms of relative performance is that, when "looking" for a target at elevations of only 2 or 3 degrees, the CPS-4 antenna can be radiating a great deal of its main-beam output (plus a good deal of sidelobe) into a surface duct at grazing angles liable to severe trapping, leading to clutter returns which would be likely to overwhelm any relatively weak signal from a point target above the duct; but the signal-to-noise ratio of the CPS-1 could be several times better in the same conditions, preserving the super-clutter visibility of the same point target.


   The absence of height-finder plots on this low-elevation target is therefore not necessarily inconsistent with the presence of a solid, radar-reflective body - at least, even though the fact remains consistent with anomalous propagation it is no longer a sufficient condition of that hypothesis. But what of the target tracked on the CPS-1? Can it be described as consistent with AP?


    Any kind of anomalous propagation is difficult to entertain in the circumstances. The correlation between X-band 3-cm airborne radar and S-band 10-cm ground radar indications has to be considered good, and the target behavior on either scope during this phase of the events is really not diagnostic of AP. The 2 x windspeed behavior of partial reflection echoes would appear to be entirely inconsistent with displayed speeds on the CPS-1 ranging from zero ("hovering occasionally") to at least 250-300 knots and probably in excess of 375 knots (430 mph) on departure, and the repeated orbital tracks cannot realistically be explained by waves propagating across an inversion surface. Low level winds were light, SSE, overlain by a SW airflow: there is no indication of any anticyclonic circulation of the order of 150 knots (175 mph), and the target headings would at various times have been transverse, or even converse, to either of these (light) winds. The presentation on the CPS-1 was sharply defined, that of "a bonafide moving target", which made two wide movements over an azimuth arc of about 40 degrees and back with a near-doubling of range, during which movements the only reported signal loss occurred due to its passage through the permanent ground clutter. This is not characteristic of an illusory track due to a series of sporadic AP returns from ground (or sea) targets. And when, on the second orbit, concurrent air-ground detection is achieved by independent fixed and mobile radars at very different wavelengths, pulse repetition frequencies and incidences, with correlating position and motion on both scopes, then AP seems out of the question.


   The hypothesis of a "ghost" reflection from the F-94 generating the target tracked on the CPS-1 appears to be ruled out. Such a ghost would always appear on the azimuth of the primary reflector (the aircraft) and at greater range proportional to the added trip time to the secondary reflector, yet it seems quite clear that the F-94 approached the target on a heading of 320 degrees (NW) and "followed" it in a starboard turn to the N, over the coast and "into our radar ground clutter". At no time during this track can F-94 and target have been on the same azimuth from Shiroi (some dozen-or-more miles NNE) and at no time can the target have been displayed at greater range - indeed the F-94 would have been at greater range almost the whole time. Furthermore the first orbit of the target was tracked before the F-94 was even airborne out of Johnson, and the notion of a "ghost" echo separating into "three smaller contacts, maintaining an interval of about mile" is incredible. Finally this hypothesis offers no explanation of the concurrent target painted by the airborne APG-33.


   No effect due to sidelobes, multiple-trip echoes, system noise, component failure, radio frequency interference, birds, insects, clear air turbulence, meteorwake ionisation, windborne debris or balloons seems even remotely adequate to the core multiple-radar episode in this case.


   Total tracking times were brief, but far from ephemeral: upwards of 90 seconds AI contact is a significant duration, with the very rapid scan and virtually continuous rate of renewal allowing ample time for the operator to estimate the target parameters and assure himself of a "bonafide" point contact. On the CPS-1 the first track was plotted in the clear (away from the ground clutter) for approximately the orbital circumference, about 5 minutes at the mean estimated speed (neglecting stationary episodes) of 125 knots (143 mph), or > 20 paints at 4 rpm - longer if the unquantified hovering episodes are taken into account. The exact timing of the sequence of events is unclear, but roughly consistent assuming a -circuit time of about 7 minutes (28 paints per plotted  track): Shiroi had the target in clear contact on its first orbit "at the time of the scramble" - about 2355 - so if it had been picked up emerging from the clutter at, say, 2354 it would have been lost in the clutter again by about 000001; the landward half of the 4-mile-radius orbit would be completed in about the same time, leading to emergence from the clutter over the coast at about 000708 (this is consistent with IR-35-52's statement that the Shiroi controller's time for this event, "0017", showed a roughly "ten minute difference" from other personnel statements due to "typographical error"); by about 0012 the target would have been a little more than half way around the seaward sector of its orbit when it split into three, and by 0015 when Shiroi vectored the F-94 onto the larger target it would have been heading towards the coast again, accelerating now and soon lost on GCI radar whilst the F-94 continued in pursuit with the target on-scope until about 0017:30.


   No great precision can be claimed for this timetable, of course, due to the large number of missing data-points and the inherent ambiguity of timings which are at best approximate and quantised to the minute; but at the same time, no serious inconsistency emerges. It is notable that ground and airborne radar operators were in no doubt that they had contact with the same "definite" and "bonafide" moving target.


   The Haneda/Tachikawa visuals are certainly the most problemmatic part of the report, and how, or even whether, they are related to the radar events is difficult to establish with confidence. In the absence of exact times for the visual movements one can only say that there is a suggestion of angular motions which correlate in a qualitative way with the radar orbits. However, the general bearing of the light from Haneda does correlate with the general location of the radar target(s) over the N of Tokyo Bay; the independent line of sight from Tachikawa does intersect in this same area; and the loss of visual contact from Haneda does correlate at least approximately with the inferred departure of the radar target. These points of correspondence must be weighed against the inability of a C-54 pilot in the air to confirm the strangeness of what appeared to him to be a "brilliant star", the statement that the light appeared to ascend at approximately the rate of siderial revolution, and the failure of observers at Shiroi and the aircrew to visually confirm the Haneda sighting, all of which points are suggestive of an astronomical source. However the brightness, angular size and detailed structure of the object - all of which can fairly be regarded as reliably scaled and described, in terms of the usual standards of witness observation - as well as the less-reliably reported angular motion, are inconsistent with the known propagation properties of the atmosphere as they could affect the image of the brightest body in the NE sky at the time.


   The question of unknown propagation mechanisms therefore arises. In this connection one is struck by certain parallels between the structure of the object and that of the Jovian system: it appeared through binoculars as a disc with three or four small, faint lights disposed in an arc below it. The planetary disc of Jupiter and the four bright Gallileian satellites are quite easily observable through 7x50 binoculars - indeed, some acute observers can see these satellites with the naked eye under favourable conditions near maximum extension, although generally their magnitudes of between +6.3 and +5.1 are obscured by the glare of the planetary disc. Plainly the observers were not looking at Jupiter's true E azimuth, and even directly contrasted the object's apparent brightness with that of Jupiter, so that some kind of mirage image has to be assumed, displaced laterally by some 40-45 degrees for much of the observation,  an angle periodically reduced as the image dimmed and shrank back towards the E true bearing of Jupiter. This image would contain some distortion to account for the fact that the four equatorial satellites, which transit Jupiter close to the ecliptic plane, were optically shifted into a pendant arc. Brightening of the disc is also implied, as is some dilation of its image by about 4 diameters from Jupiter's 40 arc seconds to a scaled angle of about 3 arc minutes without observable chromatic aberration. One imagines that such an effect might require a stable, near-stationary "bubble" of severely abnormal refractive-index-gradient performing as a near-perfect achromatic lens.  A line of sight from Tachikawa to Jupiter might graze this local discontinuity at a shallower angle, leading to an abnormal appearance for observers there also, and this requirement would place the mirage-producing zone in the region of N Tokyo Bay where extremely unusual radar propagation phenomena were observed.


   It is probably safe to say that no such phenomenon has ever been described or modelled in the field of meteorological optics. There are extremely rare reports, however, which appear to indicate dramatic propagation anomalies of a nearly analogous kind. For example, Minnaert (1968) describes a remarkable double-sun which was photographed from a ship in the Indian ocean and witnessed by numerous passengers: an exact duplicate of the solar image, perfect as to colour, shape, size and elevation, appeared offset in azimuth to one side of the true sun. There appears to be no conceivable explanation of this effect in terms of known mirage phenomena, and it indicates that some extraordinary properties of the atmosphere remain to be understood. It is, of course, not excluded that some such atmospheric phenomena might conveniently be termed "UFOs", or may be hypothesised to occur in association with other phenomena which some might choose to call "UFOs".


   If one wishes to demur from these spectacular conjectures, then Capella remains the only plausible astronomical source in the NE sky. Novel propagation mechanisms aside, it may be relevant that the details of fine structure were observed through binoculars, which suggests an alternative to diffraction due to atmospheric mist: condensed moisture on the objective glasses or eyepieces (exterior temperature was 5 degrees F above the dew point, but humid conditions in the tower or taking the binoculars out of cooler storage before use might cause the lenses to mist) or trapped moisture or other defects between lens elements which might cause diffraction, internal flares and/or improper focussing. The central image might conceivably be smeared, within an array of fainter secondary images. The likelihood of this hypothesis seems negligible, however, given that the observations lasted upwards of 50 minutes; that tower binoculars would be in not-infrequent use and chronic defects would be noted; that other stars in the NE sky are sure to have shown up the effects of any misting; that motion of the binoculars would almost certainly reveal the cause of the diffraction/blurring; and that four experienced tower controllers would not be so plain stupid.


   The issue of the non-visibility of any light source from Shiroi or the interceptor remains difficult. One point worthy of note, however, is that according to Ruppelt (1956, p.247) who headed Blue Book at the time and discussed the case with Far East Air Force intelligence personnel, one of the tower operators "had the distinct feeling that the light was highly directional, like a spotlight." It is true that the F-94 would probably never have been within less than 45 degrees of the line-of-sight from Haneda during the interception; it is also true that, if the light was the object being tracked on radar, then at the  time of the F-94's arrival in the area from the NW - when it would almost certainly have crossed this line of sight, at least in azimuth, on its way down over the bay - the target was over the land at the farthest point of its orbit and the light (ex hypothesi) was only visible from Haneda as "a small light" which was "difficult to follow closely". Thus it is conceivable that the aircrew may never have been in a position to see a directional light source. This argument is not very strong, however, and the degree of directionality implied (a very high degree of collimation with virtually zero effective scattering from the beam) would seem to conflict with the apparent simultaneous visibility of the source from Tachikawa, bearing some 50 degrees away from Haneda. No easy explanation exists for the display observed at Haneda, and one must admit the probability that radar and visual "UFOs" are in some sense related; but no satisfactory integration of all the visual details with the radar events can be achieved at this time.


   In conclusion: The possibility exists that the early detection of three or four stationary targets on the low beam of the CPS-1 was due to anomalous propagation, for which there is indirect circumstantial evidence in the form of the general weather conditions and the inferred presence of considerable lowanlge clutter on the CPS-4. It is also true, however, that the visual object was in view from Haneda at this time on the correct bearing, possibly at lower elevation than that subsequently attained, and the targets may have been related. There is insufficient information to argue this point.


   During the later radar tracks, however, the indications appear to be in every respect consistent with the presence of at least one high-performance aerial object with a radar cross-section smaller than that of a normal jet fighter, capable of jet-speed evasion of a pursuing F-94 (then probably the most advanced in the air) and erratic variations of speed. Its CPS-1 track showed it "stopping, and hovering occasionally" during a roughly 25-mile circumference orbit at 100-150 knots, implying, if not actual stationarity (the 15-second renewal rate of the PPI is too slow, and the resolution too coarse, to allow full confidence in this judgement, the controller's "considerable experience" notwithstanding, and concurrent height data are lacking) then at least a repeated deceleration to a speed (in light winds) almost certainly well below the 94-knot stalling airspeed of an F-94.


  Despite the periodic obscuration of the target(s) due to ground clutter, it is unreasonable to attempt to dissociate the target tracked on the first orbit from that tracked on the second orbit, one component of which was in multiple radar contact during intercept and achieved a speed probably in excess of the F-94's 430 mph; the two behaviors appear to be the rationally related movements of a single object (or sub-resolution cluster of objects) which had an inordinate range of speed and gave evidence of intelligent, evasive action when illuminated by the radar of the F-94.


   The splitting  into three contacts immediately prior to F-94's intercept attempt, followed by rapid departure, strikingly suggests something analogous to a tactical manoeuvre responsive to potentially hostile engagement. The fact that one contact remained brighter, with two smaller targets separating out to ranges of some 400 yards, resembles the deployment of radar decoys and might be interpreted as a test of remote drones or active jamming to confuse hostile pursuit. But it is not believed that any such free-flying drones were being deployed by any known small jets in 1952. Infra-red decoy flares were probably  under development to counter the incoming generation of IR-guided Sidewinder missiles, but flares would not answer the radar description and would also have been highly visible to the aircrew. Launched missiles would not "maintain mile separation" from the parent aircraft. The active jamming technology to achieve such false targets was almost certainly unknown in the world of 1952 analogue electronics, and the range of performance indicated is unmatched by any vehicle known to have been flying in 1952.


