Archive Index


Analysis of the Intercepted Targets over Lakenheath - 1

An Opinion by Martin Shough

Track E. The BOI-485/Perkins/Wimbledon scenario

Discussion of the reported radar contacts will be offered in terms of a number of hypotheses:

a) Aircraft
b) Anomalous Propagation
c) Specular Partial Reflection from Elevated Layers
d) Scatter from Clear Air Structures
e) Echoes from the Moon
f) Balloons/ other Windborne Objects
g) Meteors, Precipitation Cells, Lightning Channels/Sferics, Auroral Ionisation, Birds & Insects
h) System Noise, Component Failure, Remote RFI & Deception Jamming (spoofing)
i) Ghost Echoes
j) Jet Exhaust

This episode is the core of the event due to the several radars reported to be involved and the singular relationship of the reported echo movements to the position of the concurrently tracked interceptor. Track E is therefore not simply a radar track but a complex event reconstructed from several partial accounts of radar contact from different fixed and mobile locations. This is of course the source both of its interest and of its occasional ambiguity. The prima facie impression is of multiple redundant contacts with the same target, whose physical location in the air appears to be confirmed by the remote physical locations of the ground radar sites (at Neatishead and at Lakenheath) and by the simultaneous contact on the airborne interception (AI) radar.

BOI-485 clearly indicates that GCA radar at Lakenheath, in addition to the RATCC radar there, also detected what was believed both by the operators and by investigating intelligence personnel to be the same target. However the extent of concurrency here invites closer inspection, since it can be argued that BOI-485 is not sufficiently unambiguous to certify that GCA detection was concurrent during the actual interception episode (even though the context implies this, and even though the RATCC Watch Supervisor reports that confirmatory information was passed to him from GCA and relayed by him over the conference line during the event). BOI-485 states:

That three radar sets picked up the target[s] simultaneously is certainly conclusive that a target or object was in the air.

The three radars referred to would appear at first sight to be the RATCC CPS-5, the GCA CPN-4 and the AI radar on the Venom, which would mean that 2 ground radars at Lakenheath displayed a target concurrently with GCI at Neatishead and the Venom's AI radar - 4 simultaneous contacts - inasmuch as the Venom's AI radar contact only occurred during the interception episode also tracked by Neatishead. It is possible, however, to interpret this statement to mean the CPS-5, CPN-4 and TPS-1D radars (there appears to be every likelihood that the "TS-ID" radar equipment listed in BOI-485 was actually a TPS-1D Army radar operated by the 60th AAA), and the statement that "three radar sets picked up the target simultaneously" could therefore be interpreted as referring to one Army and two Air Force radars on the ground at Lakenheath.

This interpretation is quite speculative, of course. If it were correct then BOI-485 could no longer be interpreted as confirming that the Lakenheath RATCC and AI radar contacts during the interception were confirmed "simultaneously" by Lakenheath GCA. On the other hand this is arguably the interpretation favoured by the context, and is consistent with Perkins' statement that telephone cross-correlation with GCA was on-going during the affair. The "observation" by TPS-1D cannot be established as concurrent in either case, but the undisputable involvement of Army radar should be noted at this stage as a significant point warranting further investigation.

Howsoever, three electronically independent ground radars at Lakenheath alone are stated to have been involved in the event as a whole, and at least one is confirmed by two independent sources as having tracked the target during the interception; the same two sources confirm that the Venom's AI radar detected the same target simultaneously; the AI radar target is triply confirmed by a third independent source (the RAF Controller's statement) which also introduces ground radar corroboration of the same target from a further remote site. If such a three-point radar fix in three dimensions could be confirmed it would indeed be strong evidence that, as Capt. Stimson put it, "a target or object was in the air".

Some reported target positions and movements near Lakenheath
Orange vectors are from BOI-485, blue vectors from Perkins' report. The differing intercept positions in the two sources are shown as B and P respectively. The diagram is merely schematic but illustrates the very diverse headings reported at different times.


Analysis of the ground and AI radar contacts

a.) Aircraft

Indications favouring an aircraft as the cause are several: 1) The multiple ground and air contacts appear to triangulate a common location in three dimensions (or four, counting concurrency in time); 2) The consistency and strength of target presentation, and the continuity of tracking, indicates a solid, radar-reflective target with a radar cross-section comparable to that of a jet fighter; 3) The impression of rational behaviour, particularly in relation to the arrival of the Venom, is suggestive of a controlled vehicle.

According to the RAF Chief Fighter Controller, the target was personally observed by him throughout the interception episode (as well as by both his Interception Teams) and appeared as a normal-looking, strong echo which did not fluctuate. He described its presentation as "similar" to the echo of the Venom and "consistent in strength." (With reference to the question of MTI gating and clipping mentioned in Section b. below, note that none of the Neatishead radars was equipped with MTI.) "The target was obviously under control the whole time," he said, "and could thus change its speed at will." The Lakenheath RATCC Supervisor stated: "It is interesting to note that at no time did we lose radar contact with the unidentified target throughout its maneuvers. Ghosts, MTI vagaries, skip range targets and other atmospheric phenomena that cause false targets are noted for inconsistent and weak returns as well as blooming targets." According to BOI-485 the target on the interceptor's AI radar was also reported to be very clear. The RAF Fighter Controller stated that the pilot reported the code "Judy", meaning that his navigator had acquired the target solidly on both the PPI and C scopes of the Mk.21 AI radar.

Arguments against an aircraft are rather strong, however. The target towards which the Venom was vectored had been tracked for some time by Lakenheath, and according to the RATCC Supervisor its motion was not at all similar to that of an aircraft, being a sequence of steady, high-speed linear tracks punctuated by stationary episodes of up to about 6 minutes with non-inertial stops and starts. BOI-485 is ambiguous as to times and sequences but repeatedly describes the same behaviour: "Flight path was straight but jerky with object stopping instantly and then continuing." BOI-485 alludes to some other targets reported at different times and homogenizes the descriptions in a way which, without a coherent timetable, is made still more confusing by some ambiguous English; however Captain Stimson's summary concentrates on one primary "target or object", and context indicates that this singular target is the intercepted target as described by Perkins: "When the target would stop on the scope, the MTI was used. However, the target would still appear on the scope." Again: "The fact that three radar sets picked up targets [i.e., echoes] simultaneously is certainly conclusive that a target or object was in the air. The maneuvers of the object were extraordinary; however . . . radar and ground visual observations were made on its rapid acceleration and abrupt stops . . ." [emphases added]

It can be maintained that because a precise, time-flagged schedule is not offered these statements are not strictly corroborative of Perkins' account and could refer to another target. BOI-485 does not indeed specify unambiguously that the intercepted target engaged in these non-inertial manoeuvres. However the objection seems churlish, particularly given that the one internally consistent description of a target movement reported in BOI-485 which is time-flagged is closely corroborative of Perkins' account, in which it is stated that this target was the one which was later intercepted.

That single instance is the position of the target when first observed. According to Perkins it was first noticed on the RATCC scopes as a stationary echo at 20-25 miles southwest of Lakenheath, where it remained stationary for "several minutes". According to BOI-485: "GCA and Air Traffic Control Center reports radar tracking from 6 miles west to about 20 miles SW where target stopped and assumed a stationary position for five minutes."

The initial plot evidently refers to detection on the GCA CPN-4. Lakenheath GCA was reportedly the first to be alerted, then they alerted the 60th AAA Command Post and Sculthorpe GCA, and the RATCC was apparently alerted later. If the RATCC were not yet looking for UFOs and scopes were set at maximum pick-up range, then a target at 6 miles could have been inside the 7-mile minimum operating range at that range scale and would not have been observed. Subsequent movement to 20 miles SW would bring the target into the radiation pattern of the CPS-5 where Perkins states it was first called to his attention by an RATCC controller because it was stationary despite the use of MTI. He then called the GCA unit who "confirmed the target was on their scopes in the same geographical location".

Both radars held the same plot for "several minutes". BOI-485 confirms that GCA also reported this target, confirms its geographical location, confirms (though in a different passage) that it stayed on-scope despite the use of the MTI filter, and confirms the duration of its stationarity.

The direction of subsequent movement is described by Perkins as NNE and by BOI-485 as NW. Here BOI-485 is untrustworthy because, as has often been pointed out, a direction NW from a position SW is a heading away from the station, not "into the station", and could not possibly bring the target to a position "two miles NW of station". One suspects clerical confusion of reported position with reported heading. Perkins' first-hand account, however, is geometrically consistent: "As we watched the stationary target started moving at a speed of 400 to 600 mph in a North Northeast direction until it reached a point about 20 miles North Northwest of Lakenheath." It is reasonable to suppose that sloppy transcription from the original reports into BOI-485 has rendered "20 miles" as "2 miles" (which would again have been inside the CPS-5 minimum range) and confused the headings. But it seems evident despite this confusion that both sources are referring to the same movement of the same target which, broadly speaking, headed north past the airfield and stationed itself again at a position some miles to the northwest. "There was no slow start or build-up to this speed," emphasises Perkins, "it was constant from the second it started to move until it stopped."

To the extent that sense can be made of BOI-485 it supports Perkins' account, and the reported movements are qualitatively very unlike those of a fixed-wing aircraft. There is some arguable qualitive resemblance to a rotor craft, but the order of speed and initial acceleration reported are grossly excessive for any helicopter.

Perkins' statement that the target travelled at a displayed speed of 400-600 mph "from the second it started to move" is obviously a figure of speech, but it does raise the question of how sensitive the "sampling" rates of the two Lakenheath radars would be to rapid changes of speed. Are the sweep rates fast enough to certify that the "instant" stops and starts also reported in BOI-485 could have been properly discriminated?

At its maximum rate of rotation* the CPS-5 would illuminate a target once every 9 seconds, 2 of beamwidth passing it in about 1/20 second, each echo having a persistence of several seconds on the tube phosphor. The CPN-4 beam would illuminate the target for about 1/50 second every 4 seconds. The azimuth resolution of both antennas would be approximately 3500' at 20 miles range, and range resolution would be on the order of 500' for the CPN-4 (pulse length 1 microsecond), >1000' for the CPS-5 (based on a probable pulse length >2 microseconds). The reported target speed of 400-600 mph represents a travel of 1 - 1.5 miles in 9 seconds (CPS-5), or 0.4 - 0.66 miles in 4 seconds (CPN-4), with a significant radial component of motion on a NNE inbound heading. Therefore the more pertinent dimension of the resolution cell is that of range, and the pulse resolution of both radars is quite small in relation to the displacement of this target from scan to scan: On the CPN-4 the displacement would be about 5 times the resolution, and on the CPS-5 perhaps 6 times the resolution.

