Archive Index


An Analysis of the Bentwaters Fast Radar Tracks

An Opinion by Martin Shough

In addition to a cluster of slow targets (designated Track D and treated separately here), three rapid targets on the RAF Bentwaters GCA radar were reported over a period of 85 minutes on the evening of August13. These fast tracks are here designated as: Track A, 2130Z; Track B, 2200Z; and Track C, 2255Z. (Note: readers should also consult here for a supplementary detailed analysis of Track C)

The first two were uncorroborated visually and were apparently not considered remarkable enough to prompt serious action. It was the third track, concurrent with air and ground visual sightings, that apparently galvanized Bentwaters personnel to telephone an alert. We deal with visuals elsewhere; here the radar indications will be discussed in terms of the following hypotheses:

a) Anomalous Propagation
b) Electronic Artefacts and Radio Frequency Interference
c) Ghost Reflections
d) Missiles, Orbital Bodies
e) Multiple-trip Echoes
f) Meteors
g) Shockwaves

Air Information Report IR-1-56 from Captain Edward L. Holt, Air Targets Officer, 31st Fighter Bomber Wing, RAF Bentwaters, contains the following details of the first high-speed target, Track A, detected on the Bentwaters GCA surveillance scope:

The following information pertaining to an Unidentified Flying Object sighted electronically on the Bentwaters GCA at 2130Z, 13 August 1956 was submitted by A/2C John L. Vaccare Jr., GCA Operator, 1246th AACS Squadron, RAF Station Bentwaters, England.

Airman Vaccare indicated that he tracked one Unidentified Flying Object on the Bentwaters GCA screen for approximately 30 seconds at 2130Z, 13 August 1956. The size of the blip when picked up was that of a normal aircraft target. The blip diminished in size and intensity to the vanishing point before crossing the entire radar screen.

The unidentified flying object was picked up at an estimated 25 to 30 miles east south-east of Bentwaters and flew a constant course of 295 degrees to the vanishing point on the scope which was 15 to 25 miles west north-west of Bentwaters at an undetermined altitude. Airman Vaccare estimated the speed of this object to be in the vicinity of 4000 miles per hour. This speed was calculated by comparing the speed of the object on the GCA scope with speeds that the operator is familiar with on the electronic simulator. A/2C Vaccare added that some idea of the speed of the object could be computed from the fact that each time the GCA antenna completed a revolution the blip from this object moved 4 to 5 miles on the radar screen. The GCA antenna completes a revolution once every two seconds. The weather was reported as clear with unlimited visibility.

Vaccare's arithmetic is correct, whereas the preparing officer's summary introduces an error: the scan rate of the CPN-4 surveillance scope of the MPN-11A GCA system is 15 rpm (+/- 1 rpm) or one revolution every four seconds, not two, the scan-to-scan displacement yielding a bracketed speed of 3600-4500 mph, consistent with Vaccare's judgement of about 4000 mph derived from experience on the electronic simulator. Further, rationalising the "approximately 30 seconds" total transit time to a multiple of the scan rate (32 seconds) and plugging in a track length of about 40 miles (Vaccare's lower estimate) again comes out at 4500 mph. The rough consistency of these estimates with a target of such unprecedented speed can be considered fair and does not suggest careless observation.

About 5 minutes after the integrated echo of the slow target cluster (see here) had turned onto a north heading and gone off-scope another high-speed target, Track B, was reportedly tracked on a heading of 270 degrees at 2200:

This object was tracked on the radar screen for approximately 16 seconds. Course of the object being tracked was from about 30 miles east of Bentwaters to approximately 25 miles west of this station. Speed of this object was estimated to be in excess of 4000 miles per hour. All radar returns appeared normal on the scope for this object except for the last return which seemed slightly weaker than the rest. Sgt Whenry explained that [the] object suddenly disappeared off the radar screen by rapidly moving out of the GCA radiation pattern. Light conditions were night. Weather was clear with good visibility and light winds.

There is a discrepancy in the measurements here ascribed to Whenry: A track length of 55 miles in 16 seconds implies a speed >12,000 mph, which is unarguably "in excess of 4000 mph" but so far in excess as to negate the sense of the approximation. It may be relevant that a duration of "approximately 16 seconds" would permit only a few paints widely spaced more than 13 miles apart (possibly six paints if the target position at the fourth paint, at a reciprocal bearing from the 3rd after a half-rotation period of 2 seconds and at a ground range from the antenna of about 2.5 miles W, were within the beam; a maximum of five paints if the target elevation at the 4th paint were above the upper cosecant-squared beam angle and thus within the zenithal shadow). This would be a fleeting "track" which would offer little opportunity for accurate observation.