STATUS: Unknown



7.  DATE: December 10, 1952     TIME: 1915 local                CLASS: R/V air radar, air


Washington            SOURCE: Thayer, Condon 1970, 140                 

                                                    RADAR DURATION; unspecified


EVALUATION: Thayer - probable balloon


PRECIS:  The crew of a 2-seat F-94 flying between 26-27,000' near Odessa at 1915 sighted a very large, round white object which they estimated to be "larger than any known type of aircraft". It appeared to emit faint reddish light from two "windows". Its apparent movements were erratic. It performed a chandelle in front of the aircraft, seemingly able to "reverse direction almost instantly", the closed with the aircraft head-on at high speed before suddenly seeming to stop and break away. Fearing imminent collision the pilot banked the F-94 into a turn. They lost sight of the object at this time and never reacquired it visually, although the weather was clear over an undercast at 3000'. However 15 minutes later at 1930 a target was acquired on the airborne radar: It was moving generally west to east at about 75 knots. The crew believed this target was related to the earlier object, but no simultaneous visual sighting was made.


NOTES:  It appears that earlier Blue Book studies concluded that the visual object was a mirage of Venus. However, Thayer's 1968 analysis contains the Spokane radiosonde profile (release time 1900 LST) which does not indicate the presence of  temperature inversions near the aircraft altitude. A slight inversion is present at about 30,500' but this is much too far above the aircraft altitude to satisfy the small grazing angle requirement. Furthermore the small (order of 1/2 degree maximum) image wander due to mirage would seem inconsistent with the extent of motion reported, and the complete disappearance of the source - Venus being a brilliant object that would remain prominent in the clear sky even if the aircraft moved away from the critical mirage angle - is not convincingly explained.


Thayer observes that the Spokane radiosonde release time of 1900 is close to the time of the sighting, and that the appearance and motions of the object are quite characteristic of an encounter with a lighted balloon (saving some peculiar details such as "reversals" in front of the aircraft and the object's disappearance). Further, he observes that a balloon would have risen to at least 17,000' in 15 minutes and that the winds at the highest level plotted for the Spokane profile (18,000') are 260 degrees at 66 knots - not inconsistent with the reported easterly motion of the radar target at 75 knots.


On the face of it this explanation is quite convincing. A number of similar balloon "dogfights" are on record in which the rate of closure with an unexpectedly slow and unexpectedly nearby lighted balloon creates the illusion of a series of rapid, head-on passes. Typically balloon interceptions involve a series of approaches (afterwards shown to have been a slow spiralling climb over the same general area) with a small round light, sometimes blinking and sometimes steady, which eventually seems to move away. The "UFO" is the pendant tracking light below the balloon. There are certain differences in this case:


The object was described as an extremely large round object, larger than any aircraft, not a small ball of light. This might suggests that the balloon itself (probably on the order of 20' across at this altitude) was being seen by reflected illumination from its tracking light, but no such light was reported. The only other apparent source of illumination was the two reddish "windows" presumably on its surface. These are not a recognisable weather balloon feature. They could themselves conceivably have been reflections caused by the 1.5 candle battery lamp carried by such balloons (although this would normally be "white" light), but again no such pendant light was reported. If the balloon fabric was brightly lit then the source of this illumination ought to have been visible too. If the balloon's lamp had failed then it could have been illuminated by bright moonlight (the sky was clear; position and phase of moon unknown) which could explain a large, round "white" body; but the reddish light from the "windows" has no clear interpretation on this hypothesis either.


The disappearance of the object after a single "pass" is also not characteristic of a balloon which - from whatever source - was rather well lit. This would be best explained by supposing that the object seen was not the balloon itself but the small tracking light attached to an unseen balloon, which failed just as the F-94 flew past. It should be noted, however, that this coincidence could not be explained by impact with the F-94 since the Spokane profile shows that this balloon continued its ascent well past the aircraft altitude apparently without incident. In addition the object was seen to perform manoeuvres, including apparently "reversing direction almost instantly", whilst still at range ahead of the aircraft. True, there is some similarity between these types of motions and illusory motions due to balloons encountered at close quarters; but it was only after this behaviour that the object appeared to race head-on towards the F-94, and even then it reportedly "stopped suddenly" before once again pulling away. Such apparent motions would tend rather to favour the inversion-mirage hypothesis, were it not that the weather data are inconsistent.


The balloon hypothesis is therefore not quite as attractive as it at first appears. Nevertheless unusual misperceptions do occur and Thayer's quantitative argument remains to be addressed. If there is good reason to suspect that a balloon was in the area and moving in the approximate direction reported then it would be only reasonable to suspect the balloon, given at least some similarity with known close-encounters with balloons.


There appears to be some confusion here, however, between the times, speeds and altitudes relating to the visual and radar events. The maximum rate of climb of a radiosonde - about 1200 fpm - would indeed put the Spokane balloon at about 18,000' by the time of the visual sighting at 1915; but the altitude of the visually observed "balloon" at this time is fixed very accurately by the aircraft altitude as being 26 - 27,000'. On the other hand the radar target, whose motion appears to correspond with the reported winds at >18,000', was detected at 1930, by which time the balloon would have been at an altitude of about 36,000', having spent some fifteen further minutes climbing through air streams of unknown strength and direction - which negates the apparent match. (A balloon which developed a slow leak at 18,000' might continue to drift at that altitude for some time before coming down. But the Spokane profile shows that this balloon did not leak. The part profile published gives values to at least 38,000' and it was not listed as "lost".) Therefore even if the radar target could be explained as the Spokane balloon - with a little strain, its likely altitude being some three and a half miles above the F-94 - the same balloon cannot have been responsible for the visual at about 27,000' fifteen minutes before since it would have been about two miles too low.


The relative locations of Spokane and Odessa also render the balloon hypothesis dubious. Odessa is 65 miles SW of Spokane: Even at the 66-knot wind speed plotted on the Spokane profile for 18,000' the balloon could not possibly have travelled more than 15 miles from Spokane in any direction by the time of the visual near Odessa, and since this wind was from the west it would be taking the balloon not SW towards Odessa but eastwards into Idaho. In reality, of course, even if the surface and low-level winds (not given) were blowing almost opposite to the recorded 18,000' winds and thus did initially take the 1900 balloon towards Odessa, the mean speeds are likely to have been much slower at low altitudes so that there is an even larger mismatch between the probable range achieved by the balloon (much less than 15 miles) and the 65 mile range to Odessa.


In conclusion there is some doubt about whether the radar target could have been the Spokane weather balloon. There is no good reason at all to think that the earlier visual sighting was due to the Spokane balloon. Apart from appearance and behavior not convincingly like a balloon, the object was at the wrong altitude, the wrong range from Spokane and (probably) the wrong bearing from Spokane to have been the 1900 LST Spokane radiosonde balloon.


A leaking balloon from elsewhere might have been encountered at >26,000' near Odessa and might subsequently have descended towards the 18,000-foot, 66-knot airflow; but it seems a little improbable that this same balloon would be encountered a second time after 15 minutes of flight at jet speed. It is not impossible, however, if the F-94's flight path happened to take it back over the same area. This scenario cannot be evaluated without knowing at least the aircraft's home base and mission/destination, and could not be proven without plotting a specific "lost" balloon against detailed wind flow charts, which is at least impractical and probably impossible. It should also be mentioned that match between winds aloft and radar target motion is suppositious: There is no information on the actual altitude of the target.


Possibly most damaging to the theory that any single balloon could have been responsible for both visual and radar objects is the question of why a balloon which was a noteworthy radar target during the second encounter was not detected at all prior to or during the presumably much closer first encounter.


A still less plausible hypothesis is that two separate radiosonde balloons were involved, neither of which was the one released from Spokane, one having lost its radar reflector and/or instrument package so that it was not detectable. The nearest scheduled release site to Spokane would have been near Tacoma, Washington, about 230 miles W. Others within 300 miles or so include: Great Falls, Montana; Boise, Idaho; Portland and Salem, Oregon; and Cape Flattery, Washington. Given the generally W-E upper airflow indicated at Spokane then Tacoma, cape Flattery and Portland might be suspected as possible launch sites (radiosondes are released 4 times a day, usually at 6-hour intervals, but local times vary).  However these balloons do not usually travel large ground distances at low levels. About 90% reach 80,000' and shatter; some 50% reach 100,000' before pressure discovers weakness; a small number continue to 140,000' or even higher. Typical flight times of 60-80 minutes imply average windspeeds of around 200 mph (and maxima far greater) for a Tacoma release to cover the ground distance to the Spokane area before burst, and this balloon would then be at an altitude about 4 times greater than is needed to account for either sighting. Both sightings appear to require balloons with slow leaks, which could have been released at almost any of these sites at almost any time during the previous 12 hours or so given the know pattern of mesoscale wind circulation. But given that 90% of radiosondes complete their ascent to 80,000' it is obvious that the probability of any given flight remaining near 20,000' for an extended period due to slowly dropping buoyancy must be on the order of 0.1 or much less. Probability cannot properly speaking be applied retrospectively, or even calculated in such circumstances, but the coarsest of intuitive guesswork would suggest that the likelihood of two separate balloons suffering similarly and ending up near Spokane/Odessa at the same time must be on the order of <<0.01.


Very large polyethylene balloons with a volume of several million cubic feet had been flown covertly for several years by 1952. Such a balloon could certainly be described as 'larger than any known type of aircraft' and would have been unfamiliar to most pilots. (scientific balloons of this type were later required by the FAA to carry lights below 60,000' but it is doubtful if the same was true of the mainly classified earlier flights.) However it remains a problem to account for the movements described by the two airmen, and an abrupt visual disappearance becomes still less understandable in terms of so large a balloon in a clear sky. If the same balloon was encountered 15 minutes later by radar then why was it not seen visually at this time even though the crew would have been searching for it in the stated belief that the radar target was the same 'UFO'? Perhaps it was much further away at this time near the limits of radar range. But if it had been nearer earlier, why was it not detectable by radar then?


In summary, the Spokane raadiosonde does not appear to be a plausible hypothesis for either sighting. It could not have been at the altitude or location of the 1915 visual sighting, and on the basis of known times, known ascent rates and the documented 1900 Spokane profile it cannot have been near the required 18,000' level at the time of the 1930 radar contact. The generality of the account can only somewhat plausibly be accounted for by one or two unidentified balloons. The argument that there is good reason to suppose the presence of a balloon in the vicinity of the first sighting is actually invalid. The evidence of a balloon in the second instance is entirely circumstantial, based only on a rough and arbitrary match between target motion at an unknown altitude and winds at the 18,000' level. One could as well argue that there is a good match with a group of swans flying downwind at 10,000' - true, but with so many indeterminate variables such a 'correlation' is not meaningful. The corollary, of course, is that the radar target could have been almost anything and there is no positive evidence to connect it with the visual sighting. Indeed, given that the visual object apparently was not a radar target one could say there is positive evidence that the two sightings were unconnected. It is possible, however, that the airborne radar was not activated at the time of the first sighting. As regards the visual object, there are certain features that suggest a balloon but other features which are inconsistent with a balloon, implying gross misjudgments of size, appearance and motion. In terms of detection, one would have to conclude that the modus operandi  is not entirely typical of balloon cases on file.


In conclusion the core sighting cannot be identified as a balloon. The celestial body/elevated inversion/mirage model is difficult to support on the basis of meteorological evidence and the disappearance of the source in a clear sky. But some of the apparent movements are undeniably suggestive of a looming and receding mirage image. It is possible that a bright celestial body on the point of setting could be viewed at the critical grazing angle through a surface inversion on the far horizon that wasn't sampled by the local profile. The Spokane profile in fact indicates a slight surface duct below about 500', and although there is no evidence whatever that this indicates conditions near the far horizon of an aircraft at 26,000' it is possible that the setting of a bright planet (the plane of the ecliptic would dip below the horizon in the WSW) could present a brief mirage image that appeared to dance up and down, expand and contract, before disappearing. Even the reddish 'windows' could with some strain be explained as refractive separation, red light tending to be separated towards the bottom of the image. But in addition to being speculative this hypothesis suffers from the objection that the aircraft was flying above an undercast at 3000'. The case therefore merits further study.


STATUS: Insufficient Information



8.  DATE: December 15, 1952          TIME: 1915 local            CLASS: R/V  air radar/air

LOCATION:                                     SOURCE: Thayer, in Condon 1970, 126

Goose AFB

Labrador, NF

                                                           RADAR DURATION: unspecified, but brief


EVALUATION: Blue Book - Venus/radar malfunction

                           Thayer - mirage of Venus/radar malfunction


PRECIS:  An F-94B and a T-33 jet trainer were flying at 14,000' when both crews saw a bright red and white object at 270 degrees azimuth. The F-94 attempted to intercept the object but was unable to close the range. At one point during the chase the radar operator acquired a target at the correct approximate azimuth and the AI radar "locked on" briefly, but contact was soon lost. At 1940, after 25 minutes of pursuit at an indicated airspeed of 375 knots, the object faded and disappeared, at which time the F-94 was about 20 miles from its position at the initial sighting.