* NOTE: The CPS-5 was capable of sector scan and manual position control, so that by bore-sighting the target the rate of renewal of the echo would be effectively continuous; however Perkins did not believe that the sector scan facility was in use, having probably been disconnected during modification as irrelevant to the needs of Air Traffic Control. He did not recall that the antenna was ever at any time used in a tight scan or "searchlight" mode by any of his controllers, and thought it was probably left in continuous rotation throughout.

The practical limit of discrimination on the PPI would depend in part on the range scale selected. Before RATCC acquired their target Perkins said that he "had each scope set on a different range - from 10 miles to 200 miles radius of Lakenheath", and subsequently "had most of the scopes on short range, so we could watch the UFO closely on the smaller range." If during the 5 minutes of subsequent stationarity this 20-mile target was being observed on (say) a 30-mile range scale then, given a typical spot resolution of about 0.5% of the tube radius, a blip displaced radially by its own diameter would represent a travel of about 800' and a 500 mph target would therefore move more than 8 spot diameters across the PPI between sweeps. Operators would be quite used to estimating target speeds from scan-to-scan spot displacements, and the difference between, say, a 100 mph movement and a 500 mph movement would be substantial.

Therefore when Perkins stated that "there was no slow start or build up to this speed - it was constant from the second it started to move until it stopped" he was saying that the first displacement was about equal to each subsequent displacement, and these displacements would be clearly discernible on the PPI even at a range scale significantly larger than 30 miles. The selected range-scale of the CPN-4 is unknown, but assuming a maximum of 60 miles a near-radial displacement of 3000' per 4-second sweep at 500 mph represents about 2 resolvable spot diameters on the PPI, and the echo should therefore be wholely and perceptibly moved with each scan of the antenna. In other words, both the CPS-5 and the CPN-4 are capable of displaying measurable movement of a 500 mph target within a single sweep, consistent with the "abrupt" and "immediate" stop-start motion observed on both radars according to BOI-485 and Perkins.

The actual acceleration during the initial sweep cannot of course be derived independently of these reports. However, a total travel of about 35 miles from a start point 20 miles SW to a stop 20 miles NNW equates to a little more than 4 minutes at 500 mph, or some 38 sweeps on the CPS-5 and about 63 sweeps on the CPN-4. According to Perkins this was "a calculated speed based on time and distance covered on radar" and there would have been some 38 opportunities to cross-check the scan-to-scan spot displacement on the CPS-5 with the speed calculated from the overall track-length/time. Therefore there is no good reason to doubt the rough validity of Perkins' estimate. (BOI-485 contains no specific estimate of target speed for this run; the only figure offered, 600-800 mph, relates to a different episode. But the GCA CPN-4 tracks are accorded the general characteristic of "travelling at terrific speeds and then stopping and changing course immediately." Stimson's summary refers to the "rapid acceleration" of the primary target.) However, being as conservative as possible, we may conclude that from a displayed ground speed of 0 mph the target achieved a displayed speed in the region of 500 mph at least at some point on the track within a distance of some 35 miles, and within a period of less than about 4 minutes.

This is quite important, because any conventional aeroform requires that an allowance must be made for inertial deceleration to zero during the same track, so that the minimum implied acceleration is greater still. Even if the measured speed of about 500 mph was only achieved momentarily as the peak of this curve at the mid-point of the track, the object is required to accelerate to this speed from zero in about 120 seconds. The order of speed reported would only have been achievable by a jet fighter, and even to begin to approximate the reported "instant" acceleration and "constant" velocity presumably requires at least a good fraction (say 50%) of peak velocity to be have been achieved within, say, three 4-second sweeps of the CPN-4 - that is, zero to at least 2-300 mph in about 10 seconds - a performance of which no helicopter flying in 1956 was remotely capable.

It is possible in rare circumstances for a fixed wing aircraft to maintain roughly constant slant range at constant azimuth on a PPI for a brief period, thus being displayed as stationary. A steep tangential climb or dive normal to the antenna line of sight might conceivably show little or no movement above the resolution of the PPI whilst having the small range rate needed to defeat the moving target indicator. But this geometry is highly improbable even once, and the likelihood of this occurring twice on two different radial headings more than 90 apart, with sudden manoeuvres at the termini of a straight track and with the sensitive geometry maintained for 5 minutes within a range resolution of about 500', is at best negligible. When cognate behaviour continues to be observed repeatedly over the next hour or more the probability is so low as to be effectively nil.*

* NOTE: Analogous behaviour was also described - whether simultaneously or not is unclear from BOI-485 - by visual ground observers: "Objects as seen by ground observers and GCA radar has feature of travelling at terrific speeds and then stopping and changing course immediately." The visual sightings are discussed separately here.

No mechanism involving multiple-trip echoes could cause a remote aircraft to be displayed "in the same geographical location" on two different radars with very different unambiguous ranges, and the displayed speeds of multiple-trip targets must always be equal to or less than their true speed. Reduction in displayed speed is proportional to true range and the tangential component of true motion. Five minutes of displayed stationarity could only be explained by true stationarity, and the displayed speed of 400-600 mph would be less than the true speed of the target.

Aircraft detected in sidelobes are equally incapable of explaining any of the reported features of the target.

The implied performance of the target during the attempted interception by the jet fighter is obviously inconsistent with a helicopter in terms of both speed and agility. Perkins recalled that

The first movement by the UFO (circling behind the interceptor) was so swift that I missed it entirely but it was seen by the other controllers . . . . The pilot . . . tried everything - he climbed, dived, circled etc but the UFO acted like it was glued right behind him, always the same distance, very close but we always had two distinct targets.

A high-performance jet fighter might be able to evade and pursue the interceptor in a manner qualitatively similar to that described, but the target was stationary on the PPI for a significant time before falling into trail behind the interceptor. Also a jet could not subsequently stop following and remain stationary for a substantial period as also described.

The subsequent stationarity of the target was not observed by the Chief Fighter Controller at Neatishead, because when the target as seen by Perkins' team stopped after the interception the interceptor had by then been released from its mission by GCI, the target having been lost from the Neatishead scopes: "The target appeared to 'give up the chase' as though it had achieved its objective and more or less 'melted'," Wimbledon said. "My own theory is that it either went straight up at very high speed or down to ground level under our radar cover. East Anglia is notoriously flat and not much above sea level." The latter interpretation would be not inconsistent with continued detection by radar at Lakenheath given that the area of the pursuit ("about 10-30 miles all in the southerly sectors from Lakenheath" [Perkins]) would be some 50 - 80 miles from Neatishead, at which ranges a 1.0 radar horizon would be in excess of about 6000' (and if the Type 7 master radar alone was in use at this point the figure at 80 miles could be 15,000' or even higher for a small target; see here for discussion of radar coverages). Even quite small vertical movement on the part of a target close to the radar horizon could quickly cause the signal to drop below the noise threshold so that it "melted away", whilst still being within the radiation pattern of the nearer radar at Lakenheath.*

*NOTE: The Lakenheath radars painted the target even when it appeared stationary, despite the use of moving target indicators. But might GCI signal loss be explained by a more sensitive MTI filter? No, because none of the Neatishead GCI radars was fitted with MTI.

It has to be accepted that Wimbledon has not always seemed whole-hearted in his endorsement of the extremes of target behaviour observed at Lakenheath before and after the interception and "tail chase" episode. Although he has explicitly endorsed this report of sudden stops and rapid linear motion on several occasions, there have been other times when he has seemed unsure or even evasive. At the same time, despite having many opportunities for (and a track record of) criticising Perkins and "the Americans" for what he perceived as over-excitability or even flagrant invention, on this issue he has never once done so.

To some degree this difference of emphasis may be simply a question of timing due to the matter of radar horizons mentioned above combined with the rather well-defined remit of the GCI mission. But in any case, for the purposes of the present argument Wimbledon has been explicit enough: The behaviour of the Neatishead target during the time that it was observed was not consistent with a conventional aircraft.

"There is no doubt at all in my mind," he stated in 1978, "that something solid was in the air at Lakenheath that night and it wasn't a Russian, no aircraft at that time being able to perform as this object did and with the capability of crossing several hundred miles of water at nought feet under the Radar cover and return to base." The target disappeared from the forward AI coverage of the closing interceptor when range was down to "one mile or less" and within only one or two 6 rpm sweeps of the PPI positioned itself about 1/4 mile* behind the aircraft and maintained close pursuit through evasive manoeuvres for 5 minutes or longer at speeds in the region of 300 knots (about 350 mph). In a recent interview he reiterated what had occurred:

he was directed towards this object which was now stationary, near Lakenheath, and he called out ‘Contact’ which means his navigator had it on his own radar. He then called out ‘Judy’ which meant he had got it fair and square on his radar . . . and this thing, because it stopped, the Venom undershot it. At the time we didn’t know it had undershot it because prior to that this thing was moving slowly, and it stopped at the critical moment and the first Venom undershot. . . . And he said: ‘Lost Contact, More Help’. He was then told that the target was behind him, and whatever the first Venom did this thing, whatever it was, glued itself behind him.

Its acceleration and agility were both extreme and "no conventional aircraft could manoeuvre so rapidly" he said. Its performance was "30 - 40 years ahead" of the state of the art in 1956.

* NOTE: An azimuth resolution of 1.15 [r = 1.22 rad (l/D) where D = antenna width] of the Type 14 Mks.7 & 8 GCI radars would be on the order of >4000' at the ranges indicated (in the southerly sectors from Lakenheath according to Perkins), and a separation of 1/4 mile could not therefore have been resolved in azimuth. Equally the two targets could not be resolved in elevation by the Type 13 height-finders if vertical separation fell below about this figure, and the Type 7 elevation discrimination guide figure is comparable. This separation could not have been resolved even in range by the 3.8 microsec. pulse of the AMES Type 7 "master" radar. However 1/4 mile is about 1 times the range resolution of a 1.9 microsecond pulse and could probably have been discriminated on this axis by the centimetric radars. The figure of 1/4 mile should therefore be regarded as a minumum.

The Lakenheath CPS-5 antenna resolution at (say)10 miles would be about 1900' in azimuth, or better than twice as good as that of the best elements of the Neatishead system at 50 miles. If an echo separation of a couple of thousand feet was resolvable at close range on the CPS-5, but unresolvable or only intermittently resolvable at Neatishead, then the radar picture at Neatishead could well have been of a pair of targets which were rapidly 'swallowed' into a single larger echo after the initial 'circling'. The Interception Team's judgement, reported to the pilot at the time, that the target was now in close pursuit, would be known to Wimbledon who was monitoring the event from the Chief Controller's desk, and when this judgement was later corroborated by the report from Lakenheath he would of course assent to it. But he would not be in a position to confirm it from direct personal observation of (to quote Perkins) "two separate distinct targets" on the radar scope.