It is possible to suppose that "16 seconds" is a typographical error for "36 seconds", which would yield a speed of about 5500 mph, perhaps close enough to Whenry's estimated "in excess of 4000 mph" in the circumstances. Given the difficulty of accurate judgements with tracks of such unprecedented speed it would be understandable if Whenry elected to eschew any claim to precision and, when asked to estimate the speed of this one, was only prepared to say that it was faster than the last one - i.e., "in excess" of the 4000 mph earlier estimated by Vaccare. But this is merely speculation.

With the exception that Track C involves reported concurrent visual observations (and is considered in detail later) these three rapid targets pose similar problems of interpretation and can be addressed collectively.

a) anomalous propagation

Anomalous propagation (AP) effects as currently understood appear to be unattractive explanations in this case. Observed speeds on the order of k/mph do not correspond to any atmospheric structure moving under the effect of winds and the probability of sporadic ground returns imitating coherent radial tracks on three separate occasions, once coincident with multiple visual observation, appears small. It is also relevant that the nearest radiosonde profile shows no strong positive evidence of abnormal refractivity in the troposphere save for a narrow subrefractive surface layer which would be expected to reduce the range to the effective radar horizon. This is the opposite of the superrefractivity that typically results in AP echoes.

Nevertheless it is true that the region of East Anglia is commonly accepted to be more prone to AP conditions than many other parts of the UK, and the weather reported for 13/14 August 1956 - a clear, starry, summer's night after the passage of afternoon weather fronts - is not inconsistent with atmospheric stratification. However it would be fair to say that the general conditions would not lead one to expect lower-tropospheric layering of the most dramatic type often associated with torpid heatwaves, since maximum daytime temperatures in the area had been on the cool side at 19-20C and breezes remained fairly brisk. In the humid heatwave conditions of summer 1952 at Washington, Borden & Vickers [1953] found that scattering due to hypothesized narrow layers of sharp N-gradient was almost always correlated with the presence of temperature inversions at or adjacent to the inferred altitudes. In individual cases the time of appearance of the clear-air angels on the scope also correlated with the expansion of ground clutter due to the inversions. In the present case there is certainly no evidence of a low-level temperature inversion, nor indeed any strong superrefractive gradient suggestive of AP clutter. But this does not mean that Borden/Vickers-type effects can be ruled out.

A possible marginally superrefractive gradient (-28 N-units per k/ft.) over a pressure difference of 13 mbar (a few hundred feet) is indicated at about 5000'. ("Possible" because it depends on interpolating a missing temperature datum in the Hemsby profile.) This gradient is far from the sort of value (>-48 N-units per k/ft.) associated with conventional trapping effects, but would at least not be inconsistent with the presence of elevated layering. Sharply bounded layers of high N-gradient departing transiently from the trend of the radiosonde profile, and falling between the data-points, are presumably possible in principle, and moderate breezes suggest the likelihood of waves on such hypothetical layers. There may also be subtler effects associated with a humid/dry air mixing at the top of a surface subrefractive region, for example. Moreover there is the possibility of remote multiple-trip ground echoes due to a high-level tropopausal inversion to be considered.

Ground returns due to partial specular reflection or incoherent forward scattering from moving waves or turbulence on an undetected layer might generate moving point targets; however the reflection geometry leads to echoes with displayed speeds approximately twice that of the wind at the inversion surface and displayed altitudes approximately twice that of the layer surface, and these conditions are inconsistent with both the reported target behaviour and the winds aloft data. Winds recorded by the Bentwaters weather detachment in IR-1-56 are:

Level Direction Velocity
surface 230 5-10 knots
6000' 260 30 knots
10,000' 260 40 knots
16,000' 260 55 knots
20,000' 260 70 knots
30,000' 260 90 knots
50,000 260 40 knots

Bentwaters Weather Detachment winds-aloft data

(Detailed weather data including Met. Office radiosonde readings can be read here.) Maximum wind speed recorded at Bentwaters for any altitude was 90 knots at 30,000', leading to maximum possible echo speeds of about 200 mph. It will also be apparent that a partially reflecting layer at such an altitude is inconsistent with the reflection geometry, which will not permit a displayed slant range of less than twice the true altitude of the layer (i.e., if the elevation angle from antenna to reflector moves through 90 degrees; although this is an unreal situation and no returns would be detected in practice at such an elevation due to the much lower angle subtended by the top edge of the surveillance beam and the effectively zero reflectivity of the layer at normal incidence). Therefore a partial reflection echo from a layer at (or above) 30,000' could not in principle be displayed at a slant range of less than about 11-12 miles and would in practice be detectable only at far greater displayed range.