NOTES:  The Air Force conclusion that the radar malfunctioned was not based on evidence of a fault (the AI radar would usually be checked before and after every mission) but simply on the fact that the contact was brief. Venus was setting in the E at about the time of the event and seemed a probable explanation, in which case there would be every reason to dismiss the radar contact as a malfunction. Thayer agrees that the aircrew were probably chasing Venus, but adds that the image could have been a mirage of Venus which would be more likely to mislead experienced pilots and would explain the description of the light as a source with "no definite size or shape". An astronomical explanation is supported by the witnesses' statement that the relative elevation of the object appeared to remain constant despite the varying altitude of the aircraft, by the inability of the F-94 to close on the object, and by the fact that no ground radar contact was reported with the object. The weather was clear with unlimited visibility.


   There is a serious flaw in this otherwise plausible hypothesis, however: after 25 minutes of pursuit at an IAS of 375 knots (about 430 mph) the F-94 had covered only about 20 miles on the ground. If it were chasing a stationary celestial body it should have covered some 180 miles during this time. The actual distance would imply a ground speed of about 48 mph (about half the stalling airspeed of an F-94), and headwinds of 380 mph at 14,000' are presumably impossible. This curious anomaly in the report is not explained by Blue Book or by Thayer. It suggests that the pursuit was not confined to a single azimuth and that the F-94 had thus been brought back to a point near its original position when the object disappeared, in which case it could not have been Venus or any other celestial body. A lighted balloon might lead a pilot on an erratic chase, back and forth over a restricted area, but the F-94 would rapidly overtake a balloon and none of the characteristic "dog-fight" behavior of balloon interceptions occurred in this case.


    Unless this contradiction can be resolved the object cannot be identified as Venus or any other object with any confidence, and if it was not Venus then a reassessment of the AI radar "lock on" becomes in order. The reported absence of ground radar contact is suggestive, but not wholly probative without information on the range of the aircraft from Goose and the completeness of any report or investigation at that site. The case would appear to merit further study.


STATUS: Insufficient information



9.  DATE: February 4, 1954              TIME: 2300 local           CLASS: R/V  ground radar/
                                                                                                                   radar/ground visual

LOCATION                                     SOURCE: Good ATS 1987 274

Carswell AFB, Texas

                                                         RADAR DURATION: unspecified


EVALUATION: unspecified


PRECIS: This report was submitted in a PRIORITY message classified CONFIDENTIAL from the Commanding Officer 19th Air Division, Carswell AFB, to Commander 8th Air Force, Carswell; the Director of Intelligence, Washington, D.C.; Air Defense Command, Ent AFB, Colorado; and ATIC, Wright Patterson AFB, Ohio; with distribution to the Joint Chiefs of Staff, CIA and NSA.


   At 2300 Carswell GCA radar picked up an inbound target at 13-15 miles. At < 10 miles the target appeared as a one-inch return on the scope. Because the target appeared to be unknown traffic heading towards the base GCA notified the control tower and the Officer of the Day. Personnel in the control tower looked out and saw the object approaching. It passed directly over the tower at an estimated altitude of 3-4000' but no sound could be heard. All personnel in the tower saw the object, and the chief tower operator observed it continuously through binoculars. It resembled an aircraft with a long fuselage, elliptical wings and a stabilizer, but with no visible or audible propulsion unit. There were very bright lights in the nose and tail, and two yellowish ventral lights on the fuselage. One observer believed he could make out a light on each wing tip. None of the observers could identify it as a conventional aircraft. An investigation was later unable to identify any responsible flight and there was "no unusual activity, meteorological, astronomical, or otherwise" which could account for the sighting. The intelligence report concluded that all witnesses were "completely reliable", and that the details of the sighting were "probably true".



NOTES: A possible explanation is that a conventional aircraft flew over the tower, and that the observers, for whatever reason, were mistaken about its appearance and failed to hear it due to strong winds aloft and/or the masking effect of local noise. The case does not appear in the final Blue Book listing of unknowns, and one suspects that the official reflex here would have been to file it under "probable aircraft" notwithstanding the lack of positive ID. This hypothesis is not wholly convincing, however: the visual sighting, by multiple trained aircraft observers in a tower designed to afford an excellent view of local airspace, with both the naked eye and binoculars, does not argue strongly that the object was a conventional aircraft. It is noteworthy that although the observers' expectation was presumably of a conventional aircraft none of them, despite their professional familiarity with aircraft and the object's generally aircraft-like appearance, was able to identify it as such.


   Given that some type of aircraft did apparently over fly the field it is reasonable to suspect that the inbound radar target was related to the aircraft.  The radar description is frustratingly sketchy, however. The intelligence report states:


Object was first detected by the Carswell GCA Sta at a distance of 13 to 15 mi fr Carswell attn was drawn to the object because of the large rtrn presented on the scope. Object when viewed on 10 mi scope gave a rtrn of 1 inch. Obr scanned from 10 degs to 02 degs on search and object retained the same rtrn dur this opn. Because of the unusual rtrn, and because object was approaching directly over the fld the GCA opr notified the Airdrome Off of the Day and the tower.


   This is confusing. It is not immediately clear whether "the 10 mi scope" is another radar electronically independent of the scope (presumably surveillance) on which the target was first observed at 13-15 miles, or simply another display fed from the same antenna but adjusted to a different range scale. If the latter then there may be an anomaly in the statement that the target appeared as a one-inch return at a displayed range of <10 miles.


   A typical GCA surveillance scope with a range of about 50-60 miles (such as the S-band CPN-4 commonly used with electrically-scanned blind-landing radars in the MPN-11 GCA system of mid-'fifties vintage) would not be a very large display, probably no more than about 15" inches across. Adjusted to a 10-mile range scale, an approximately 1" target-arc with a radius of 7" (10 miles) subtends an angle of about 8 degrees, and if displayed anywhere on the scope other than at the extreme periphery would subtend a still larger angle. This is clearly not a normal aircraft return which would typically be only a couple of degrees or a few millimetres wide: for a typical surveillance beam width of about 2 degrees between the half-power points a 1" target arc anywhere on a 10-milescale PPI of < 15" diameter implies a reflective solid target at least several thousand feet wide.


   It is possible that the display sensitivity might be adjusted so that a strong aircraft target would be displayed by its main-beam return flanked closely by smaller echoes from the two major sidelobes, but an experienced operator would be familiar with this type of presentation, which is quite different from the much broader and integrated arc implied. It is conceivable that the operator was so inexperienced that he failed to realise that the display sensitivity was turned up to the point where the screen was fizzing with system noise and clutter, and took the consequent abnormal target presentation to be the norm; but this seems hardly likely, especially in view of the Air Force investigation which noted that all personnel were "completely reliable". It is also possible that the operator's statement was simply misunder-stood or misquoted by the preparing officer.


   If it is assumed that the "10 mile scope" is a separate instrument from the airfield surveillance scope on which the more distant target was (presumably) initially detected then it is possibly one of the landing system displays - fed from fixed azimuth and elevation sector-scanning antennae and used to guide aircraft to a blind touch-down. Information on the exact nature of the GCA setup at Carswell in 1954 would be needed to make even broad inferences about the radar cross-section implied by "a 1-inch return" on the landing radar. Certainly a 10 mile range would not be inconsistent with the range of a landing radar. The report does appear to make a distinction between the "10 mile scope" and the "search" (surveillance) radar mentioned immediately afterwards, possibly supporting the hypothesis that two instruments are involved here. If this is the case then the probability that the the echoes, however anomalous in presentation, were from a real aerial target is considerably strengthened, given that surveillance and landing radars would be electronically independent and operate at very different wavelengths and pulse-repetition-frequencies (tending to rule out RFI, internal noise, component failure, anomalous propagation, multiple-trip echoes and some other possible causes of false blips). It is also relevant that the operator "scanned from 10 degs to 02 degs on search and object retained the same rtrn [return] during this opn [operation]." (This may refer to surveillance beam elevation which can sometimes be altered, either by mechanically elevating the antenna or by shifting the antenna feed, to offer "high" and "low" beams.) A spurious ground return detected on low beam can often be lost by increasing the elevation so that the amount of energy radiated at near-grazing angles required for ducting is reduced.


   In summary the radar report is less than adequate. It is not possible to definitely relate the radar target to the visual object except in the most general way, even its displayed bearing being unknown. (The report's entry against question 2 [a], which in the required Air Force format reads "Angle of elevation and azimuth of object when first sighted", is: "Not avail. GCA radar could not furn[ish] this info.") All that can really be said with confidence is that the operator observed a target which he interpreted to be an inbound object, and which was notable for the strength of its radar return, indicating unusual size. This is in essence what the intelligence report says.


   The visual reports are somewhat more detailed. According to the intelligence report the "aircraft shaped" object approached Carswell from the SW on a heading of 30 degrees. Its angular subtense was about that of a baseball at arm's length. It was dark grey in colour and appeared to be at least as large as, or possibly larger than, a B-36 (a very large bomber). It left no visible trail or exhaust, had no visible propulsion units or cabin lights and travelled silently over the tower at 3-4000', maintaining its 30 degree heading. After passing over the tower it was observed to continue the same course, the bright light in its tail visible for a further 5 minutes as it appeared to pass 5 or 6 miles N of Meacham Field near Fort Worth. It was lost to sight at an elevation angle of about 5 degrees.


   A silent, low-altitude aircraft with elliptical wings and a tail fin might suggest a glider - perhaps some type of towed aerial gunnery target that had come adrift during an exercise. The reported lighting pattern is very curious, however, and this hypothesis does nothing to improve the fit with the radar indication of a very efficient reflector since a glider would a be a light-weight, probably largely wooden, construction with a very poor radar cross-section. Further, the flight pattern is not very typical of a glider - especially an unpiloted runaway, and the estimate of size is grossly inconsistent with any known glider.


   This estimate is, however, reasonably consistent with the visual estimate of angular subtense. Allowing for the fact that angles are almost always overestimated, a "baseball" at arm's length might equate to, say, a couple of inches at about 24" from the eye, or about 5 degrees. The object was estimated to be within 3-4000' of the tower. At 4000', 5 degrees equates to about 350 feet. A Convair B-36 had a wingspan of about 230 feet. Thus, roughly speaking, these estimates are consistent with accurate observation of size, angular subtense and altitude of an object "larger than a B-36". An object of unusually large size is also roughly indicated by the radar report, which in this respect is also consistent.


   The state-of-the-art in ECM "active jamming" techniques in 1954 is uncertain, but the remote possibility exists that the object was some sort of experimental ECM drone carrying emitters capable of simulating or enhancing radar targets. Indications of rather accurate visual observation, however, argue quite strongly against this hypothesis.


   If there was in reality no "aircraft" at all but only an imaginary outline, an illusion encouraged by a fortuitous pattern of lights - perhaps a formation of jets flying at high altitude - then the silence would be explained. It is perhaps possible that the radar echoes could be explained as the unresolved, integrated returns of several high-altitude aircraft flying abreast, since the PPI would display slant range and cannot distinguish between a low-level target at 10 miles ground range and a target at, say, 40 degrees elevation and an altitude of 35,000' with a ground range of less than 8 miles. However this hypothesis could not be valid if, as appears likely, the target(s) appeared on the landing radar display since the high elevation would be shown on the height scope, and might even be above the elevation limit of the antenna. Furthermore the detailed visual observations over a period of some minutes - including visibility virtually to the horizon - cannot be accommodated without a deal of strain.


   In conclusion, this is an interesting and unusual report which would appear to merit further investigation. A great deal more detail on the radar instrumentation, scope presentations, and radar/visual times, ranges, bearings and elevations, is required to finally evaluate it. But some weight has to be given to the experience and stated reliability of the several Air Force witnesses involved, and it is only reasonable to concede that some object was seen and detected by radar. Although the case cannot be said to be probative, therefore, the rough consistency of observations by numbers of trained personnel does indicate the likelihood of some type of very quiet or silent aircraft, having abnormal size, radar cross-section, configuration and lighting, which was not identifiable by flight plan or visually despite close observation for some minutes both with the naked eye and binoculars.


STATUS: Unknown



10.  DATE: December 2, 1954    TIME: 1410 local          CLASS: R/V  ground radar/air

LOCATION:                                       SOURCES: Vallee CS 1966 186

French military

radar site,

Ceuta, Morocco

                                                             RADAR DURATION:  60 mins. approx.