This is consistent with the evolution of Wimbledon's evidence. In his first statement he merely acceded to the Perkins/BOI-485 description of the extended tail-chase by implication (i.e., his report of what he observed went no further than the 'circling' manoeuvre of the target; he allowed the description of the ensuing pursuit to stand unchallenged but did not explicitly endorse it). In later statements he did explicitly endorse it.

Note that linear azimuth resolution of the Lakenheath radar would always be better than Neatishead's in this situation; nevertheless it would still be in the region of 3500' or more at the mean of the "10 to 30 miles" range reported by Perkins, with range resolution probably more than 1000'. So again the implication is that the mean real separation of the target and the aircraft was more than 1/4 mile if the two echoes were "always" resolved. (Interestingly, though, Perkins recalled a factory specification of 500' as the approximate minimum resolution of the CPS-5 at these ranges - 200' to 600' in practice - which is a factor 5 or 6 smaller than the figure derived from wavelength and antenna width above. This discrepancy is presently unexplained.)

When the interceptor broke off the interception and headed back home the following target stopped again at a point some 10 miles south of Lakenheath (now no longer being painted by Neatishead GCI) and remained stationary on-scope for an extended period whilst a/c #1 had radio contact with inbound a/c #2 and Perkins consulted the conference line authorities for instructions. He was told simply to keep watching and keep them informed. Only after this did the target begin to move again, he said, finally heading N at a constant 600 mph:

The UFO target did not leave the scope until 30 minutes or more after the 2nd interceptor returned to home station - we continued watching him and relaying this info to all concerned as long as he was on the scopes. He did not disappear to the north immediately, he stuck around for quite a while, but no one seemed inclined to initiate any further action to find out what it was.

It goes without saying that the authorities knew of no conventional aircraft which could explain this object. BOI-485 states that "Other aircraft in the area were properly identified by radar and flight logs as being friendly" and concludes that the phenomenon was "unexplainable".

Perkins explained that both US and British authorities on the conference line quickly dismissed suggestions that it could be a domestic product of either nation. The unusual intelligence arrangements apperently made disclose a need to acquire information, and part of that need clearly relates to "foreign technology" concerns; but it was evidently accepted that foreign technology, at least in the usual sense, was not involved and although the "object" remained in UK airspace "no further interception activities were undertaken" - at least none within the cognizance of the 3910th Air Base Group, Strategic Air Command or, it would appear, operational RAF personnel at Neatishead. (That is to say, not at this time. Further interception actions against what appears to have been a different radar target were undertaken by 23 Squadron Venoms beginning at 14:0200Z, in which neither Neatishead nor any other British ground radar site was involved.)

In summary the hypothesis of conventional fixed-wing/rotor craft appears extremely improbable.

b.) Anomalous Propagation

Some of the same features which might, prima facie, have been held to suggest an aircraft are seriously inconsistent with AP. Multiple concurrent detection of any target by several fixed and mobile radars on the ground and in the air, at different relative azimuths and incidences and at four different wavelengths between 3cm and 1.5m, would be exceedingly difficult to explain as an anomalous propagation echo. The behaviour of the target as reported seems on the face of it to place unsupportable demands on any type of anomalous propagation as currently understood, if the same propagation condition is being appealed to in order to explain an apparent causal relation between the targets on these various radars (but see Sections c. and d. below). This is especially so in view of the largely unremarkable middle-troposphere conditions suggested by the Hemsby radiosonde data. Nevertheless it is possible that different conditions operating on different ray paths might cause sporadic AP echoes which, by chance, appeared to correlate with one another.

Philip Klass [1974] proposed that "the erratic behaviour of the [Lakenheath] radar-UFOs is characteristic of spurious targets that have appeared on scopes since the earliest days of radar", suggesting that a random succession of such "commonplace" AP blips could easily be interpreted by an imaginative operator as the coherent track of a single object "which can fly at fantastic speeds, stop instantaneously . . . and perform other seemingly impossible feats." Such spurious echoes might, he believed, have been ground returns due to AP conditions and/or multiple-trip AP echoes from remote ships in the Channel or the North Sea.

At the start of the incident the target was reportedly being monitored by both RATCC and GCA units "in the same geographical location" SW of Lakenheath for 5 minutes and then tracked to a position NW of Lakenheath. The implication of BOI-485 is, as has been shown, that this target was initially picked up by the GCA CPN-4 (possibly at a point within the minimum range of the RATCC CPS-5) and tracked by GCA to the point 20 miles SW where, as Perkins reported, it was then independently noticed by RATCC controllers. The RATCC then for the first time telephoned GCA, who "confirmed the target was on their scope in the same geographical location." In 5 minutes the GCA antenna would scan the target some 75 times, the RATCC scope some 45 times, quite long enough for the operators to be very confident of the range and bearing on both scopes. These circumstances argue quite persuasively that the same reflector was returning signals to both radars for the following reasons..

There is meteorological evidence (from temperature alone; humidity data are not available for this altitude) for a possible high-level superrefractive or scattering layer at about 31,000' or above, which could conceivably cause multiple-trip AP echoes from ground targets far beyond the normal ranges of both radars. However, multiple-trip AP returns from the same remote ground reflector could almost certainly not be displayed "in the same geographical location" on two radars with different PRFs and unambiguous ranges differing by a factor of more than three. For example, an efficient reflector at a true range of 220 miles could be displayed as a second-trip echo on the CPS-5 at a spurious range of 20 miles (true range minus unambiguous range of the set); but could only appear on the 60-mile CPN-4 as a very much weaker third-trip echo at a different displayed range of 40 miles, and thus not "in the same geographical location" as reported by Perkins but fully twice as far from the site. Note that this concurrent geographical match on the two scopes was confirmed independently in contemporary intelligence report BOI-485. The alternative is two quite different ground reflectors, detected by two different multiple-trip ray paths on the two radars, which were at different real ranges but happened to be at the same azimuth and happened to display by chance at the same apparent range. In conditions of "little or no traffic or targets on the scopes" this notion is not seductive.

First-trip ground returns from a reflector at a true ground range of about 20 miles are more likely, which would overcome the problem of concurrent detection of the stationary target at the same range - despite the very different powers and frequencies of the two radars which would render them differently susceptible to superrafractive conditions. Indeed it is consistent with the implication of BOI-485 that the onset of superrefractive AP would occur sooner for the shorter wavelength (10 cm) CPN-4 than for the longer wavelength (23 cm) CPS-5.

But this same differential susceptibility to AP means that it is much less conceivable that changing AP conditions would cause ground echoes from a target at this location to disappear simultaneously on both radars in order for "the target" (that is, presumably other sporadic AP echoes) to then be "tracked" concurrently north past the station on both radars.

Perhaps, despite the telephone contact between the two locations, these motions were not truly concurrent. However there remains the problem that there is no clear evidence in the refractivity N-profile for the marked superrefraction in low levels of the atmosphere that would be required by this hypothesis - in fact, the evidence suggests that if anything the surface conditions were markedly subrefractive, which would prejudice rather than favour the detectability of ground targets. Moreover the short path lengths of signals reflected from local static sources of clutter would very likely not be sufficiently variable to introduce phase differences capable of defeating the MTI filter (multiple-trip AP echoes are more likely to do this; see below).

If a static ground reflector seems an unlikely candidate for these initial echoes, a mobile reflector (such as, for example, a moving empty lorry), which shifted to present a less efficient aspect, could both defeat the MTI and better explain the common loss of the echo on the two radars just as "the target" began to move. But then what of the subsequent echoes, displaced scan-to-scan in simulation of a 500 mph target moving off on a straight heading to the NNW of Lakenheath? It is surely becoming fanciful to suppose that (for example) a series of lorries strung out along 35 miles of the A10 Cambridge-to-Kings Lynn road would be successively picked out by fluctuating AP conditions over 100 of azimuth in 4 minutes to suggest such an illusion?

It is noteworthy that there was "little or no [aircraft] traffic or targets on the [CPS-5] scopes as I recall" according to Perkins, which is not suggestive of widespread intermittent AP ground echoes from which a spurious "track" might be imaginatively interpolated."We were using full MTI on our radar which entirely eliminated all ground returns and stationary targets" - except, of course, the apparently-stationary "UFO". The CPN-4 of the MPN-11A GCA system at Lakenheath was also fitted with MTI. As mentioned above, it is possible that returns from a fixed ground reflector in superrefractive conditions might sometimes defeat an MTI and be displayed.

An analogue MTI of this vintage is essentially a delay line consisting of a mercury column fitted with transducers which allows phase comparison of consecutive reflected pulses: if the phase of pulse 2 is identical to the phase of pulse 1 the circuitry decides that the target has not moved and no signal is fed to the display; but if the phase has shifted by some given fraction of a wavelength the signals are interpreted as returns from a moving target and displayed. This discrimination is therefore very sensitive to within fractions of a wavelength (centimetres) but relies on the assumption that the signal path length is constant from pulse to pulse and that any phase shift is due only to variations in the true slant range of the target. If the path length were inconstant, as might conceivably be the case due to abnormal refraction or scattering of the radar wave by a shifting atmospheric structure, it is possible that tiny phase differences might occur between a sufficient number of consecutive pulses to generate an output which will then be fed to the display as a "blip".

Although this mechanism is itself sporadic it is true that phase-shifted AP echoes which did defeat the filter over a very large number of scans might appear as unexpectedly strong and consistent targets. This is because the MTI delay line introduces a sort of 'digitisation' into an analogue radar by offering a gated and clipped amplified signal to the display. This has the effect of rejecting extremely weak signals and limiting strong signals, tending to flatten the normal wide variation in echo brightness on the PPI to a condition of "on" or "off". An operator unfamiliar with the use of MTI might be surprised by the strength of an echo in these circumstances and misled into rejecting the possibility of a spurious return. In this case the MTI unit on the CPS-5 was apparently a recent field retrofit, and it has been suggested that this is significant because operators could have been unfamiliar with it.

On the other hand, although this MTI unit was new there is no reason to suppose that the RATCC controllers were not familiar with MTI in other settings. USAF personnel were typically posted around from site to site periodically and would have experience on other radars where MTI was used. US radars commonly incorporated MTI at this date, unlike UK radars. Perkins, the RATCC Supervisor, had been a radar approach controller at Westover AFB using the MPN-11A GCA since 1952 before coming to Lakenheath. This was the same GCA set-up in use at Lakenheath employing CPN-4 search radar fitted with MTI. He stated that he had helped set up the RATCC at Lakenheath in 1953 using a remote feed from the GCA CPN-4. MTI on the CPN-4 was not a field retrofit but a factory feature. And of course the GCA controllers in August 1956 would have been familiar with its use on a daily basis. BOI-485 states that all the controllers were "experienced", and that "technical skills were used in attempts to determine just what the objects were. When the target would stop on the scope, the MTI was used. However, the target would still appear on the scope." In other words, the MTI was being turned on and off, implying that the controllers would have been well aware of presentation anomalies that might have been caused by the MTI. There is no prevalent or conceivable defect in early MTI technology (such as loss of phase coherence or standing waves in the mercury delay column) that would be likely to cause spurious echoes correlated with the antenna scan period so as to generate successive displacements with consistent range and azimuth rates, simulating the regular, localised motions of the target reported here.