EVALUATION: No official


PRECIS: A French military radar site at Ceuta tracked a target which was simultaneously observed visually by the crew of a fighter in the air. The speeds and altitudes of the target were recorded for a continuous period of about I hour from 1410 to 1510, during which time its altitude varied between 7 km (23,000') and 18 km (59,000') with variations in speed from 10 kph (6 mph) to 220 kph (137 mph). The altitude/speed diagram shows a rate-of-climb of >1500 fpm maintained for some 17 minutes (speed 100 mph slowing latterly to 40 mph) up to an altitude of nearly 59,000', levelling off with speed dropping briefly to near-zero, then accelerating steadily to 137 mph over about 12 minutes at the same altitude before suddenly dropping 20,000' in 1 minute down to 39,000' and again levelling off, still at 137 mph, for about 10 minutes, at which point the target dropped, with a simultaneous rapid deceleration, at a mean rate of about 5000 fpm for 3 minutes down to another level of 23,000' at <60 mph, maintaining this speed and level for some 10 minutes until contact was lost.



NOTES: No information is available on the radar type(s) or the nature of the simultaneous air-visual, nor are ranges and bearings given for the target, and the scope presentation is not described. Evidently at least one nodding-fan height finder was involved, but whether there was a concurrent PPI track is unknown. Nevertheless an aircraft would seem to be ruled out by the combination of altitude, somewhat abrupt manoeuvrability, and speed range. Apparently slow speeds, or even brief hovering, might be displayed on the PPI track of a fixed-wing aircraft during a radial climb at constant slant range, but not on a height-finder simultaneously (although a height-finder operated alone is susceptible to a related blind effect - see below). An altitude of nearly 60,000' was at the very limits of the state of the art for fixed-wing flight in 1954, and coincidentally it was on December 1 - the day before this incident - that Eisenhower gave his authorisation for the $19 million USAF/CIA program to develop a revolutionary plane - the U-2 - which could exceed it in extended flight. No helicopter could achieve the height/speed domain of the target. In short, if the record accurately reflects target movements, this would seem to be unbelievable performance for any fixed wing or rotor craft known to have been flying in 1954.


   It is possible that multiple-trip returns from an aircraft beyond the unambiguous range of the set could display spuriously slow speeds, since the angular rate of the target is preserved but at much less than the true range.  True radial velocities would be accurately displayed, but motion with a component normal to the line-of-sight would be displayed at spuriously slow speeds to a degree proportional to the tangential vector. On a PPI display, true tangential motion of a multiple-trip target could be reduced still further by a simultaneous steep climb, since the 2-dimensional display only indicates the change in azimuth. On an RHI scope the situation is a little different and changes in elevation would not enhance the multiple-trip effect. However, because the elevation angle is preserved it is obvious that multiple-trip returns displayed at a given altitude by the RHI scope must relate to a more distant target at a greater true altitude. Therefore an aircraft is a still less likely explanation of the target if detected on the 2nd trip. Additionally, of course, any such hypothesis would be conditional upon the nature of the undescribed air-visual corroboration of the target.


   Whether or not the RHI indications cited were supplemented by a PPI track is very important to an interpretation of the altitude/speed diagram. A nodding-fan height finder has a poor azimuth resolution, with good discrimination in altitude and range; a surveillance PPI on the other hand does well with range and azimuth but has essentially no resolution in elevation. If the diagram was derived from RHI indications alone, therefore, it may not reflect the true azimuthal vector of target motion and thus the speeds cited would be minima. This would mean that when the diagram shows the target slowing to about 6 mph, this could be merely the rate of change in range - but there could be a simultaneous undisplayed tangential vector in the plane of the resolution cell, which, for a typical beam width of about 4 or 5 degrees, would be on the order of 2 miles across at a slant range of (say) 50 miles. At the highest recorded speed of the target in this case, it would take more than 50 seconds to cross this cell laterally, during which time it would be displayed as stationary. Its echo would then disappear and the operator would be obliged to realign the antenna azimuth left or right to find it again, by which time it could have begun to turn towards a radial heading and would now show indications of increasing speed. In this way the speed diagram might show spuriously abrupt transitions and episodes of near-stationarity. Of course it is normal for a height finder to be operated in tandem with a surveillance PPI, and so it may have been in this case although no track information is presented which positively requires this to be so.


   Although angles of elevation cannot be inferred without range data, wide variations in altitude up to nearly 60,000' seem intuitively inconsistent with anomalous propagation. This is not necessarily so, however. For a radar with a range of (say) 200 miles, a target displayed near maximum range with a displayed altitude of 60,000' would represent an elevation angle of only about 3 degrees, and a diminishing amount of trapping or partial reflection might be possible up to about 10 degrees. The altitude variations of the target (representing a couple of degrees at long range) might fairly be described as erratic, and could result from sporadic echoes of ground targets beyond the operating range of the set detected due to superrefractive conditions. Such echoes might be mistakenly interpreted as a single coherent track, and a bright star or planet low on the horizon in the same rough direction - possibly exhibiting abnormal scintillation or perceptible image-wander due to mirage caused by the same atmospheric conditions - could have been seen by the visual observers in the air. Granted, such an hypothesis may not be especially probable and may imply an unusually anisotropic atmosphere (with the target[s] confined to within a maximum azimuth arc of about 20 degrees for a radar range of 200 miles, giving a track length consonant with average speeds of about 75 mph over  60 minutes), but the limited data available do not definitely exclude it. Again, confirmation that the RHI was operated in tandem with an electronically independent PPI scope - and the detailed track information from that scope - would help in assessing this hypothesis, since different operating frequencies and beam shapes would argue against similar sporadic AP echoes occurring in consistent patterns on both scopes.


   In conclusion, this potentially very interesting report deserves further investigation - particularly of radar operating characteristics, the ground track of the target and the visual sighting - but unfortunately does not support any definite interpretation as it stands.


STATUS: Insufficient information



11.  DATE: June 23, 1955  TIME: 1245 local                       CLASS: R/V ground radar/
                                                                                                                      multiple air visual

LOCATION:                           SOURCES: Thayer (Condon 143)

Utica, N.Y./ Albany,

N.Y./ Boston, Mass.                RADAR DURATION: unspecified


EVALUATIONS: Thayer - unknown


PRECIS: A Mohawk Airlines DC-3 was cruising at 3000' in good daylight visibility below a 4000' overcast, about 15 miles E of Utica, N.Y., on a heading ESE to Albany, N.Y. at 160 knots. At about 1215 both pilot and copilot saw an object come over the top of their aircraft from behind, an estimated 500' above their altitude, on a heading that made a 20-degree angle with the vertical as it crossed the windshield. They estimated the length of the object at about 150'. It was described as:


"light gray, almost round, with a center line . . . . Beneath the line there were several (at least four) windows which emitted a bright blue-green light. It was not rotating but went straight. [The lights] seemed to change colour slightly from greenish to bluish or vice versa [as the object receded]. A few minutes after it went out of sight, two other aircraft (one, a Colonial DC-3, the other I did not catch the number) reported that they saw it and wondered if anyone else had seen it. The Albany control tower also reported that they had seen an object go by on Victor-2 [airway]. As we approached Albany, we overheard that Boston radar had also tracked an object along Victor-2, passing Boston and still eastbound."


NOTES: Thayer's study of this case notes that the crew computed the speed of the object, based on the times of the contacts near Utica and Boston, at 4,500 - 4,800 mph, and he questions the "absence of a devastating sonic boom" which should have been caused by a 150' ellipsoid exceeding Mach 6 below 4000'. On this basis Thayer concludes that the Boston GCA radar report was probably coincidental, and whilst he evaluates the residue as "a most intriguing report that . . . pending further study . . . defies explanation by conventional means" the lack of a related radar track clearly must reduce the interest of the case.


   There is an inconsistency here, however. The total travel time for an object flying the 220 miles between Utica and Boston at 4,500 mph is only 3 mins., yet "a few minutes" had already passed before the crew heard reports from other aircraft and Albany control tower, by which time the object should already have been beyond Boston and probably well out to sea. The likelihood seems to be that the error lies in the estimate of speed, which is itself plainly inconsistent with the visual sighting from the DC-3 crew, who watched the object "for several miles" as it moved ahead of them, had time for a clear view and were not rocked by the turbulence of a near air-miss with a large, hypersonic body.


    Firstly, we should note that the times given in the report for the beginning and end of the above-described sequence of events, 1215 - 1245, are consistent with the DC-3's trip from Utica to Albany at approx. 160 knots. If the crew heard the report of the Boston tracking as they "approached Albany" at or near 1245, as stated, then the implied average speed of the object is in the region of 450 mph, not 4500 mph., a factor-ten error which suggests a (not unique) corruption somewhere in the Blue Book reporting chain. This value is in turn consistent with the Albany fly-past, reported "a few minutes" after the Utica encounter: at 450 mph the object would pass Albany about 8 minutes later. There does not appear to be any evidence that the object was hypersonic, and thus there is no reason to discard the Boston report on this basis.


   We are left with consistent multiple reports of an object flying between Utica and Boston at approximately 450 mph and detected by radar. The object seen from Albany control tower and two other commercial aircraft is undescribed, but presumably was unidentifiable by the reporters. The radar target is similarly undescribed, but again presumably was unidentified if only because it failed to conform to any flight plan and/or did not respond to radio interrogation.


   The Utica sighting, however, is much more circumstantial and prima facie does indeed defy explanation. The overall performance is not itself inconsistent with a jet aircraft, but the description is difficult to reconcile in this way. An aircraft flying just within the overcast might conceivably be difficult to identify, and might even create unusual cloud turbulence in its wake which would obscure its true shape, but the rather specific configuration and lighting pattern are grossly at odds with such an hypothesis.


   If the overcast were thin the lighting pattern might relate to a military air refueling tanker (always brightly lit) just above the clouds; but such an operation would never be conducted on a commercial airway, much less without warning, and the hypothesis would still leave a great deal to be added by the witnesses' imaginations. Presumably there could have been unusual towed drones and targets, as well as classified military airborne radar experiments with unusually configured radomes that might have been unfamiliar to commercial pilots in 1955, but this is highly speculative. The Grumman WF-2 Tracer was the first known Airborne Early Warning platform to use a saucer-shaped dorsal radome; the glassfiber structure was massive in relation to the aircraft and in certain circumstances might have given rise to "UFO" reports. But the prototype WF-2 did not fly until March 1957, and no hypothetical early version can easily be squared with the object described.


   In summary, the specificity of the Utica description, the independent visual reports from ground and air, the radar tracking (albeit unconfirmed), and the internal consistency of reported times, distances and headings all suggest the probability of a large, unidentified, ellipsoidal object travelling at about 450 mph low over the Eastern Seaboard of the United States.


STATUS: unknown



12.  DATE: July 17, 1957     TIME: 0430 CST(approx.)         CLASS: R/V ground-(air?)
                                                                                                                           radar/air visual

LOCATION:                            SOURCE: McDonald (Sagan, Page 1972) 56

South Central U.S.                                    Craig (Condon 1970) 56, 261

                                                                   Klass (1974) 186

                                                                   Thayer (Condon 1970) 136


                                                 RADAR DURATION: (ground) few minutes


EVALUATION: Blue Book - aircraft

                            McDonald - unknown

                            Craig -  tentative unknown

                            Klass - aircraft/meteor/stars/electromechanical fault

                            Thayer -  tentative unknown


PRECIS:  A USAF ECM (electronic countermeasures) reconnaissance RB-47H of the 55th Reconnaissance Wing, Forbes AFB, Topeka, Kansas, with a crew of six, was undergoing an extended night-time exercise involving tests of navigation, gunnery and on-board radar monitors prior to a European mission. Its scheduled course was to take it from a gunnery range in the Gulf of Mexico northward through S Mississippi to Meridian, then west across Louisiana and E Texas making practice ELINT intercepts of ground-based air defense radars and communications stations, turning N for the home run to Forbes at approximately Waco, Texas. The imaging ECM monitors displayed the relative bearings of radar sources whilst associated equipment performed signature analyses of these signals. Two types of monitor were involved: #1, an APD-4F direction-finder, with two hard-mounted wingtip antennae; #2, an ALA-6, with two back-to-back antennae scanning in azimuth at 150-300 rpm in a ventral housing. Signal processing was by APR-9 receiver and ALA-5 pulse analyser. (The #3 monitor operated outside the bandwidth in question and was not directly involved. The only role of this station was in wire-recording intercom and radio traffic.)


   The first, purely electronic, event occurred during the northward leg from the Gulf into Mississippi. A second sequence of events, visual and electronic, began when the RB-47 was westbound, on its assigned heading of 265 degrees, altitude 34,500' at Mach 0.75, over the South Central states. The first event involved passive ECM monitor #2 aboard the aircraft (the only monitor then operating, the ECM exercise-proper having not yet begun); the second sequence of events involved electronic detection by passive ECM monitors #1 and #2, air-visual observations by pilot and co-pilot from the flight deck, ground radar detection by FPS-10 air defense radar of the 745th ACWRON, Duncanville, Texas, and (reported but unsubstantiated) detection by airborne active radar aboard the RB47.