The Watch Supervisor stated in 1975:

I have had the same experience as other radar operators with ghost echoes, birds, skip-range targets and all the problems with MTI. We are not completely uneducated on the subject as these phenomena have caused accidents. But to suggest that several radars can be simultaneously chasing a ghost or a vagary of MTI is ludicrous.

Furthermore, a defeat of the MTI filter by AP is only likely to occur in conditions where path lengths are abnormally long, so that very small variations in the curvature of the refracted ray have the opportunity to introduce detectable differences of path length and hence of phase. Abnormally long path lengths will occur with very distant reflectors beyond the unambiguous range of the set and the returns are then displayed as second- or third-trip echoes. But, as has already been shown, it is extremely improbable that such an echo would be displayed on two radars with different PRFs "in the same geographical location". Moreover, despite the fact that AP tends to favour shorter wavelengths the detection of AP echoes by third-trip returns on the CPN-4 would be very much less likely than detection by second-trip on the much more powerful CPS-5 due to the inverse 4th power attenuation of the signals from a reflector which is required here to behave as a point target. Yet the target was detected at least as early on the CPN-4, if not earlier. The subsequent rapid motion of the echoes over a large azimuth angle on both scopes remains unexplained.

The a priori probability that the very singular manoeuvres of the target when approached by the interceptor could be due to misinterpreted sporadic AP echoes must be vanishingly small, and that independent operators at remote ground sites should experience the same extraordinary illusion cannot be seriously entertained. Furthermore the wavelength of the Neatishead AMES Type 7 is ten times as long as that of the Lakenheath CPN-4, that of the CPS-5 being different from either. The simultaneous detection of a "clear" echo from the interceptor's AI radar beam on a fourth frequency at different and changing angles of incidence adds yet another order of unlikelihood. With three fundamentally different ray geometries and four different frequencies there is essentially no probability that AP echoes would show up concurrently, even in merely similar displayed positions, let alone in such a singular fashion as to suggest the same illusory "track" in the imaginations of different operators.

In summary, if the events witnessed were anything like those reported then the evidence strongly suggests that the echo behaviour reportedly displayed on these various radar systems could not be due to operator misinterpretations of sporadic AP echoes. It appears much more probable that the various radar indications were systematically related by a common physical cause.

c.) Specular Partial Reflection from Elevated Layers

In the simplest case partial reflection involves reflection of radar energy at the boundary of a layer of sharp refractive index gradient. The incidence angle is narrow - only a few degrees - and energy is reflected forward to ground targets further away. The energy returned via the same path may be detectable at the receiver, causing the ground targets to be displayed on the PPI. Slight rippling of the layer under the influence of winds can cause a region of efficient reflectivity to move across its surface, with the result that the reflected beam impinges on different ground targets or surfaces at different ranges, generating what might appear to be a moving echo or echoes. The efficiency of this process is roughly inversely proportional to the 6th power of the cosecant of the elevation angle, and thus proportional to increasing displayed range (see Analysis of the Bentwaters Slow Cluster, Part 2 (d).).

In the present case the reported target speeds alone are sufficient to rule out this hypothesis, since partial reflection echoes of this type are always displayed at roughly twice the true slant range to the point of reflection and therefore move at roughly twice the speed of the wind at the altitude of the layer. The displayed slant range of the primary target as described in BOI-485 moves from 6 miles to 20 miles, so that the highest geometrically possible layer altitude (assuming reflection to occur at normal incidence for the sake of argument) would be 3 miles. Winds aloft data for the Lakenheath area from midnight to 140600Z are given in BOI-485, and windspeed at 16,000' was 45 knots. Even at 50,000' (20/2 miles) windspeed was only 75 knots. The echo speeds resulting from these very generous maxima are still much too low to account for the order of speed reported.

In practice of course no energy is emitted at near-normal incidence due to the radiation pattern; the highest practical elevation angle for detection of any target on the CPN-4 is about 45 at low altitudes, leading to an implied maximum layer altitude of something like 10,000' for a target at a displayed slant range of 6 miles and a windspeed therefore of only about 35 knots. Moreover due to the inverse proportionality of reflectivity to elevation (and thus to incidence angle), and given typical power-reflectivity coefficients together with the low sensitivity of a surveillance set such as the CPN-4, a partial reflection echo is most unlikely to be detectable above a grazing angle of about 10, which would imply a layer at <3000' with displayed echo speeds of only about 40 knots - an order of magnitude too low.

Whilst such echoes move in response to the wind they do commonly move at a moderate angle to the wind. However the initial target headings of approximately SSW and NNE are both across the 230/290 winds, the former with a component of motion into the wind, and are essentially 180 apart. This behaviour is unintelligible in terms of any atmospheric structure, or combination of atmospheric structures, responding to the effect of winds.

The complex behaviour of the target during the interception is evidently independent of any conceivable motion due to winds and is in every respect utterly unlike that of a partial reflection echo. It is also noteworthy that the efficiency of a reflecting layer is a function of the frequency of the incident energy and would be very different at S-band, L-band and VHF leading to very different intensities of back-scattered energy at the receiver. Yet both at Neatishead and Lakenheath it was described as having a presentation similar to that of the interceptor (note that no Neatishead radars were equipped with MTI, so that the sort of flux-limiting effect described in Section b. above cannot be involved here).

If the Neatishead interception team had been using the full range of available radars they would have been able to "see" the target at S-band and VHF alternately at 12 rpm due to the reciprocal orientation of its Type 7 and Type 14 Mk. 8 surveillance antennas. A spurious target would be expected to betray itself by flashing on and off in these circumstances. The fact that the target remained "consistent" in strength on the Neatishead scopes might be explained because Wimbledon recalls that they were using "mainly" the Type 7 alone. Nevertheless, the Type 13 heightfinders, which obviously were in use in this interception control situation, are themselves the same centimetric wavelength as the Type 14 and very different from the metric Type 7. A good return on both PPI and RHI scopes is therefore itself not diagnostic of partial reflection (or AP in general). Moreover, there is no aerological evidence for a possible marked scattering layer below 31,000'. The lowest possible height indication from such a layer would display as 62,000'. No jet fighter could operate at anything like this altitude in 1956, hence GCI could never even have begun to bring the aircraft to interception under its control. (This is not to say that an undetected low-altitude layer is ruled out, however.)

Finally, Neatishead is some forty miles or so further from the area of the interception than is Lakenheath, and it is obvious that the radar pulses transmitted by and returned to each site must have a totally different propagation history. The displayed slant range of a partial reflection echo is in general twice the range to the point on the layer at which reflection occurs, so that in order to display echoes in "the same geographical location" fixed now by the position of the interceptor, each radar must be receiving reflections from quite different remote regions of the same layer or from quite different layers (and by backscatter from completely different parts of the landscape). There is no common geometry and no common propagation mechanism, so that the displayed proximity of the echoes to that of the Venom is reduced unsatisfactorily to a matter of pure happenstance. The additional target detected by AI radar aboard the jet presses coincidence too far, and to say that such an event is highly improbable is to understate the case. In summary, partial reflection echoes appear to be ruled out.

d.) Scatter from Clear Air Structures

We can distinguish forward scatter and direct back scatter.

Most of the arguments against partial specular reflection also apply to forward scattering from clear air structures. In such a case regions of turbulence, sometimes quite highly structured in bands or cells, may move across the layer boundary, again with the wind or at a moderate angle to the wind, and again echoes typically move at about twice the speed of the wind. Dielectric inhomogeneities associated with the regions of turbulence can scatter energy incoherently to the distant ground and return it by the same path. (See Analysis of the Bentwaters Slow Cluster, Part 3.) Separate radars which satisfy the scattering geometry could conceivably detect the same structure concurrently and from different bearings, but not from different bearings and in the same geographical location. Typically these will be extended amorphous echo regions or fluctuating patterns of smaller echoes rather than a discrete and isolated aircraft-like point target. And once again in this case the absence of a common propagation history, the motions of the target, the relationship to the interceptor, the order of speed reported, and the multiple directions of target motion, are all irreconcileable with scattering features moving under the influence of winds.

The efficiency of the scattering process is also sensitive to frequency, varying in inverse geometric progression: Greatest reflectivity occurs at S-band up to about 5 GHz, but at the low VHF frequencies of the Neatishead Type 7 master radar efficiency would be relatively very poor indeed, precluding echoes of comparable strength.

Direct backscatter from sharp clear air structures with extreme power-reflectivity coefficients is occasionally detectable even at near normal incidence by sensitive research radars operating at optimum frequencies. Rare effects may presumably occur which are beyond the currently understood limits of radio propagation in the atmosphere, such that a very discrete domain of clear air turbulence might conceivably return direct signals to remote instruments, even at widely differing and inefficient frequencies, thus being painted in the same triangulated true location. In the present instance, however, none of the radars involved was a sensitive research instrument and this hypothetical mechanism would remain incapable of explaining any of the reported target movements.

The classic investigation of clear-air angels by Borden and Vickers [1953] found that even the strongest echoes were not sharp and distinct in profile but had slow rise and decay times unlike aircraft targets, were strongly correlated with wind direction and velocity and with the presence of low-level temperature inversions, and that individual echoes persisted over average track lengths of only about 2 miles. Whilst it is impossible to make very meaningful comparisons with the target presentation observed in this case (notwithstanding qualitative statements that the Venom and the unknown appeared comparable on the scope), each of the other conditions identified by Borden and Vickers is rather reliably violated.

In general the reported continuity of complex motion by a single well-defined target over an extended period is quite unlike any echo behaviour likely to be caused by scattering from known or conceivable clear air structures (which, it might also be noted, are apt to be disrupted by the passage of an aircraft due to atmospheric mixing, and have certainly never been known to "attach" themselves to passing jets and be towed around the sky at >300 knots).

e.) Echoes from the Moon

Radar echoes from the moon were first detected by 'Project Diana', an experiment designed by the US Army Signal Corps in January 1946. At the Evans Signal Laboratory, Belmar, N.J., two 32-dipole SCR 271 early-warning antennas were mounted side-by-side on a 100-foot tower and directed towards the rising moon. Calculations indicated that a mere 3kw of peak power delivered in pulses more than 0.02 sec in length should return echoes about 20db above the noise level of the receiver. The experiment succeeded.

Later pulsed radar peak energies are very high compared to the 3 kW beamed at the moon by Project Diana, and although the pulse lengths are very short (order of 10-5 of Diana's 0.02 second) there will be several hundred per second. Therefore owing to this spread of pulses the total dwell-time within the beam is still sufficient for the integrated spots of excitation on the scope to represent an echo from the whole hemisphere.

In 1959, with the 'Oxcart' programme to supercede the U-2 getting underway, the CIA's Project Melody began plotting the placements and characteristics of Soviet air defence radars from the signals reflected by their missiles duriing flight tests. This led in the early 'sixties to the idea of using signals reflected from the moon, and building directly on research stemming from Project Diana the CIA succeeded in mapping in detail the entire Soviet 'Tall King' radar network from a giant radar antenna in New Jersey [Poteat 1998].

There are many special factors affecting the efficiency of radar echoes from the moon, such as a short-term fall off in energy due to libration and a more long-term decline due to Faraday rotation in the ionosphere. The latter is the rotation of the plane of polarisation of electromagnetic waves passing through a magnetic field, which reduces the detectability of the signal at the antenna. I frankly don't know how those variables apply in the present case, but there are other variables involved about which something can be said.

It appears that such echoes were occasionally detected on early surveillance radars of even modest performance (see here) and that electronic filters were sometimes installed to exclude them. An unusually dramatic incident of this type occurred in 1960 when the threat computer of the newly-operative BMEWS FPS-60 early warning radar at Thule in Greenland was swamped with inbound signals. Bomber crews at US air bases were brought to alert but were stood down when it was realised that the unfamiliar power of the radar was generating echoes from the moon which the computer had not been programmed to recognise. The problem was solved by incorporating an electronic 'moon gate'. This involved shifting the receiver frequency every 1 seconds, so that multiple-trip returns from the moon (2.4 light seconds out and back) would fall outside the receiver bandwidth and be rejected.

In the present case the first thing to say is that for visual observers at Lakenheath the moon had fully set at 235 azimuth (W by SW) by 2208Z. (This is allowing for normal atmospheric optical refraction of about 0.5 at the horizon, so that in fact the moon is at this time about its own diameter below the horizon.) But the 4/3-earth curvature of the radar line of sight is not limited by the visual horizon, and although the usual radar horizon only exceeds the geometric by about 15% this could be greatly increased in a superrefractive atmosphere. By 2240Z the moon was -5 below the visual horizon and declining at the rate of about 9/hour elevation, reaching -16 at midnight and -34 by 0200. The first Lakenheath UFO was detected some time after 2255. Could the moon have been detected even at these negative elevations?

The lower atmosphere shows no evidence of superrefractivity but there is the possibility of a superrefractive elevated duct at some 31,000'. If we suppose that this ducting layer is very large in horizontal extent, then given a fairly normal N-gradient through the troposphere a radar beam from an antenna 10 m above MSL would intersect this layer at a marginal grazing angle of 10 at a ground range of about 35 miles, and at 0 (i.e. tangentially to the layer) at a range of about 260 miles. The solid angle between these elevations would probably contain most of the radar energy, and 260 miles is not very far beyond the normal design range of the CPS-5. The likelihood of ducting or scattering increases the smaller the angle, and so it is possible that strong signals might be ducted or scattered to large negative elevations far over the horizon from a layer at about 31,000' over the southwest Channel approaches.*

* NOTE: It would of course be pure speculation to infer, from the Hemsby temperature-profile on the coast NE of Lakenheath, the existence of a layer tens or hundreds of miles SW. The night-time sounding for the most appropriate coastal station, Camborne in Cornwall - still not necessarily representative - shows no dewpoint data above the 400 mb pressure level, as at Hemsby, so a reliable N-profile for the appropriate altitude cannot be drawn here either and no detailed reduction of the figures has been attempted. But a cursory check of the temperatures shows that an inversion of at least 5 begins somewhere between about 30,000'-36,000', which is quite similar to the height of the first inversion at Hemsby and consistent with a widespread sub-tropopausal layer over southern England.

So, from 31,000' the refracted or scattered ray path would then reach a secondary 4/3-earth radar horizon some 500 miles further away off the Bay of Biscay, so that this position represents 0 elevation for the radar beam which is now effectively 'looking' from the altitude of the layer 260 miles southwest of Lakenheath. The time of moonset from the surface of the sea at this point will therefore represent, nearly enough, the time at which it ceases to be possible for detectable energy to be reflected back over the same ray path from the moon to Lakenheath.

On this basis we can say that the moon would cease to be detectable by radar at Lakenheath, even by this mechanism, at about 2300Z. By midnight the moon would be 8 below the minimum threshold deduced above (i.e., 16 below the optical horizon at Lakenheath) and by 0200Z it would be 26 below the threshold for detection (34 below the optical horizon at Lakenheath).

There are more complex cases, of course, which would in principle allow an even more remote radar horizon. Trapping can cause multiple 'bounces' of the radar line of sight, ducting radar energy to extreme ranges before, in principle, freeing it moonward. There is no evidence at all of trapping gradients in the lower atmosphere. It is possible that a similar effect could arise within an elevated duct and there is suggestive evidence (temperature only; unfortunately the partial pressure of water vapour is the more important contributor to the refractivity gradient) of possible tropopausal ducts. However the inversions of >6/75mb at Hemsby and >5/34mb at Camborne are not sufficient to demonstrate the necessary trapping gradient. Even at the higher level shown for the Hemsby data where the temperature gradient suggests a more respectable -30 N-units per thousand feet, this is still only somewhat superrefractive. As it is, -30 N/kft. is not at all a trapping gradient, which requires more than -48 N/kft. Humidity could quite conceivably result in a trapping gradient here, but this is not necessarily so and certainly isn't demonstrated for any level.

It is possible for the beam to experience multiple partial internal reflections inside an elevated duct before finally escaping into space towards the moon and then returning by the same devious route. In this mechanism a trapping gradient is not necessary because a portion of the energy incident on the layer is always reflected, whatever the refractivity might be. But the point here is that a condition of total internal reflection, even for an extremely sudden refractive index discontinuity of 36 N-units (a change from standard to trapping conditions), is likely to be an incidence angle of order 0.1 (proportional to the sine reciprocal of the smaller index divided by the larger). It seems highly doubtful that in practice sufficient energy would survive the large attenuation losses due to any long series of partial reflections in order to return, by the same path, from the moon to the receiver.

The evidence is therefore probably against the moon being a detectable target on Lakenheath radars at the time of any of the events reported from there, since according to BOI-485 the first telephone alert was received there not before 2255Z and the first detection of the Track E target occurred (according to Perkins) between 10-20 minutes after that, by which time there was already a vanishing chance of detecting the moon even by scattering from an hypothetical elevated layer.

However, allowing that it is possible let us consider how such a moon echo might display on the PPI. Firstly the moon appears stationary but obviously is not. The azimuth rate is small, near setting about 0.2 or 12 minutes of arc per minute, which is for practical purposes stationary. But the range rate of the setting moon will be about 900 mph. If this were displayed truly it would show the echo streaking radially away from the centre of the scope at this speed, heading out to maximum range before reappearing back at the scope centre and repeating the process. This is clearly not an explanation of a target whose first motion was at 400-600 mph in approximately the opposite direction. But in fact it would not be displayed truly because by the time the 1200th-trip echoes from this rapidly receeding reflector have returned to the receiver they will no longer be simply correlated one-to-one with the signals that were sent out; they will be scrambled, and the displayed range rate of the blip will have a very complicated relationship to the real range rate of the moon

To explain this: If the moon echo is picked up at the antenna peak gain like the echo from any other target (sidelobe gain will be discussed in a moment) then it would always appear on its true azimuth but would not display on every scan. Unlike with a normal target, the radar pulses emitted when the beam swings past the moon on one scan will not get back to receiver while the antenna is still pointing moonwards, but 2.4 seconds later when the CPS-5 antenna is already more than a quarter of the way around its circular scan and when the CPN-4 is more than half way around. Evidently if the out-and-back signal time were exactly the same as the rotation period then the moon echo would be picked up every time and displayed consistently at its true azimuth, simply delayed by one rotation. Equally evidently where the rotation period of the antenna is 9 seconds (CPS-5) then the two periods are only going to be in phase rather infrequently. Where the scan period is 4 seconds (CPN-4) the in-phase condition will occur with a different degree of infrequency. But each time the antenna rotates past the moon, even though there may be no returning signal to detect, another spray of pulses goes out, 1 of beamwidth passing a reflector in 0.0375 sec., and as the observation goes on the number of signals en route between the antenna and the moon at any one time clearly increases. Thus the probability of any one of these signal transit periods being in phase with the rotation period at the crucial moment increases as time goes on, and after an indefinite time there will be an indefinitely high likelihood, at which point one will always find a delayed moon echo showing up at the true azimuth (though not the true range of course) every time the antenna rotates past it.

Clearly whether this condition of continuous echoing occurs depends on how long the radar has been 'looking' at the moon. This will depend on when the radar was switched on and the time of onset of the changing atmospheric conditions that have led to this unusual sensitivity arising (the assumption being that the operators do not see 'moon UFOs' flitting around their scopes on a regular basis). These conditions might be expected to come and go. When 'looking' at the moon at low elevation the radar beam is subject to unpredictable attenuation and scattering during a long path through the atmosphere, to refractions and partial trappings by intermittent atmospheric layers and then scattering from grazing incidence at the ionosphere. It is remarkable that any energy gets to the moon and back into the atmosphere at all, let alone back to the receiver. Adding to these circumstances the complicated phase relationship of emission and detection due to the long delay of the trip, I don't intend to make any guess at all as to when or for how long a hypothetical moon echo on August 13 1956 might have appeared weak or strong, occasional or consistent.

In the above scenario, whatever the presentation of the echo and however often it appears, it never departs from the true azimuth, which means essentially that the echo can shift about up and down the same radius on the scope. This sort of mobility doesn't answer any of the descriptions of target motion in the reports. If the returned echo were unusually strong and picked up at the much lower gain of one of the antenna sidelobes might it be possible for more motions to be displayed on different azimuths.

Consider a pulse emitted by the CPS-5. By the time the echo has returned from the moon the antenna and the scope trace would be 96 in advance of the azimuth of the moon, i.e. if the antenna revolution is clockwise then it is now pointing at about 330 degrees in the NW (or if it is anticlockwise it will be pointing at 138 in the SE). The antenna does have some gain through 360, even though tiny compared to the main beam, and it would not be unusual to find significant spillover beginning at around 96 behind an old parabolic reflector. So if we imagine that the echo is weakly detected in this spillover zone then it would be displayed on the scope at 96 azimuth. As the beam rotates further varying outputs from minor lobes sweep past the moon but none have power enough to return a detectable signal, until the the main beam comes around again and another spray of pulses heads out, to return once more and be picked up in the spillover lobe, displaying again on a 96 azimuth. Thus we can imagine a weak echo which is picked up consistently, scan after scan, but not painted on the true azimuth of the moon. The echo would vary unpredictably in range along this 96 radius, moving on roughly the same restricted set of trace radii with small changes due to varying atmospheric path length.