   The contemporary intelligence summary was compiled by the Wing Intelligence Officer, 55th Strategic Reconnaissance Wing, Forbes AFB, from immediately post-mission crew interrogation, real-time on-board partial wire recordings of interphone and command position conversations, real-time written notations by #2 ECM operator, and a TWX filed several hours after the incident from the 745th ACWRON, Duncanville at 1557Z (0957 CST), July 17 1957 (a CIRVIS flash-report  was also transmitted electrically from Duncanville prior to 1055Z whilst events were still in progress). Later information was contained in a 12-page Airborne Observer's Data Sheet (AISOP 2) completed by the aircraft commander Major Lewis D. Chase on September 10. Copies of some of these materials were finally forwarded on October 17 1957 from Air Defense Command Hq., Ent AFB, Colorado, to Project Blue Book, ATIC, Wright-Patterson AFB, where receipt was logged a further 7 days later on October 25. Items among these known materials which were not forwarded, and are apparently not extant, include the original Duncanville CIRVIS [Communications Instructions for Reporting Vital Intelligence Sightings] flash-report, the original ECM #2 log, and the partial wire recording of interphone and radio traffic.


   The essence of the COMSTRATRECONWG 55 intelligence summary follows (all times Zulu):


  ECM reconnaissance operator #2 of Lacy 17, RB-47H aircraft, intercepted at approximately Meridian, Mississippi, a signal with the following characteristics: frequency 2995 MC to 3000 MC; pulse width [length] of 2.0 microseconds; pulse repetition frequency of 600 pps; sweep rate of 4 rpm; vertical polarity. Signal moved rapidly up the D/F scope indicating a rapidly moving signal source; i.e., an airborne source. Signal was abandoned after observation.


  At 1010Z [0410 CST] aircraft comdr first observed a very intense white light with light blue tint at 11 o'clock from his aircraft, crossing in front to about 2:30 o'clock where it apparently disappeared. Aircraft comdr notified crew and ECM operator Nr 2 searched for signal described above, found same approximately 1030Z at a relative bearing of 070 degrees; 1035Z, relative bearing of 068 degrees; 1038Z, relative bearing 040 degrees. At 1039Z aircraft comdr sighted huge light which he estimated to be 5000 [feet] below aircraft at about 2 o'clock. Aircraft altitude was 34,500 ft, weather perfectly clear. Although aircraft comdr could not determine shape or size of object, he had a definite impression light emanated from top of object.


  At 1040Z ECM operator #2 reported he then had two signals at relative bearings of 040 and 070 degrees. Aircraft comdr and co-pilot saw these two objects at the same time with same red colour. Aircraft comdr received permission to ignore flight plan and pursue object. He notified ADC site Utah [Duncanville] and requested all assistance possible. At 1042Z ECM #2 had one object at 020 degrees, relative bearing. Aircraft comdr increased speed to Mach 0.83, turned to pursue, and object pulled ahead. At 1042.5Z ECM #2 again had two signals at relative bearings of 040 and 070 degrees. At 1044Z he had a single signal at 050 degrees relative bearing. At 1048Z ECM #3 was recording interphone and command position conversations.


  ADC site requested aircraft to go to IFF Mode III for positive identification and then requested position of object. Crew reported position of object as 10NM northwest of Ft. Worth, Texas, and ADC site Utah immediately confirmed presence of objects on their scopes.


   At approximately 1050Z object appeared to stop, and aircraft overshot. Utah reported they lost object from scopes at this time, and ECM #2 also lost signal.


  Aircraft began turning, ECM #2 picked up signal at 160 degrees relative bearing. Utah regained scope contact, and aircraft comdr regained visual contact. At 1052Z ECM #2 had signal at 200 degrees relative bearing, moving up his D/F scope. Aircraft began closing on object until the estimated range was 5 NM. At this time object appeared to drop to approximately 15,000 feet altitude, and aircraft comdr lost visual contact. Utah also lost object from scopes.


  At 1055Z in the area of Mineral Wells, Texas, crew notified Utah they must depart for home station because of fuel supply. Crew queried Utah whether a CIRVIS report had been submitted, and Utah replied the report had been transmitted. At 1057Z ECM #2 had signal at 300 degrees relative bearing, but Utah had no scope contact. At 1058Z aircraft comdr regained visual contact [with] object approximately 20 NM northwest of Ft. Worth, Texas, estimated altitude 20,000 ft at 2 o'clock from aircraft.


  At 1120Z aircraft took up heading for home station. This placed area of object off the tail of aircraft. ECM #2 continued to [receive] D/F signal of object between 180 and 190 degrees relative bearing until 1140Z, when object was approximately abeam Oklahoma City, Oklahoma. At this time, signal faded rather abruptly. 55 SWR DOI [Director of Intelligence, 55th Strategic Reconnaissance Wing] has no doubt the electronic D/F's coincided excatly with visual observations by aircraft comdr numerous times, thus indicating positively the object being the signal source.



NOTES:  After receipt of summary materials from the Director of Intelligence, ADC, some three months later on October 25 1957, Blue Book opened a file which contains minimal analysis. A preliminary evaluation by V. D. Bryant of the ATIC electronics branch, Wright-Patterson AFB, dated October 30, reads:


  This report is difficult to evaluate because there is such a mass of evidence which tends to all tie in together to indicate the presence of a physical object or UFO. With the exception of rather abrupt disappearance of returns on the electronic equipment, an indication that the object travelled at relatively high speed, there are no abnormal electronic indications such as are usually present in reports of this type - extreme speeds, abrupt changes of course, etc. These abnormal indications are usually the basis for considering anomalous propagation, equipment malfunction, etc., as responsible for the "sightings".

  The electronic data is unusual in this report in that radar signals (presumably emanating from the "object") were picked up. These intercepted signals have all the characteristics of ground-radar equipment, and in fact are similar to the CPS-6B. This office knows of no S-band airborne equipment having the characteristics outlined. Since the type equipment on the ground (at "Utah") is not known, and since there are no "firm" correlations between the ground [radar] intercept  and the sightings from the aircraft, it is impossible to make any determination from the information submitted. On the other hand, it is difficult to conclude that nothing was present, in the face of the visual and other data presented.


   The investigation progressed no further, however, and in November the then Blue Book Officer, Captain G. T. Gregory, closed the file with this note:


In joint review with the CAA of the data from the incident, it was definitely established by the CAA that object observed in the vicinity of Dallas and Ft. Worth was an airliner.


The official conclusion thus reads: "Identified as American Airlines Flight 655."


   The contents of the file reveal exactly what the CAA "definitely established". The "joint review with the CAA" consists of a letter from Roy Keeley, Director, Flight Operations and Air Worthiness, CAA, to Brigadier General Harold E. Watson, ATIC, which states:


The second incident mentioned occurred on July 17, 1957, near El Paso, Texas, and involved American Airlines Flight #655. Investigation of this incident definitely established the fact that the unidentified flying object was American Airlines Flight #966, which had previously departed from El Paso, Texas, en route to Dallas, Texas.


   The flight which was "definitely established" as the cause of this incident was #966, not #655, and this was furthermore a quite separate incident reported by the crew of Flight #655 whilst in the vicinity of Salt Flats, near El Paso, Texas. The file contains a wire report of this incident:


The American Airlines DC-6 air coach [#655] with 85 aboard narrowly averted collision near Salt Flats, Texas, in the pre-dawn darkness of July 17, 1957. Capt. Ed Bachner dived the airliner from its 14,000 ft altitude when he saw a green light ahead. Ten passengers were injured when thrown from their seats. Though the weather was clear, the crew said the other aircraft appeared without warning.


And a Blue Book file comment adds some early speculation that flight #655 encountered a fireball meteor:


July 17 - 50 miles E of El Paso, Texas - 3:30 a.m. (MST) [0830Z] Amer. Airlines Flight #655 almost collides with huge green UFO! (Shot E.) (Fireballs mounting)


The CAA thus "definitely established" that flight #655 was westbound approaching El Paso, 450 miles W of Dallas, shortly after the eastbound take-off from El Paso of flight #966, whose own ETA at Love Field, Dallas, was 1100Z (which would be about 5 minutes after the RB-47 commander's decision to abandon pursuit of the "UFO" near Dallas and turn for home). Yet incredibly, the conclusion, "definitely established in joint review with the CAA" in 1957, that the Dallas radar-visual "UFO" was flight #655, remained the official Blue Book evaluation 12 years later as at 1969 when the project was disbanded.


    By 1967 the RB-47 commander, Lewis D. Chase, happened to have been assigned as an air base UFO Officer, and in that capacity attended a conference sponsored by the University of Colorado pursuant to its UFO-study contract with the Air Force. Chase recalled the incident and an approximate date in September 1957, and staff of the UFO project, who had authority to request access to classified Air Force reports, requested that Blue Book attempt to locate records in its files. Major Hector Quintanilla, incumbent Blue Book officer, responded that there were no records of such an incident. Neither could ADC find any reference in its intelligence files or operations records (said to be routinely destroyed after 3 years), and the 55th SAC Strategic Reconnaissance Wing declared that "a thorough review of Wing History" disclosed no such incident. However interviews were obtained with Chase, with his former co-pilot James H. McCoid, and with #2 monitor operator Frank B. McClure who, like Chase, was at that time still an Air Force officer on active service. All three men "remained deeply impressed by the experience", according to CU physical chemist Roy Craig, and were "surprised" that no records could be located inasmuch as they recalled debriefing by intelligence personnel; but the evaluations by Craig and by Thayer (an ESSA radar and optical propagation specialist on the staff of the UFO project) for the Condon Report were therefore based solely on the somewhat detailed recollections of these three witnesses.


   In 1969, however, University of Arizona atmospheric physicist Dr. James E. McDonald obtained further interviews with all six of the aircrew, and in the same year the formerly SECRET case file was located in the newly-declassified Blue Book records under its true date of July 17 1957. McDonald was thus able to present new and much more detailed expositions, in the Journal of the American Institute of Aeronautics & Astronautics in July 1971, and in the proceedings [1972] of an American Association for the Advancement of Science UFO symposium held in December 1969. McDonald's evaluation and the now-public case file, together with further correspondence with Chase and McClure, became the bases of a highly influential re-evaluation of the case by avionics journalist Philip J. Klass in 1974.


   This history is of some relevance. The intelligence summary is, quite typically, concise to the point of obfuscation; but the fairly detailed narrative jointly recounted by the aircrew in 1967 is found to be supported in almost every significant respect by this official 1957 record, a fact which bears on the dependability of those observational and circumstantial details which do not appear in that summary record. For example, Craig's 1968 evaluation had remarked on McClure's, McCoid's and Chase's corroborative accounts of the event itself, but had questioned their divergent recollections of what happened afterwards. A retrospective look at this issue is educative.


   In their 1967 interviews with Craig, both McClure and McCoid recalled "intensive interrogation" by Wing intelligence personnel immediately on their return to Forbes AFB, but stated that they never heard any more about the incident after that. They also believed that no film or wire-recorded data from the ECM monitors had been taken from the aircraft, since it was merely a practice run and no film or recording wire had been taken aboard.


   On the other hand the aircraft's pilot, Chase, did not recall being "extensively" questioned when met after landing by the intelligence personnel, but did clearly recall that some time much later (possibly "weeks" later) he was required to fill in a rather lengthy questionnaire, including sketches and a  narrative account of the event, for the Air Defense Command. Further, he believed that film and wire-recorded data had been removed from the "back end" of the aircraft on landing.


   These accounts appeared suspiciously contradictory in 1967, and Craig emphasised the "serious lack of agreement" on these issues, concluding that "if" an official report was ever submitted then "it apparently is no longer in existence", and that electrical-optical recordings recalled by Chase "apparently never existed."


   An intelligence file did exist, however, and careful study of it now appears to reconcile the different accounts. According to that file Chase, alone of the crew, did indeed fill in a detailed 12-page Airborne Observer's Data Sheet on September 10 1957 - some 7 weeks later - which did indeed go to ADC intelligence, and this fact would be consistent with his not having been as extensively questioned at the time (for whatever reason) as McClure and McCoid. Also, although Chase was (presumably) mistaken as to the extent of the recorded data removed from the "back end" of the aircraft, his recollection that some wire-recordings were removed is found to be consistent with the contemporary report, which plainly states: "ECM #3 [operated by another, peripherally involved airman] was recording interphone and command position conversations." (This issue will be returned to later.)


   With this in mind it is useful to interpolate the contemporary intelligence records (1957) with the crew's collective account(s) of the incident, as given by Craig and Thayer (1968), McDonald (1972) and Klass (1974). Each interpolation is identified by date, and it should be remembered that those due to Craig and Thayer predate the discovery of the official file, whilst those due to McDonald and Klass are respectively concurrent with, or postdate, its discovery and may include details abstracted from it. The relevance of this exercise will later be apparent. (Note: The account supplied by Klass contains few differences of substance from that reconstructed by McDonald; generally speaking Klass's account differs in interpretation, reflecting the hypothesis for which he is arguing, and this interpretation is to a large extent embedded in his narrative. For these reasons Klass's exposition will be considered separately and more fully later in this report, and in the context of his hypothesis.)


    The Initial ECM Contact over S. Mississippi (prior to 1010Z, 1st North leg)


1.[1957] "ECM reconnaissance operator #2 . . . intercepted at approximately Meridian, Mississippi, a signal with the following characteristics: frequency 2995 MC to 3000 MC; pulse width of 2.0 microseconds; pulse repetition frequency of 600 pps; sweep rate of 4 rpm; vertical polarity. Signal moved rapidly up D/F scope indicating a rapidly moving signal source; i.e., an airborne source. Signal was abandoned after observation."


2[1957] "The electronic data is unusual in this report in that [the] intercepted signals have all the characteristics of ground-based radar equipment, and in fact are similar to the CPS-6B."


3[1968 Craig] "The mission had taken the crew over the Gulf of Mexico and back over South Central United States . . . . Radar monitoring  unit number two, in the back end of the B-47, picked up a strong signal, at a frequency of about 2,800 mHz., which moved up-scope while the plane was in straight flight. (A signal from a ground station necessarily moves down-scope under these conditions, because of the forward motion of the airplane.) This was noted, but not reported immediately to the rest of the crew. The officer operating this unit suspected equipment malfunction, and switched to a different monitoring frequency range. . . . the frequency received . . . was one of the frequencies emitted from ground radar stations (CPS-6B type antennas) . . . ."


4[1968 Thayer] Quoting McClure: " [This signal] had all the characteristics of a ground site - CPS-6B."


5[1972 McDonald] "Having completed the navigational exercises over the Gulf, Chase headed north across the Mississippi coastline, flying at an altitude of 34,500 feet, at about Mach 0.75 . . . . Shortly after they crossed the coast near Gulfport, McClure detected . . . a signal painting at their 5 o'clock position (aft of the starboard beam). It looked to him like a legitimate ground-radar signal, and, upon noting that the strobe was moving upscope, McClure tentatively decided that it must be a ground radar off to their northwest, painting with 180 degree ambiguity for some electronic reason. But when the strobe, after sweeping upscope on the starboard side, crossed the flight path of the RB-47 and proceeded to move downscope on the port side, McClure said he gave up the hypothesis of 180 degree ambiguity as incapable of explaining such behavior. Fortunately, he had examined the signal characteristics on his ALA-5 pulse analyser before the signal left his scope on the port side aft. In discussing it with me, his recollection was that the frequency was near 2800 mcs, and he recalled that what was particularly odd was that it had a pulse width and pulse repetition frequency (PRF) much like that of a typical S-band ground-based search radar. He even recalled that there was a simulated scan rate that was normal. Perhaps because of the strong similarities to ground-based sets such as the CPS-6B, widely used at that time, McClure did not, at that juncture, call this signal to the attention of anyone else in the aircraft. . . . He was puzzled, but at that point still inclined to think that it was some electronic difficulty."


6[1974 Klass] "Shortly after the RB-47 turned north from the Gulf . . . McClure decided to turn on his equipment and exercise it to assure that it was functioning properly before the aircraft reached Meridian and headed west. . . . As the RB-47 approached Biloxi, McClure tuned his APR-9 receiver to . . . S-band . . . and observed a signal with the familiar characteristics of a CPS-B-type air-defense radar. But, curiously, the bearing shown to this signal source was approximately 5 o'clock [and] was moving "up-scope" - in the reverse direction to normal. McClure concluded that his equipment was probably malfunctioning . . . . McClure told me that a CPS-6B radar was installed at Keesler Air Force Base, near Biloxi, and used for training electronic countermeasures equipment operators. It was operated by the USAF's Training Command . . . . It was because McClure knew  that there was a CPS-6B installed near Biloxi that he had tuned to its frequency as the RB-47 flew north from the Gulf."


   The Initial Visual Contact over Louisiana (1010Z, West leg)


7[1957] "At 1010Z aircraft comdr first observed a very intense white light with light blue tint at 11 0'clock from his aircraft. crossing in front to about 2:30 o'clock position, copilot also observed passage of light to 2:30 o'clock where it apparently disappeared. Aircraft comdr notified crew . . . ." [Aircraft location at this time recorded in Airborne Observer's Data Sheet as 32 degrees N, 91 degrees 28 minutes W, near Winnsboro, Louisiana; heading 265 degrees, altitude 34,500', true air speed 500 mph; winds W, 50 mph, weather clear.]


8[1968 Craig] "The pilot saw a white light ahead and warned the crew to be prepared for a sudden maneuver. Before any evasive action could be taken, the light crossed in front of the plane, moving to the right, at a velocity far higher than airplane speeds. The light was seen by pilot and co-pilot, and appeared to the pilot to be a glowing body as big as a barn. The light disappeared visually . . . ." "Plane's altitude: above 30,000 ft . . . . Witnesses recalled seeing . . . lights of cities and burn-off flames at gas and oil refineries below. They have no recollection of other than clear weather."


9[1972 McDonald] "They turned into a true heading of 265 degrees . . . Major Chase, in the forward seat, spotted what at first he thought were the landing lights of another jet coming in fast from near his 11 o'clock position at, or perhaps a bit above, the RB-47's altitude. He called McCoid's attention to it, noted absence of any navigational lights, and, as the single intense bluish white light continued to close rapidly, he used the intercom to alert the rest of the crew to be ready for sudden evasive maneuvers. But before he could attempt evasion, he and McCoid saw the brilliant light almost instantaneously change direction and flash across their flight path from port to starboard at an angular velocity that Chase told me he had never seen matched in all of his twenty years of flying, before or after that incident. The luminous source had moved with great rapidity from their 11 o'clock to about their 2 o'clock position and then blinked out."


10[1972 McDonald] "Immediately after the luminous source blinked out, Chase and McCoid began talking about it on the interphone . . . . McClure now mentioned the unusual signal he had received on his ALA-6 back near Gulfport, set his #2 monitor to scan at about 3000 mcs to see what might show up."


   1st ECM/Visual Contacts SE of Dallas (1030-1040Z, West leg)


11[1957] "Aircraft comdr notified crew and ECM operator Nr 2 searched for signal described above, found same approximately 1030Z at a relative bearing of 070 degrees; 1035Z, relative bearing of 068 degrees; 1038Z, relative bearing 040 degrees. [Note: bearings moving up-scope.] At 1039Z aircraft comdr sighted huge light which he estimated to be 5000 [feet] below aircraft at about 2 o'clock [about  60 degrees]. Aircraft altitude was 34,500 ft, weather perfectly clear. Although aircraft comdr could not determine shape or size of object, he had a definite impression light emanated from top of object."


12[1968 Craig] ". . . number two monitor was returned to the frequency at which the signal was noted a few moments earlier and again showed a target [sic.], now holding at the 'two-o'clock' position. The pilot varied the plane's speed, but the radar source stayed at two o'clock . . . After the UFO had held the two o'clock position through various test changes in aircraft speed, the number two monitoring officer informed the pilot that the target was starting to move upscope. It moved to a position dead ahead of the plane, holding a ten-mile range, and again became visible to the eye as a huge, steady, red glow."


13[1968 Craig] "The monitoring officer recalled that the navigator [Thomas H. Hanley], who reported receiving his own transmitted radar signals reflected from the target, not only had a target on his screen, but reported target bearings which coincided exactly with the bearings to the source on the monitoring scope. He also indicated that the officer [J. Provenzano] operating the number one radar monitoring unit, which was of a different type, having a fixed APD-4 antenna . . ., also observed the same display he observed on unit two."


14[1972 McDonald] "[McClure] found he was getting a strong 3000 mcs signal from about their 2 o'clock position, the relative bearing at which the unknown luminous source had blinked out moments [sic.] earlier. Provenzano [#1 ECM operator] told me that immediately afterwards they checked out the #2 monitor on other known ground stations, to be sure that it was not malfunctioning; it appeared to be in perfect working order. He then tuned his own #1 monitor [APD-4] to 3000 mcs and also got a signal from the same bearing . . . . as the minutes went by and the RB-47 continued westward at about 500 mph, the relative bearing of the 3000 mcs source out in the dark did not move downscope on the monitors, as should have occurred with any ground radar, but instead kept up with the RB-47, holding a fixed relative bearing. . . . Chase varied speed, going to maximum allowed power, but nothing seemed to change the bearing of the 3000 mcs source. . . . [Then] it moved upscope and reappeared visually."


   ECM/Visual/Ground-Radar Contacts during turn NW towards Dallas (1040-1050Z)


15[1957] "At 1040Z ECM operator #2 reported he then had two signals at relative bearings of 040 and 070 degrees. Aircraft comdr and copilot saw these two objects at the same time with same red colour. Aircraft comdr received permission to ignore flight plan and pursue object. He notified ADC [radar] site Utah [Duncanville] and requested all assistance possible. At 1042Z ECM #2 had one object at 020 degrees, relative bearing. Aircraft comdr increased speed to Mach 0.83, turned to pursue, and object pulled ahead. At 1042.5Z ECM #2 again had two signals at relative bearings of 040 and 070 degrees. At 1044Z he had a single signal at 050 degrees relative bearing. At 1048Z  ECM #3 was recording interphone and command position conversations."


16[1968 Craig]  "The pilot then requested and received permission to switch to ground interceptor control radar [Duncanville] and check out the unidentified companion. Ground Control in the area informed the pilot that both his plane and the other target showed on their radar, the other target holding a range of ten miles from him."


17[1957] "ADC site requested aircraft to go to IFF Mode III for positive identification and then requested position of object. Crew reported position of object as 10 NM northwest of Ft. Worth, Texas, and ADC site Utah immediately confirmed presence of objects on their scopes."


18[1968 Craig] "The pilot went to maximum speed. The target appeared to stop, and as the plane got close to it and flew over it, the target disappeared from visual observation, from monitor number two, and from ground radar. (The operator of monitor number two [and co-pilot McCoid] also recalled the B-47 navigator's having this target on his radar, and the target's disappearing from his radar scope at the same time.)


19[1957] "At approximately 1050Z object appeared to stop, and aircraft overshot. Utah reported they lost object from their scopes at this time, and ECM #2 also lost signal.


20[1972 McDonald] "Chase, in reply to my questions, said he recalled that there was simultaneity between the moment when he began to sense that he was getting closure at approximately the RB-47 speed and the moment when Utah indicated that their target had stopped on their scopes. He said he veered a bit to avoid colliding with the object, not then being sure what its altitude was relative to the RB47, and then found that he was coming over the top of it as he proceeded to close. At the instant that it blinked out visually and disappeared simultaneously from the #2 monitor and from the radar scopes at site Utah, it was at a depression angle relative to his position of something like 45 degrees."


   ECM/Visual/Ground-radar re-acquisition during turn W of Dallas (1050-1058Z)


21[1957] "Aircraft began turning [port radius around Mineral Wells], ECM #2 picked up signal at 160 degrees relative bearing. Utah regained scope contact, and aircraft comdr regained visual contact. At 1052Z ECM #2 had signal at 200 degrees relative bearing, moving up his D/F scope. Aircraft began closing on object until the estimated range was 5 NM. At this time object appeared to drop to approximately 15,000 feet altitude, and aircraft comdr lost visual contact. Utah also lost object from scopes."


22[1968 Craig] "The pilot began to turn back. About half way around the turn, the target reappeared on both the [#2] monitor and ground radar scopes and visually at an estimated altitude of 15,000 ft. The pilot received permission from Ground Control to change altitude, and dove the plane at the target, which appeared stationary. As the plane  approached to an estimated distance of five miles the target vanished again from both visual observation and radar."


23[1968 Thayer] "One of the most disturbing features of the report is [McClure's] insistence, referring to ground and airborne radars [monitors], that '. . . this would all happen simultaneously. Whenever we'd lose it, we'd all lose it. There were no "buts" about it. It went off.'"


24[1972 McDonald] "Chase put the RB-47 into a port turn in the vicinity of Mineral Wells, Texas . . ., and he and McCoid looked over their shoulders to try to spot the luminous source again. All of the men recalled the near-simultaneity with which the object blinked on again visually, reappeared on the #2 scope, and was again skin-painted by ground radar at site Utah . . . . Chase added [that] he requested and secured permission from Utah to dive on the object when it was at lower altitude . . . . He told me that, when he dove from 35,000 feet to approximately 20,000 feet, the object blinked out, disappeared from the Utah ground scopes, and disappeared from the #2 monitor, all at the same time. McClure recalled that simultaneous disappearance too."