This all sounds quite encouraging, but some problems are: a) it seems incredible that a moon echo could be detected in a minor sidelobe at maybe 1/1000th or 1/10,000th of the main beam gain without moon echoes being a very common occurrence on the radar in question, especially by the implied lossy ray path of trapping or elevated ducting; b) it is impossible for multiple-trip echoes from the moon to display simultaneously in the same geographical location, or at the same range or the same azimuth, on CPS-5 and CPN-4 radars with different unambiguous ranges and different scan rates. The observation of a stationary target "in the same geographical location", by these two radar sites in telephone contact, cannot be explained. Neither can the non-radial 400-600 mph movement on both radars to a position NW of Lakenheath.

All of the above should be considered in the context of the conclusion above that the moon had already become an extremely marginal and worsening potential target, even assuming an hypothetical high-altitude superrefraction or reflection mechanism for which there is only ambiguous evidence, by the earliest time at which the Lakenheath radar observations could have begun - i.e., shortly after 2300Z.

f.) Balloons/other Windborne Objects

Considered in isolation, certain phases of the reported target movement could be interpreted with some strain as echoes from a balloon. For example the initial detection of the stationary target some 20 miles SW of Lakenheath could be explained as a balloon. The reported location would be quite close to Cambridge airport, and the CPS-5/CPN-4 radar horizon at this range would be about 2000'.

A radiosonde balloon fitted with a foil or mesh tracking reflector and climbing at a typical rate of >1000 fpm would, during five minutes of detectability, have risen to about 7000' altitude subject to average winds of about 20 knots. Ground distance covered would be about 2 miles, initially NE then tending E at around the 6000' level, so that the slant range from Lakenheath would at first be changing significantly in terms of the fractional-wavelength sensitivity of the MTI, allowing it to be displayed as a moving target, but the discernable motion on the PPI would be very small and might for practical purposes appear to be nil. A subsequent drift to the E, then S of E under the influence of 290 winds above 6000', could conceivably take the balloon onto a trajectory at right angles to the line of sight from the Lakenheath radar antennas resulting in constant slant range for a short period. If range happened to be constant enough the MTI could now briefly be deceived into rejecting an apparently stationary target.

The balloon may subsequently have developed a leak and dropped back below the radar horizon before its echo could be reacquired, or, distracted by events elsewhere on the scope, operators may simply have not noticed its subsequent reappearance. This problem, known as tangential fade, can occur with simple MTI units of this vintage, although it is unusual and might be less likely with an erratically drifting balloon than with an aircraft. Concurrent detection of the target by two radars, each fitted with MTI, could thus be explained. However the subsequent target motion to the NNE at high speed, for which no satisfactory interpretation exists, remains as a further and uncomfortable coincidence.

Surprisingly, perhaps, the tail-chase is in some ways the easiest episode to explain in terms of a balloon. A long series of AI radar contacts made in the same area by two Venoms after 0200, designated here as Track F, appeared to be stationary at a few thousand feet and was interpreted as a possible static balloon. However likely or unlikely this may be it raises the idea of a tethered balloon. Let us suppose another, similar, unlighted balloon at the time of Track E, carrying a radar reflector and tethered by a light-weight cable several thousand feet in length over the airfield. The interceptor would attempt to come in "below and behind" its target and in attempting to turn behind it unsuccessfully its wing could slice through the cable at 300 knots, the loose end of the tether becoming snagged in some part of the airframe. The balloon and reflector might then be towed around the sky behind the jet at up to several thousand feet distance until the tangled cable broke free and it appeared to 'give up the chase'.

Problems with this hypothesis are numerous, not least the likelihood that USAF and RAF authorities, who had consulted at length over what to do about this target, would have known about the danger posed by a tethered 'barrage balloon' over the airfield. If they had even suspected this possibility it is inconceivable that they would have sent a jet into a potentially fatal collision with it. Note that the RAF interception was coordinated by the 60th AAA at Lakenheath whose Air Defence Command Post had direct tactical responsibility for airfield defence.

It is just possible that tethered balloons were being flown at Lakenheath for covert purposes unknown to the local authorities, perhaps connected with a prototype "Palladium"-style radar spoofing experiment by the CIA (see Section h. below); but still, the behaviour of the target before the arrival of the interceptor and after the departure of the interceptor would require the balloon(s) echo(es) to be somehow turned on and off in coordination with the insertion and removal of an electronic phantom in a way that is simply incredible. It is in any case no less inconceivable that the CIA would take such a risk with RAF (or USAF) planes and aircrews in these circumstances. And how likely is it that a fast jet could become entangled with a balloon tether in this way without even noticing - indeed, without fatal results?

A related possibility that avoids some of the objections above is a stray experimental neutral-buoyancy or tethered balloon rig from elsewhere. Balloons of many kinds were flown at the time for different purposes, and some were tethered, for example those used by the RAF for parachute training. Such a balloon might have broken away and drifted with the winds, trailing its tether to a point over Lakenheath where it perhaps became temporarily snagged.

Intriguingly Paul Fuller has uncovered information in PRO files about a programme codenamed Balthum which appears to have been a series of tethered balloon experiments operated out of the Met Office Research and Development Establishment at Cardington, Bedfordshire, around this time in 1956. One stated user of the data - Marconi's Research Station at Slough, Bucks. - and a description of their interest - "computation of modified refractive indices for centrimetric radio propagation" - reveals the meaning of Balthum: presumably this is BALLOON TEMPERATURE HUMIDITY, so these were basically just tethered weather balloons for measuring temperature/humidity gradients, possibly with onboard instrumentation which could be recovered and which was connected by wire to recording equipment on the ground. The Balthum balloons were flown on 5000' cables.

It is very interesting that Cardington is about 40 miles SW of Lakenheath and thus, given the general low-level airflow over the UK at the time, only a couple of hours directly upwind of the location of the UFO incident. File PRO MJ1/436, RAF Cardington, indicates that an average number of Balthum flights in excess of some 70 per month were made during this period, and states that some balloons were lost, on dates unknown. Cardington was also the home of the RAF Balloon Unit, but Paul Fuller found in PRO IR 29/2986 that although escapes of tethered parachute-training balloons did occur there were apparently none at this time since there were no launches anywhere between July 19th and August 16th. This may have been partly due to weather, since flights were banned if winds exceeded 18 knots.

The Operations Record Book for RAF Horsham St Faith (Norfolk) has an interesting reference to two successful searches in September 1956 by helicopters of 'D' Flight 275 Squadron for lost "Atomic Research" balloons utilising their SARAH (Search and Rescue and Homing) devices. The exact nature of these balloon flights is unclear, but may have been connected with airborne particle collection for nuclear test detection or monitoring.

It is conceivable that there may have been other experimental balloon launches of various kinds which would have been unknown to operational RAF and USAF authorities at the time, some of which may remain secret even now - perhaps experimental aerostatic or tethered platforms used for directed over-the-horizon radio relay, or for RADINT/SIGINT intercepts. It is possible to imagine that a high-altitude free powered aerostat, though untethered, might carry a long trailing aerial.

However the idea that the crew would not even notice that their fighter had survived a 300-knot impact, even with a cable, never mind the inertia of a balloon or aerostat with its possibly substantial instrument/aerial payload, remains difficult to entertain. And even if one entertains it, there is still no explanation at all of the prior and subsequent high speed movement of what was believed to be the same radar target. Bear in mind that what BOI-485 describes as the "extraordinary maneuvers" and "rapid acceleration and abrupt stops" of the UFO were observed on ground radars for an extended period long before the intercepting jet was even called for, which reduces the appearance of a very unusual balloon in the midst of this UFO event to the status of pure coincidence, and combining this unlikely juxtaposition with an almost-incredible snagging of the balloon by the jet surely stretches credibility too far.

g.) Meteors, Precipitation cells, Lightning channels/sferics, Auroral Ionisation, Birds, Insects

There is no resemblance whatsoever, either in displayed speed, duration, presentation or motion, to any of the above, considered singly, multiply or in combination with any other electronic, propagation or reflection effect.

Weather reports from UK Met. Office and USAF weather observers show that there were no thunderstorms anywhere in the region, and no hail or other precipitation.

Auroral ionisation is considered a very marginal target at L-band [Plank 1956] and would be undetectable at S-band or X-band, radar cross-section for plasmas varying in proportion as the square of the wavelength [Skolnik 1962]. The two ground and one airborne radars at Lakenheath could not have been been detecting auroral streamers. The metric Type 7 at Neatishead might detect auroral ionisation. But the southeasterly bearing of the target area from Neatishead, and the generally southerly bearings from Lakenheath, are in the diametrically opposite scope sectors from any possible auroral echoes. Moreover records of the Balfour Stewart Auroral Laboratory at the University of Edinburgh show that there was no significant auroral activity reported on the night of 13-14 August 1956.

To expand this a little: For mid-latitude N hemisphere observers the origin of the visual auroral display will always be in the N polar quadrant. Occasional visible streamers can extend into the S sky; but not only will the highest electron-densities be most frequent and persistent in the N core of the display, radars will not generally detect the zenithal streamers in any case (even if capable of observing the zenith, which a surveillance radar is not) because the governing factor is that the intersection of the radar line-of-sight with the magnetic field lines needs to be close to orthogonal [Leadabrand 1965].

The region in which this off-perpendicular angle is small defines a curved surface, which varies in altitude and extent as a function of latitude, but which for a radar N of the equator is always N of the radar. Even a radar which is at a latitude N of the auroral zone (i.e., N of about 67 degrees geomagnetic latitude, where the visual aurora appears mainly in the S - such as at Pt. Barrow, Alaska where the classic investigation of this effect was done in 1954 [Dyce 1955]) radar echoes still occur almost exclusively in the N.  So, occasional echoes from streamers extending past the zenith to low elevations in the S scope sectors might occasionally occur, but any radar in East Anglia will preferentially, if not exclusively, detect auroral echoes in the N quadrant. Hence one can be confident that if insensitive air traffic control radars operating at far-from-optimum frequencies could detect auroral echoes at all, they would certainly not do so preferentially in the S quadrant as was reported for the Neatishead and Lakenheath radars.

The same preference of the 'UFO' targets for southerly scope sectors is also relevant to the possibility of echoes from Perseid meteors, whose origin appears in the NE sky. If these were detectable they would be point echoes visible only for a single scan and found preferentially on WNW and ESE bearings on the Neatishead and Lakenheath scopes where the trails would be favourably orientated in the so-called radiant condition [Planck 1956] normal to the antenna line of sight. Most of the 'action' observed from Lakenheath was in the S sectors, the target departing due N, and all of the action observed from Neatishead was in the S sectors. These are exactly the least likely sectors in which Perseids might be detected, even if the highly-patterned echo behaviour was at all suggestive of sporadic meteor returns (see Analysis of the Bentwaters Fast Tracks for a detailed discussion of meteor echoes).

Echoes of birds at close ranges can present on scope with comparable strength to echoes of aircraft at moderate ranges, and because the inverse 4th power range/energy relation is somewhat counterintuitive an inexperienced operator might mistake a nearby bird for a plane or a UFO. But measured speeds of several hundred miles per hour rule out birds. It is always possible to suppose outrageous combinations of consecutive unrelated bird echoes that simulate a high-speed target. But to introduce such a theory marks the point at which one gives up reasonable inference, and can only be justified by an a priori confidence that "a UFO" would be impossible. (A discussion of some parameters of bird and insect 'angels' is given here)

h.) System Noise, Component Failure, Remote Radio Frequency Interference & Deception Jamming (spoofing)

Klass [1974] proposed that the Lakenheath CPS-5 target might have been due to a faulty MTI unit. This was suggested to him by the apparent stationarity of the echo despite the use of MTI, and by the inference that the analogue circuitry of an early MTI system only recently retrofitted to the CPS-5 during a field modification may have been prone to teething faults, possibly exacerbated by summer humidity. (See Section b. above for more on MTI effects.)

Although Klass does not address the reported concurrent tracking of the primary target by GCA radar at all, the CPN-4 also had MTI, in this case factory fitted. However such faults could not plausibly account for concurrent tracks on these two radars as reported by Perkins and confirmed by BOI-485. It is true that MTI was still a relatively new technology in radars of this vintage, but those very same technical problems which might conceivably have led to problems in these radars (poor inherent stability, imperfect analogue delay lines and phase sensitive detectors) had led to MTI not being fitted on any of the RAF radars at Neatishead. (The first UK radars to operationally use MTI were not introduced until after this date - the Marconi S.232/264 series of 600 MHz (50 cm) ATC radars [Bruce T. Neale, consultant, Marconi Radar Systems Ltd., personal communication, December 4 1986].)

Internal faults unique to a given piece of equipment are in general not appropriate to a target observed concurrently on several remote, electronically independent equipments. It is conceivable that false targets might be displayed due to coincidental independent component degradation or noise on two instruments, approximately at the same time and in a similar indicated position for a short period. However if the observed behaviour is at all complex, and at all persistent, the probability must intuitively be extremely low indeed.

Remote RFI is a different case and might similarly affect receivers with similar bandwidths, even, occasionally, receivers with different bandwidths if the source is powerful enough. However the display products would differ very markedly on radars with different scan rates, different PRFs, different range scales and in this case different display timebases. No single or multiple source of RFI pulses could possibly generate display products on each of the several remote radars which not only convincingly resembled a normal aircraft target but also correlated in displayed range and azimuth to within a few thousand feet through a complex series of movements.

This leads to the possibility of deliberate spoofing. In 1962, about the time when Bud Wheelon became the CIA's new head of the Directorate of Science and Technology, a radar spoofing programme was developed to be operated in tandem with the agency's bi-static Soviet radar mapping programme. This mapping programme had begun in 1959 with Project Melody and progressed to using radar echoes bounced from the moon (see Section e. above). Having charted the Soviet radar fence in detail and found it unexpectedly forbidding, the CIA, and through them particularly Strategic Air Command, now needed to know how to make spyplanes or bombers with radar cross-sections small enough to squeeze through. This was the beginning of 'stealth', and the CIA's radar spoofing was designed to provoke reactions from Soviet radars so that NSA COMINT intercept specialists could then decrypt their communications and estimate the minimum detectable cross-sections in various conditions. Wheelon dubbed this effort Project Palladium [Poteat, 1998].

Palladium engineers were probably among the few Americans who were delighted when the Soviets installed defences along the Cuban coast in 1962 as it offered a chance for some high-calibre ELINT measurements of their SA-2 missile radar. In this instance the Palladium gear was on the deck of a destroyer stood off Cuba, its antenna just peeking over the horizon, and a simulated jet fighter blip was 'flown' through the Soviet surveillance radar cover towards the island. At the same time a submarine surfaced offshore just long enough to release a callibrated series of balloon-borne metal spheres of different sizes. As expected the phantom overflight caused the Soviets to light up their SA-2 targeting radars, illuminating the test reflectors, and decrypted communications intercepts then revealed the performance of the radars.

(An interesting report that may conceivably be related to the Palladium programme is an unconfirmed report from 1967 that communications intercept operators of the Air Force Security Service's 6947th Security Squadron at Key West Naval Air Station overheard an attempt by two Cuban MiG 21s to intercept a UFO detected on radar entering Cuban airspace at 33,000'. They allegedly saw a bright "metallic sphere" and were instructed to engage but when the flight leader locked his weapons on the target his jet disintegrated in mid air. The UFO climbed out of radar coverage at 98,000'. An AFSS Intelligence Spot Report to NSA on the incident went unacknowledged; a follow-up report resulted in an order to ship all electronic and other records to NSA. In 1978 researcher Robert Todd requested information from NSA and CIA under FOIA legislation and found himself the subject of investigation by the FBI. An agent reportedly told him that the Bureau was acting on behalf of NSA and that "Some of the information is classified. Most of it is bullshit." [Fawcett & Greenwood 1984])

Similar covert tests were reportedly arranged all over the world. The Palladium gear was portable in a van and went with a CIA operating team, an NSA COMINT decryption team and a military support team. Given the value of Soviet radar defences to Strategic Air Command offensive planners, and the involvement of both CIA and NSA, one might suppose that if an early prototype test of a Palladium-type operation had ever been contemplated in Europe in 1956 then forward-basing of the equipment at a SAC airfield where CIA and NSA had already established a secure presence would be very natural. RAF Lakenheath was a SAC bomber base where in May 1956 the CIA had chosen to deploy one of the first U-2 spyplanes. Is it possible that the August UFOs in that area could have been a CIA spoof?

The first thing to say is that in this case a deception-jamming signal is faced with (at least) two local receivers with very different bandwidths, S-band (CPN-4) and L-band (CPS-5). Note that the reason two surveillance radars can share the same airspace like this is exactly because they are 'blind' to each other's frequencies; if they were not they would be inoperable due to interference. So the spoofer needs an extremely powerful out-of-band signal, tuned between the two, strong enough to generate normal-looking echo strengths on both scopes.

Secondly, if the signal were strong enough to 'leak' into both then the display products would differ very markedly on the two radars. The different scan rates of 9 and 15 rpm mean that a ghost 'echo' timed to appear as a stationary target on one scope would appear as a target skittering back and forth between opposite sectors on the other. The very different p.r.f. means that a solid-looking target arc on one scope would appear as just a radial scatter of dots on the other. There is just no way that these two radars could be spoofed simultaneously with the same transmitter.

So we start with two Palladium sets and two crews, then. They could handle things maybe - until the interceptor shows up. Then what about the AI radar? That would need a third transmitter - and a highly mobile one too, with some very careful radio coordination with the other teams. Then there's Neatishead's GCI installation, 40 miles away, where we have not only a combination of surveillance radar outputs at different frequencies with a third different rotation rate, but also Type 13 height-finder radars with a very different pattern of antenna gain and a vertical scan - not to mention Type 7 height-finding by a '3-D' system that calculates the height by comparing the signals reflected from the target by direct and ground-reflected ray paths. Have the spooks thought about that one?

Generally speaking, if you want to fool the radar you have to talk when its antenna is listening (a surveillance radar's receiver circuitry is obviously shielded to keep out stray interference and it 'listens' only through its antenna) and its antenna has relatively very low gain except for the bore-site azimuth and its major sidelobes, so that a narrow 'window' scans past your position every few seconds. This severely limits your ability to feed phony signals to the radar from a fixed location because whatever delay and doctoring you choose to do to the signal, generally speaking the radar only paints a target on the scope in the direction it is looking when it 'hears' you, which of course is always the same direction. In other words, even if you are able to use the sidelobe gain to get very powerful signals into the receiver when the antenna is pointing a little bit to the one side or the other of your position, on the whole you're limited to sending your ghost echo up and down this narrow azimuth sector.

If you site your equipment very close indeed to the antenna, or use an extremely powerful output modulated to compensate for the fluctuations in gain as the antenna rotates, you might be able to 'shout' into the antenna even when it isn't listening to you. Alternatively you might be able to plug your output directly into the radar electronics in a 'friendly' simulation, which would give you complete freedom. Only in these ways would it be possible to generate convincing phony echoes that moved between widely divergent azimuths - as this 'UFO' apparently did . But in either case, the need to coordinate a number of separate spoofing rigs and separate operators at these different radar sites 40 miles apart becomes still more obvious.

(Poteat says that Palladium could insert a ghost and 'fly it along any path [emphasis added]' but I don't believe that this should be misunderstood to mean 'anywhere in any sector of a surveillance radar scanning through 360 degrees'. Palladium was simply a receiver, a variable delay line and a transmitter, and was dependent on the antenna link of the remote 'host' radar to work. On a surveillance radar, a system like this could only generate a phantom on the main-beam azimuth of the antenna at the time the radar 'hears' the delayed echo returned to it by Palladium, so for a given Palladium transmitter output the scope sector available for spoofing would always be limited by the extreme anisotropy of the input antenna gain.)

Given that it would be too difficult to spoof all radars both in the air and many miles apart in this case, what if there was no attempt at a properly coordinated spoof on all radars? Suppose instead that they opted to generate a 'phantom' UFO only on local radars at Lakenheath and to augment the deception with a drone of some kind - maybe a tethered or neutral-buoyancy balloon (see Section f. above). The first requirement of this theory would be two sets of Palladium-type gear sited more or less under the antennas of each of the Lakenheath surveillance radars, or plugged straight into the electronics. Is there any evidence that such 'tampering' might have taken place?

In fact there is evidence that field modifications to both GCA and RATCC radars did take place just before the UFO incident. The RATCC was still a new facility, whose official dedication was only 'a few days old' at the time according to an August 1956 edition of the Lakenheath Base Newspaper. According to Perkins the RATCC was actually up and running that June, but an extensive field modification of the CPS-5 radar was made just before the UFO incident which lasted some days. During this time they reverted to the old set-up and the GCA approach control radar was used for Air Traffic Control, with a remote feed taken from the GCA trailer to the tower building. Perkins' recollection is consistent with the fact that MTI is known to have been installed in a CPS-5 field mod at some stage (that it was fitted is confirmed in BOI-485 but it was not part of the CPS-5 factory spec) and an adjustment of the scan-rate from 4 to 6.66 rpm was probably also done at this time. That something else was done presumably cannot be ruled out.