   Final ECM/Visual Contacts (1055-1140Z, completion of turn and 2nd N leg)


25[1957] At 1055Z in the area of Mineral Wells, Texas, crew notified Utah they must depart for home station because of fuel supply. Crew queried Utah whether a CIRVIS report had been submitted, and Utah replied the report had been transmitted. At 1057Z ECM #2 had signal at 300 degrees relative bearing, but Utah had no scope contact. At 1058Z aircraft comdr regained visual contact of object approximately 20 NM northwest of Ft. Worth, Texas, estimated altitude 20,000 ft at 2 o'clock from aircraft. At 1120Z aircraft took up heading for home station. This placed area of object off the tail of aircraft. ECM #2 continued to [receive] D/F signal of object between 180 and 190 degrees relative bearing until 1140Z, when aircraft was approximately abeam Oklahoma City, Oklahoma. At this time, signal faded rather abruptly."


26[1968 Craig] "Limited fuel caused the pilot to abandon the chase at this point and head for his base. As the pilot levelled off at 20,000 ft. a target [sic.] again appeared on number two monitor, this time behind the B-47. The officer operating the number two monitoring unit, however, believes that he may have been picking up the ground radar signal at this point. The signal faded out as the B-47 continued flight."


27[1972 McDonald] "McCoid recalled that, at about this stage of the activities, he was becoming a bit worried about excess fuel consumption resulting from use of maximum allowed power, plus a marked departure from the initial flight plan. He advised Chase that fuel limitations would necessitate a return to the home base at Forbes AFB, so they soon headed north from the Fort Worth area. McClure and Chase recalled that the ALA-6 system again picked up a 3000 mcs signal on their tail, once they were northbound from Fort Worth, but  there was some variance in their recollections as to whether the ground radar concurrently painted the object."



   Saving one or two anomalies to which attention will be drawn in due course, it is evident that the >10-year-old recollections of the aircrew and the contemporary intelligence summary are in generally good agreement, as far as they each go. The crew recollections tend to be somewhat approximate as to quantitative values such as exact times, frequencies and so on, whilst building a more coherent narrative with more vivid, qualitative detail than does the terse intelligence report. In some cases, however, the crew recollections of ranges and altitudes are even exact, whilst the episodes reported by them are structurally very close to, and often identical with, the same episodes as reported in the then-SECRET intelligence summary (compare, for example, paras. 21 & 22).




            1st ECM signal, S. Mississippi


   It is notable that McClure's stated first interpretation of the first signal detected during the 1st N leg from the Gulf into Mississippi was in terms of an ordinary S-band radar signal. The near-identity with a CPS-6B output was also noted immediately by the ATIC electronics specialist, who refrained from positively identifying it as such due to lack of positive information about ground radars in the area and visual and other features of the report suggesting an airborne source. In 1967 Craig learnt from McClure that at the time he had "suspected an equipment malfunction" which somehow caused a ground radar signal to be displayed moving up-scope, and that "the frequency received . . . was one of the frequencies emitted from ground radar stations (CPS6B type antennas) . . . nearby". Later McClure told Klass that it was precisely because he knew of the CPS-6B at Keesler AFB near the coast that he had tuned to that frequency with the idea of checking out the ALA-6 before the scheduled ELINT test on the W leg. (Paras. 1 - 6)


   However, although McClure "suspected" a malfunction, and although similar anomalous behavior was in fact specified in the ALA-6 instruction manual (as Klass later pointed out) as a symptom of certain malfunctions, McClure did not conclude that the signal moving up-scope was an erroneously displayed signal from a CPS-6B. Craig related in 1967 that ground radar signals received from time to time during the later events had, in McClure's view, "confused the question of whether an unidentified source . . . was present", but that at the same time these ground signals counterindicated the hypothesis of ALA-6 malfunction:


On original approach to the area, however, a direct ground signal could not have moved up-scope. Up-scope movement could not have been due to broken rotor leads or other equipment malfunction, for all other ground signals observed that night moved down-scope. [Craig, 1968]


   McClure further explained to McDonald why, although at the time he was still "inclined to think that it was some electronic difficulty", he could not understand the movement of the signal in terms of known ALA-6 faults which might induce a 180-degree bearing error to a ground radar source. After the strobe moved up-scope on the starboard side it then "crossed the flight path of the RB-47 and proceeded to move downscope on the port side", at which point "McClure said he gave up the hypothesis of 180-degree ambiguity as incapable of explaining such behavior." [McDonald 1972]


   Klass's 1974 exposition, however, returns to this 180-degree-error scenario. At the 34,500' altitude of the aircraft, he suggested, trapped moisture could have frozen and temporarily immobilised the spring-loaded pivot arm in either of two relays which, when actuated by closure of the operator's antenna selector switch, should "tell" the display which of the two back-to-back antennae is in operation. Indeed, the ALA-6 instruction book specifically warns the operator that a 180-degree error in bearing indication can be caused in this way. When Klass contacted McClure the officer recalled that he had turned on the set with  the specific intention of looking for a CPS-6B signal from Keesler AFB near Biloxi on the Gulf coast, in order to satisfy himself that his equipment was working properly before they began the assigned test run against ADC radars on the westbound leg of their mission, due to start after the turn near Meridian. (The CPS-6B at Keesler was not an active air defense radar but a set used by the USAF Training Command for ECM training purposes, which McClure evidently recalled from his own training; ADC had none of its operational CPS-6/FPS-10 sets in the vicinity of Biloxi.) On this basis Klass proposed that the RB-47 was at a position SSE of Keesler AFB approaching the coast when this signal was first observed at a bearing of about 150 degrees ("5 o'clock"), corresponding to the CPS-6B radar site at a real bearing of 330 degrees.


   This evidence of proximity to Keesler AFB is somewhat persuasive, but also somewhat speculative. Klass states that "When I plotted the flight path . . . and the approximate bearings to the source of the radarlike signal, as McClure had subsequently recalled them for intelligence officer Piwetz [E. T. Piwetz, WIO COMSTRATRECONW 55, Forbes AFB], it became apparent that the radarlike signal could have been coming from the CPS-6B at Biloxi . . ."; but this is a little disingenuous insofar as only one specific estimated bearing (the initial one, 5 o'clock or 150 degrees) is recorded anywhere in the literature as having been offered by McClure, and there is no record of other bearings in the intelligence summary (neither Klass nor McDonald, each scrupulous to advance their arguments by appeal to original sources, list any other values or hint at the existence of other sources).


   The aircraft's position at this time is also entirely uncertain. It may have been approaching the coast SSE of Biloxi, as Klass's scenario requires, or it may "shortly" before have crossed the coast "near Gulfport" (nearly 20 miles W of Biloxi and thus >20 miles W of the longitude required by Klass), as McDonald was earlier given to understand. The contemporary intelligence report actually states that the signal was "intercepted at approximately Meridian, Mississippi", which puts the location close to the scheduled turn, 100 miles or so north from either Gulfport or Biloxi. (At the RB-47 speed this spot might be 10 minutes from the coast or a little more, and in the context of an exercise with a duration measured in hours this might just be consistent with McDonald's "shortly after crossing the coast", and is certainly consistent with his statement that Chase and McClure "were quite definite in pointing out to me that the initial ECM contact was made in Southern Mississippi", but is almost certainly not consistent with a location over the inshore waters of Mississippi Sound.)


   There appears to be no support therefore for the supposition that the aircraft's position relative to Keesler AFB at the time of the first signal can be established with any accuracy. Since McClure did not at the time consider the anomaly a reportable incident there is no record of its exact time of occurrence, and no record or estimate of elapsed time which would enable a back-calculation from the known times and positions of the RB-47 after 1010Z. Since no time-report was offered by McClure in immediately post-mission interrogation we are left with what he evidently did offer in 1957 - an estimate of location "approximately at Meridian" - together with his 1967 recollection that he had some expectation of picking up the Keesler signal prior to the ECM practice leg due to commence after the turn near Meridian. These facts are perfectly consistent with each other, since the radiation pattern of the relevant S-band vertical-center beam of the Keesler CPS-6B (main beam and lower sidelobe) would, as McClure presumably knew, extend more than 160 miles - well past  Meridian - and whether the RB-47 crossed the coast near Biloxi or near Gulfport makes no difference to the fact that it would be flying well within the main beam coverage for more than 100 miles while it crossed S Mississippi and began its turn to the W. It is also psychologically consistent that McClure would check his equipment shortly before the turn for the scheduled ECM run W from Meridian.


   In short, there is no basis for disputing the contemporary record that the signal was detected inland, "approximately at Meridian", in which case it is readily apparent that a signal moving up-scope from 150 degrees could not be due to the CPS-6B at Keesler whether the K-301 relay in the ALA-6 malfunctioned or not. However the fact that the RB-47 would have been within the radiation pattern of a CPS-6B sited at Keesler does raise the question: if the signal detected by McClure was not from Keesler, then where was the Keesler CPS-6B signal which he was hoping to detect? A simple answer to this question would be that the Keesler set was not turned on. Remember that it was not an operational air defense radar, but a Training Command set used only for ECM training purposes. McClure decided to look for it because he happened to know it was there, not because it was in any way involved in the scheduled ECM exercise. This was not a training flight, but an equipment-test against operational ADC radars and communications stations by an experienced crew immediately prior to a mission in Europe. There is no overriding reason to suppose, therefore, that the Keesler AFB training radar was up and running at that time on that particular night.


   However let us continue to assume, for the sake of the argument, that the CPS-6B at Keesler AFB was operating, that the aircraft's course took it E of Keesler, and that the anomalous signal was detected by McClure at this time. The hypothesis of ALA-6 malfunction still does not directly address McClure's statement to McDonald that the signal crossed the axis of the aircraft and proceeded down the port side of the scope. It is geometrically impossible for the same 180-degree error to cause the same ground signal to transit in this way. But it is possible that the frozen relay released itself during the observation of the signal, in which case the bearing indication would discontinuously jump 180 degrees and then progress down-scope to port. Whether this answers the description of the event given by McClure is arguable. Since McClure's first interpretation of the signal moving up-scope was in terms of precisely such a malfunction, then one would expect that this development would merely confirm that diagnosis, not cause him to "give up the hypothesis" of 180-degree ambiguity.


   It is true, however, that this description of the transit of the signal does not appear in the (admittedly brief) 1957 intelligence summary, and does not appear in McClure's first account as reported by Craig in 1967. Indeed, Craig states that McCoid and McClure both recalled that:


the target [sic.] could be tracked part of the time on the radar monitoring screen . . . but, at least once, disappeared from the right side of the plane, appeared on their left, then suddenly on their right again, with no trail on the scope to indicate movement of the target between successive positions. [Craig, 1968]


Although both men are here referring to a much later phase of the incident near Dallas one would have to say that some confounding of distant memories is  possible, and this may be circumstantial evidence of intermittent ALA-6 failure which (it is not ruled out) may have occurred during observation of the first signal. This issue therefore remains unresolved, despite McClure's rather specific recollection of a signal transit as given to McDonald. It is possible that Craig's necessarily-limited early inquiries simply failed to elicit this information, or that it escaped emphasis in his brief 2-page account for the Condon Report; nevertheless, whilst giving due respect to McClure's testimony, a highly pertinent detail of which there is no record prior to 1969 should not perhaps be allowed too much weight in the argument, and given that the signal did cross the scope to disappear on the port side aft it is still possible that the signal was not observed to transit continuously.


   With some reservations, therefore, it is possible that the signal behavior could be consistent with a malfunctioning relay which corrected spontaneously during the observation, and the principal remaining difficulty is finding a CPS-6B source consistent with the reported position of the aircraft.  The location of the aircraft at this time, as has been shown, is somewhat uncertain but appears to have been over southern Mississippi. Given this uncertainty some "fudging" might be justified in order to rescue Klass's plausible identification of Keesler AFB as the signal source. The near identity with a CPS-6B output, the aircraft's flight path which at all events passed not very far from Keesler AFB, and the fact that Keesler was the site of the only CPS-6B (or similar FPS-10) set anywhere in the area, all suggest a strong prima facie likelihood that this radar was the source. Therefore one might allow Klass the benefit of the doubt and follow him in relocating the incident "near Biloxi", and further allow that the aircraft could have been "approaching Biloxi", that is, just coming up to the coast a little to the SSE of the radar instead of "shortly after crossing the coast" as McClure had earlier recalled to McDonald. It is not too unreasonable to suppose that McClure, enclosed in the back end of the aircraft, could have mistaken their position by a few miles (eliding meanwhile the contemporary record that the event occurred nearly 100 miles north near Meridian).


   However a position just off the coast can be shown to be inconsistent with the S-band vertical-centre radiation pattern of a CPS-6B which might have been operating at Keesler AFB. (Note: only this beam of the CPS-6B is of the appropriate 3000 mHz frequency. Klass himself explores this pattern in some detail when considering the relationship of the identical FPS-10 coverage to later events near Dallas.) At the RB-47's altitude of 34,500' this is (in plan) a triple concentric annulus pattern with the inner edge of the innermost annulus due to the upper sidelobe falling at a ground range of some 28 miles from the antenna. This is a thin ring approximately two miles broad, encircled by a  7-mile null zone, beyond which the annulus formed by the coma lobe and main beam (which are continuous) commences at 37 miles. This main beam annulus extends to a ground range of some 120 miles, followed by another 20-mile null beyond which occurs the outermost annulus some 15 miles wide due to the lower sidelobe. Thus, a 150-degree bearing from Keesler AFB intersects the upper sidelobe more than 20 miles off the Mississippi coast, which is therefore the closest ground range at which the ALA-6 could have been detecting this signal due to the zenithal radar shadow. This lobe, however, is so narrow that the northbound RB-47 would have crossed through it in something like 20 seconds; and given the 4 rpm scan rate of the CPS-6B (with the beam rotating towards the aircraft only once every 15 seconds) it is improbable that McClure would have chanced to detect this signal at all, virtually impossible that he could have observed it more then twice (i.e., an insufficient number of times to determine a "rapid" up-scope motion), and  certainly impossible that he could have had the time also to examine its frequency, pulse length, and p.r.f. on his ALA-5 pulse analyser.


   The signal would therefore have to be that of the main beam, which a 150-degree bearing would intersect at a ground range from the coast of more than 30 miles, and because the aircraft must remain within this main beam coverage for a significant number of 15-second antenna revolutions before flying into the null zone and losing the signal it is plain that its position when the signal was first detected would have to be significantly further S still. If this position were 50 miles from the coast, the RB-47 would be flying N within the main beam for a further 20 miles or so, or about 2 minutes at Mach 0.75, during which time the monitor would be able to receive perhaps ten scans of the signal moving upscope by about 20 degrees to 130 degrees - still well-aft of the starboard beam - where it would disappear; and this should perhaps be considered the minimum duration and movement compatible with the intelligence report of a signal "moving rapidly up the D/F scope", not to mention McClure's own later testimony. (In order for the signal to progress up-scope as far as 90 degrees - abeam to starboard - or further, requires a flight path far enough east of the Keesler radar to avoid the null and remain in the main beam, implying a first-contact position some 70 miles out over the ocean on a heading which would probably take the aircraft N into Alabama rather than Mississippi.)


   These values are in very serious conflict indeed with the crew's statements and the contemporary intelligence report, which consistently implicate a location over land during the N approach to Meridian, Mississippi. Despite the attractiveness of the Keesler CPS-6B as the source, therefore, a "fudge" of some 150 miles or more seems difficult to justify. Further, a map of the mission derived from Chase's 1957 Data Sheet [McDonald 1972] shows that at 70 miles from the Gulf coast the RB-47 had only just completed its turn from the gunnery range onto the 1st north leg. This indicates that the (Keesler) source would have been detected by McClure very shortly (a few minutes at most) after this turn and the gunnery/navigation exercise, and when interrogated upon landing he would be expected to recall this recent manoeuvre as the most natural time-reference. One might expect to find, therefore, that the record based on this debriefing would give the position as "shortly after the gunnery exercise", or "over the Gulf" or even "south of Biloxi", but presumably not "approximately at Meridian, Mississippi" which is given as the reference for the turn onto the west leg of the mission.


   In summary, therefore, the inconsistencies introduced by the hypothesis that the up-scope signal came from Keesler AFB are unattractive, and in the absence of evidence that the training radar at that site was operational in the early hours of July 17 1957 Klass's argument is not wholly persuasive and the signal source should probably be considered as yet unidentified.



2nd ECM episode, East Central & Northeast Texas


   When a similar signal was later detected close to the Louisiana-Texas border, some minutes after visual observations of the rapid, bluish-white light from the flight deck, the plane was once again in straight flight (this time on a heading of 265 degrees) and, according to the intelligence report, once again the bearings to the source moved up-scope, this time remaining to starboard of the aircraft:  at approximately 1030Z it was being displayed at 70 degrees; at 1035Z it had shifted slightly to 68 degrees; three minutes later at 1038Z it had moved up to 40 degrees, staying at that bearing for a while, and it was at about this time (1039Z) that Chase and McCoid saw the "huge light" to starboard at a visually estimated bearing of "about 2 o'clock" [60 degrees]. This sequence appears to corroborate - in its essential features - the account given by Chase, McCoid and McClure to Craig in 1967, before the contemporary record was known: ". . . the number two monitoring officer informed the pilot that the target was starting to move up-scope. It moved to a position dead ahead of the plane . . . and again became visible as a huge, steady, red glow."


   By this stage, recalled McClure, he had already checked the performance of his ALA-6 on other known ground radars back near Meridian and it appeared to be working perfectly when the 70-degree signal was picked up at about 1030Z. According to Provenzano's statement to McDonald, after he and McClure now checked the ALA-6 again, he tuned his own #1 APD-4 fixed-antenna monitor to the same frequency and obtained a signal which confirmed the bearing. This was also recalled by McClure in 1967; he told Craig that "the officer operating the number one radar monitoring unit, which was of a different type . . . also observed the same display he observed on unit two." A recurrence of the hypothesised ALA-6 relay failure could not therefore account for the up-scope motion recorded at this time. Additionally it appears from Klass's inquiries to ADC that there were no CPS-6/FPS-10-type radars operating to the port side aft of the aircraft (in S Louisiana), and by this time the CPS-6B at Keesler AFB would have been well out of range - even if it was operating.


   However there was an FPS-10 at Duncanville near Dallas (site "Utah"), broadly forward and to starboard of the 265-degree course being flown by the RB-47, and Klass proposes that the signal came from this site. One's instinct is to agree that this seems highly plausible, and one can show that a heading of 265 degrees from near Meridian (fixed by the known map coordinates of the 1010Z visual near Winnsboro) at 500 mph airspeed, with an approximately 50-mph headwind leading to a true groundspeed of about 450 mph, would place the aircraft approximately level with Timpson, Texas by the time of the first signal contact at about 1030Z. Timpson is about 160 miles from Duncanville, very close indeed to the point at which the RB-47 would have entered the FPS-10's vertical-center lower sidelobe at its cruising altitude of 34,500'. Given a signal with near identical characteristics to those of an FPS-10's vertical-center beam, the probability of all these various parameters matching by chance is presumably negligible and one has to conclude that the signal was very probably related to the output of the FPS-10.


   There remain some inconsistencies, however, for which no easy explanation currently exists. The positions of the aircraft at various times during this leg can be inferred with some accuracy, as is here confirmed by the close match between the projected course and the first detection of the Duncanville signal. The ALA-6 had recently been checked on other known radar sites and its accuracy was at this time reaffirmed by a further check and the simultaneous corroboration of the independent APD-4 monitor. Yet the true bearing to Duncanville at this time was 30 degrees, whilst the relative signal bearing was recorded as 70 degrees. (A bearing of 70 degrees to Duncanville would place the aircraft at approximately Teague, Texas, a further 115 miles and 15 minutes of flight time beyond Timpson, which is quite inadmissible.) Therefore one is led to suppose either an error of 40 degrees in observation of the monitor(s), at least  by McClure and probably also by Provenzano, or a typographical error in the intelligence report, which is certainly not unheard-of. (Note: the rough signal bearings indicated by Klass on his chart of this portion of the flight path [1974, pl.16] are in error by 10-15 degrees favouring the direction of Duncanville.)


   Five minutes later at 1035Z the aircraft would have been over the Angelina River approaching Rusk, Texas, and the bearing to Duncanville had moved downscope to 40 degrees, a one-third increase in starboard displacement; but the signal bearing had changed very little, and if anything had moved up-scope a fraction to 68 degrees - a figure which, incidentally, would imply that the values were being read off the bearing ring with some care, since it is in the nature of the ALA-6 display that its broad fan of closely-spaced strobes, painted with only limited persistence on the tube phosphor once in each 15-second revolution of the source antenna, means that more-than-casual attention is required to estimate a displacement of only 2 degrees. The implied error here - whether observational or, again, typographical - is 28 degrees.


   By 1038Z, with the aircraft passing the town of Rusk towards Palestine, the bearing to Duncanville had dropped back to 45 degrees and the signal bearing was still moving up to meet it, reaching 40 degrees at this time. Some sixty seconds later, Chase and McCoid first saw the "huge light" off to starboard (which was to remain in visual contact for 11 minutes). About 1 minute later when the aircraft was approximately at Palestine, Texas, at 1040Z, the bearing to Duncanville had now fallen back to 50 degrees but the 40-degree signal was still there, having remained constant for some two minutes.


   At this time a second signal briefly appeared at a bearing of 70 degrees; but two minutes later at 1042Z, when the aircraft was passing Palestine and heading towards the area of Fairfield and Teague, the strobes indicating the second source had disappeared, and the remaining source was now still further up-scope at 20 degrees, with Duncanville now well aft of this bearing at about 55 degrees. According to the intelligence report it was at this time, 1042Z, that Major Chase, having requested and received CAA permission to deviate from the 265-degree flight plan and having requested radar assistance from ADC Duncanville, "increased speed to Mach 0.83, turned to pursue, and object pulled ahead." (Klass's sketch map, in contradiction to the officially reported sequence [see para.15 above], shows this turn well underway before 1042Z and also indicates a signal-bearing at this time of about 40 degrees instead of 20, which errors jointly create a bearing close to that of Duncanville. The true mis-match here would appear to be 35 degrees. Note that the signal bearing, from an initial angle of 40 degrees east of Duncanville, has now swung to an angle 35 degrees west of Duncanville, moving in an opposite sense to the relative down-scope progression of Duncanville.)


   This sequence of events from 1010Z to 1042Z was qualitatively described by members of the crew [e.g., paras. 12 & 14] in a way which one can see is quite accurately supported by these 1957 figures. The source appeared to remain at an essentially fixed bearing for some minutes, then began to move up-scope by 30 degrees, at which time visual contact was made from the flight deck. The source then moved up-scope a further 20 degrees, at which point Chase turned to starboard in pursuit onto a heading of approximately 320 degrees true.


   During this turn the bearing to Duncanville would swing rapidly towards the bow, but by 1042.5Z the signal bearing was swinging back in the opposite direction, remaining to starboard of the aircraft at 40 degrees (at this time a second signal was detected, as once before, on a bearing of 70 degrees, remaining on-scope for about 1 minute before disappearing). The RB-47 was now traveling at Mach 0.83 on a NW heading which would take it between Dallas and Fort Worth, approximately over Arlington, Texas, only a few miles from Duncanville. The Duncanville radar site was now, at 1044Z, only fractionally off the bow at a range of about 70 miles, but the signal source, rather than narrowing the angle with the bow towards zero, was now at 50 degrees.


   By approximately 1048Z the aircraft, still at 34,500' and travelling at Mach 0.83 (TAS 553 mph, estimated ground speed approximately 530 mph, wind 50 mph at 50 degrees off port bow) would have been leaving the main-beam/coma-lobe coverage of Duncanville's FPS-10 to enter the null zone in which no signal from the FPS-10 could be detected. (About 1 minute later it would pass rapidly through the narrow upper sidelobe in a matter of seconds, at which point the ALA-6 could have received a brief signal if the FPS-10 antenna happened to be pointing to the aircraft at the right moment.) The intelligence report states that McClure lost the signal off his #2 monitor at 1050Z, which seems close enough, once again apparently reaffirming the accuracy of the projected flight path and the evident relationship between the Duncanville coverage pattern and the detectability of the ALA-6 signal.


   However at this time this spatial relationship starts to deteriorate. Having passed by Duncanville the aircraft began the port turn which was to take it N of Fort Worth towards Mineral Wells, and the signal was reacquired at 160 degrees. At 1052Z this signal had moved clockwise, aft of the aircraft, to a position of 200 degrees, and initial reacquisition must therefore have been at some time prior to 1052Z. Klass estimates a time of approximately 1051Z and indicates a map location which accords with the aircraft's brief re-emergence into the upper sidelobe NW of Duncanville. But if this is correct, then at the time of the 1050Z signal-loss one minute earlier the aircraft would already have been over half way across the radar shadow cone. This zenithal "blind" zone over Duncanville is just over 50 miles across for the RB-47's projected course and altitude as reconstructed by Klass, and the signal should have remained absent for at least 6 minutes if the final contact south of Duncanville occurred with the upper sidelobe, or at least 7 minutes if (much more probably) it occurred with the main beam. (Note: the aircraft had overflown the visual "object" when it "appeared to stop" at the time of the 1050Z signal loss, and would thereafter have been slowing from its Mach 0.83 pursuit into the port turn which would bring it back over the area. Consequently Klass' model itself requires that the estimated 6-7 minute signal hiatus should be considered a minimum.)


   This discrepancy invites closer scrutiny of the times and positions indicated on Klass' chart. It becomes evident that during the crucial period 1042-1051Z the aircraft is shown with an average ground speed of approximately 720 mph, even exceeding 750 mph in the 2 minutes from 1042Z. These excessive rates arise from the attempt to correlate the 1051/1052Z reacquisitions with the known coverage pattern at 34,500', and Klass, evidently aware of the introduced inconsistencies, therefore omits to indicate on his chart the one time (signal loss at 1050Z) which would immediately have drawn attention to them.



                                                                           ------------end of Part 1 ---------------