Assuming that these two radars could be rigged a date would have to be chosen on which to try out the deception. It might help to choose a clear night near the maximum of the Perseid meteor shower. Hence on the evening of August 13 the spoof team set up in their van out on the airfield or wherever they're coordinating things from, and power up their gear. To help start things off the radar crews would have to be alerted to be watching their scopes; hence a call is put through purporting to come from RAF Bentwaters suggesting that a UFO might be heading west after a dramatic radar-visual sighting there. This explains the existence of such a story in BOI-485 independently confirmed by Perkins, and also explains the fact that no report of such an event appears to have been made from Bentwaters. (The idea is to keep a lid on the affair at base level if possible; at this stage it is not anticipated that the events will get out of hand and leak to Blue Book.) Soon the spoofers have successfully inserted a similar phantom on both the Lakenheath screens and start to experiment by moving it around. This creates quite a bit of excitement on the military telephone lines and after a while they hear that an interception is planned, which is the signal to start phase two.

Because extending the electronic deception to a remote RAF radar site is both technically impractical and politically awkward, they've decided to use a bogus real target, a simple tethered balloon carrying a small reflector which the interceptor's AI radar will be able to detect at close range but which won't offer much of a target to the ground radar. When the jet arrives at the designated point it overshoots the balloon, but a brief AI contact is sufficient to raise the temperature all round. Now the spoofers start to have some fun and take their ghost for a spin around the sky, chasing the Venom's tail for several minutes, an event which makes an understandably vivid impression on everyone in the RATCC, although of course neither the interceptor crew nor their Neatishead controllers will have shared quite the same experience. Later in the night, when everyone has 'gone home', Lakenheath GCA notices a small stationary target over the airfield - the Palladium team's balloon failed to cut free as planned and is still tethered there. This explains the target to which the two Venoms are vectored in the early hours.

Well, apart from anything else this scenario fails to account for the Neatishead radar-UFO, tracked on PPI and heightfinder. If that was a different coincidental interception event at a different hour, or on a different date, why was it apparently not observed or reported at nearby Lakenheath, only at Neatishead? (The 23 Squadron scrambles at 0200 and 0240 were certainly observed at Lakenheath, but apparently only at Lakenheath - these were definitely not the same event as the Neatishead radar-UFO observed by Wimbledon.) Only by ignoring the evidence of GCI radar contact entirely can a deception-jamming scenario even be contemplated. Otherwise the idea of mounting a deception against such a complex of geographically remote radars of different types seems to me too mind-boggling to accept, even though I do like this theory. Of course by assuming that the reports of the radar events are grossly exaggerated tosh then one can re-engineer the evidence into a form consistent with the theory that a much humbler kind of deception was successful; but this is no kind of valid evidential procedure. Discounting the multiple radar corroboration would remove the strangeness that leads us to entertain a 'Palladium hypothesis' in the first place.

Finally, what was the likely status of any Palladium-style programme in 1956? Palladium was started under Bud Wheelon who was DDS&T from 1962. Palladium was certainly the result of a design devised in 1962 by CIA DST electronics engineer Eugene Poteus with assistance from Science Consultants of the CIA's Oxcart Program Office, Oxcart being the U-2 successor (eventually the SR-71). Electronic technology was by 1962 entering the age of miniaturisation, radar was changing fast and probably Palladium took advantage of much recent technology. There is no indication that Palladium was continuous with a long series of similar prototype developments going back to 1956, and the existence of such a forerunner seems both technically and historically doubtful.

i.) Ghost Echoes

The hypothesis of a ghost echo due to signals bounced from a secondary ground reflector via the Venom was espoused by Menzel [Menzel and Taves, 1977] to account for the "tail chase" observed in this case. This hypothesis was in fact advanced (and discarded) by Thayer [1969] in the Condon Report. Ghosts can theoretically behave in such a way as to appear to "intercept" an aircraft and can attain quite high angular speeds. Some superficial similarity to a ghost echo is therefore apparent and the hypothesis does seem intuitively quite attractive.

Such echoes always appear on the same azimuth as the primary reflector (the aircraft in this case) and always at a displayed range proportional to the total trip time via the secondary ground reflector. In other words the ghost can never appear closer than the aircraft, and if the aircraft turns away from the antenna the ghost, instead of seeming to chase the aircraft, will move ahead of the aircraft.

This appears inconsistent with a reported 5-10 minutes of evasive flying by the Venom at speeds of not less than about 300 knots, during which time the target was observed to "'lock on' to fighter . . . and follow all maneuvers of the jet fighter aircraft" [BOI-485], appearing to be "glued right behind him, always the same distance" whilst the pilot "climbed, dived and circled" [Perkins] "unable to 'shake' the target off his tail" [BOI-485]. Furthermore the separation between the aircraft and its ghost must vary as the range from the aircraft to the secondary reflector, and given the area of the chase ("10 to 30 miles all in the southerly sectors from Lakenheath" [Perkins]), the turning radius of the Venom, and the several miles of ground track covered each minute at several hundred knots, it is inconceivable that a ghost from a fixed ground reflector could remain "always the same distance, very close".

For example, if the displayed separation is (say) 1 x the azimuth dimension of the resolution cell so that there are two "very close" but "always distinct" targets - that is, about 5000' - then the path length to the ground reflector is 5000' and the aircraft altitude must therefore be equal to or less than 5000'. But in only 60 seconds of straight flight at 300 knots the aircraft will have travelled nearly six miles, and the displayed separation of the ghost from the aircraft echo will have increased sixfold to about 30,000'. Only if the aircraft circles around the reflector at constant altitude and on a constant radius for 5 minutes (a curious kind of evasive flying!) can the echo separation remain constant for 5 minutes; but if it circles around the reflector then for some 75% of each orbit it will either be flanking or pursuing its ghost echo, only being itself "pursued" briefly around the inbound quadrant. Further, to remain as close as about 5000' (slant range) from the ground reflector whilst circling requires an incredibly tight radius of turn which, if within the scope of the Venom's performance, would have to be conducted virtually "on the deck" at an altitude much less than 5000'; yet the Neatishead Type 7 radar horizon at these ranges is itself about 5000' and the aircraft could not have been detected by this radar. The Type 14 low cover gap-filler (if in use, contrary to the Fighter Controller's recollection) could theoretically have detected an inbound bomber down to 1000' over the sea according to available guide figures, but a small fighter over land is a different prospect, and in any case the idea of the sort of evasive aerobatic flying described ("he climbed and dived . . ." etc) being conducted at such dangerously low levels is hard to credit.

In short, although it will in general be the case that the ground track of a ghost from a ground reflector maintains some broad relation to the track of the aircraft (as in the imaginary example shown below) there will, equally generally, be large and highly variable differences in range between the aircraft and the ghost. There is no way for the separation in range to be small and held constant unless the aircraft is hovering at low altitude over the ground reflector (i.e., it is a jump-jet or a helicopter), and there is no way for the aircraft to be pursued continuously unless it is continuously inbound towards the antenna. Furthermore there is no way that such a ghost echo could be displayed as becoming stationary, as reported, whilst the aircraft continued on its way outbound from the antenna at several hundred knots. It is obvious that the conditions for a ghost echo of the type invoked by Menzel are, variously, inconsistent and aeronautically impossible.

Diagram of theoretically possible ghost track due to an aircraft and a stationary ground reflector

A second type of ghost echo involves a moving secondary reflector such as another aircraft. In this case the motion and separation of the ghost is not restricted by the fixed position of the secondary reflector on the ground and can in principle behave in a more complex fashion - but still remaining always at greater range than the primary reflector. It should be noted, however, that the reflection geometry required for even a simple ghost echo is very sensitive to changes in aircraft aspect relative both to the antenna and to the varying efficiency of the secondary reflector at changing incidences. Ghosts are inherently quite rare and typically intermittent. It would be highly unusual for such a ghost to persist as a strong echo for a significant period, and still more unusual for efficient geometries to be maintained between the aircraft and two remote ground radar sites simultaneously for several minutes without any fading or loss of the ghost echo.

With a multiple air-to-air reflection these tolerances will tend to become yet tighter since the relative positions of the reflectors can change much more rapidly, in addition to which a typical aircraft configuration is a much less efficient reflector at differing incidences than the kind of static corner-reflector (metal building structure, fencing, empty lorry or similar cube-corner conductor) which produces a good ground-reflected ghost. Air-to-air reflections will therefore tend to be still more fugitive. Further, a secondarily-reflecting aircraft would still have to remain in constant close proximity to the Venom in order for its ghost to also be displayed in constant close proximity - in other words it would have to be doing essentially what the "ghost" was displayed as doing, which is absurd and renders the whole hypothesis redundant.

Finally, the target was obviously detected both at Lakenheath and at Neatishead before the Venom was even scrambled, since its presence was the cause of that scramble, and was on-scope at both locations during the period in which the Venom was being vectored towards its position, remaining at closer displayed range than the inbound Venom during this operation. This target, which was then corroborated by AI radar on the Venom, plainly cannot have been a ghost. There appears to be no question of transference of attention from some other, prior target (for which no interpretation exists) to a coincidental ghost of the Venom (for which there is no support). In summary the "ghost" hypothesis appears to be of no value.

j.) Jet Exhaust

Thayer [1971] observed that radar will sometimes track a jet exhaust plume. This very interesting hypothesis merits further study, but is dismissed by Thayer in the present case because the movement of the target exhibited clear independence of the Venom both before and after the period of interception. On this hypothesis there would be no clear interpretation of the target towards which the Venom was vectored, or of the AI radar contact then acquired by the Venom. Further the simultaneous loss of this target by the Venom and the several ground radars, followed immediately by ground-radar acquisition - for the first time - of the Venom's jet exhaust plume, is too uncomfortable a coincidence to contemplate. It seems highly unlikely that an exhaust plume would present a consistent target with a strength and presentation comparable to that of the Venom on radars with very different operating wavelengths (S-band to metric) for several minutes. And it seems even less likely that this plume could then detach itself from the aircraft and behave in the extraordinary fashion reported, or that yet another spurious target (for which no interpretation exists) would appear on-scope just as radar for some reason ceased to track the Venom's jet plume.

In conclusion there appears to be no single conventional hypothesis, or plausible concatenation of conventional hypotheses, electronic, propagational, meteorological, psychological or physical, which suggests an explanation of the multiple-redundancy of ground and air radar contacts reported in this instance. If the target behaviours reported by the several sources are approximately accurate, then the simplest and most natural interpretation would seem to be in terms of a radar-reflective aerial "object" capable of unconventional performance and quasi-rational behaviour.

Martin Shough 2003

Go to: An Analysis of the Intercepted Targets over Lakenheath - Part 2

Archive Index