Spuriously small ranges could be displayed on the CPN-4 by multiple-trip echoes detected by forward scatter from a high-altitude layer which is intercepted at the grazing angle by the radar beam at slant ranges longer than 30 miles. But again wind speeds and directions at any level are very grossly inconsistent, and it would not be possible for tracks generated in this way to cross the scope continuously between opposite sectors. (Multiple-trip mechanisms are considered further below.)

It is characteristic of such echoes that signal attenuation is roughly inversely proportional to the 6th power of the cosecant of the elevation angle, meaning that there is an acutely non-linear drop in echo intensity with proximity to the antenna, approaching effectively to zero at elevations far below 90 degrees, and typically below the maximum elevation of the radiation pattern so that a symptom diagnostic of partial reflection is close-in disappearance of the echo at greater slant range than would normally be the case for a solid target entering the zenithal shadow. A close-in disappearance at 2 miles slant range is noted in the BOI-485 report of Track C, and this has been taken by some analysts to indicate an abnormal circumstance diagnostic of partial reflection inasmuch as commonplace beam-angle signal loss might not be thought noteworthy. (No interruption is reported in IR-1-56 for Tracks A and B.) However the reflection geometry here demands a true inversion altitude considerably less than about 5000' (2/2 miles), and winds at 5000' are known to have been less than 30 knots leading to echo speeds of less than 70 mph. Thus there is an inconsistency of two orders of magnitude between the values for windspeed and inversion-altitude which emerge from plugging the report data into a forward scattering model.

Equally noteworthy in the case of all three tracks is that the smallest angle made with the heading of the wind at any altitude is 115 degrees. The targets all have a component of motion into the wind, and the track with the smallest such component (A) is only 65 degrees away from a heading into the eye of the least unfavourable (surface) winds, which are at the same time the most unfavourable in terms of velocity.

There may be some doubt as to the accuracy of estimated target speeds, particularly in the case of Track B although the operator's estimate here can be shown to be roughly consistent with the value derived from track-length-over-time if a typographical error is assumed in one digit (such errors are by no means unknown in USAF reports). It is also possible that the reported target headings are approximate (in the case of Track A, for example, a radial heading of WNW equates to 292.5 degrees, not 295 degrees, and Tracks B and C may equally have been a few degrees away from the cardinal headings given). Nevertheless the qualitative differences between the order of speed and the approximate headings reported on three occasions and the values required for partial reflection echoes are so extreme as to be implausibly explained by operator error. In short there appears to be no possibility of reconciling the reports with simple partial reflection from hypothetical sharp N-gradient layers at any level.

Sporadic ground echoes due to superrefractive trapping can appear and disappear as propagation conditions change from scan to scan. Thus operators have sometimes been misled by the illusion of movement between consecutive returns from different ground reflectors into reporting unidentified "tracks". However due to the random nature of this process such illusory tracks are typified by very erratic apparent speeds and courses, which are indeed the abnormal features which tend to prompt a report of a "UFO" target. In the case of Tracks A, B and C there are two common features which indicate a high degree of order, and so argue strongly against this hypothesis.

In each case the tracks were on straight headings, and in each case they appeared to pass (approximately) over the radar site - that is, they crossed the scope centre on diametric headings, twice E-W and once ESE-WNW. There are thus systemmatic relationships a) between consecutive paints and b) between each separate track and the CPN-4 scope centre. These are not symptoms diagnostic of sporadic AP returns. Track A implies some 9 or 10 paints (depending on the asymmetry of target reacquisition on the reciprocal westerly bearing); Track B is more fleeting, but implies between a few and 10 or 11 paints (if the hypothesis of typographical error is correct); and Track C (in respect of which there is no operator-estimate of duration) is 60 miles long and at the mean estimated target speed of 3000 mph corresponds to 18 paints in 72 seconds. A sporadic ground return might display as a discrete spot target. The probability of some thirty to forty such ground returns generating this degree of space-order and time-order by chance (even allowing for operator-selection from an hypothetical random background of unreported AP echoes) must be small.

This conclusion is reinforced by the reports that (in the case of Tracks A and B) the targets appear to have been followed somewhat closely from scan to scan:

All radar returns appeared normal on the scope for this object [Track B] except for the last return which appeared slightly weaker than the rest.

and similarly:

The size of the blip when picked up [Track A] was that of a normal aircraft target. The blip diminished in size and intensity before crossing the entire screen.

In the case of Track C some 70 seconds is quite long enough to confirm the target's heading and presentation, and the internal consistency of the figures for Track A suggests a systematically related pattern of measurements.

Comparison of target presentation with that of a "normal aircraft target" is also not diagnostic of sporadic ground returns (although it is true that an MTI filter can tend to flatten variations between the raw signals from different sources before they are presented to the display). Ducted returns from different ground reflectors on opposite azimuths with different propagation histories would be expected to vary erratically, not systemmatically, in presentation. And in the case of Track C the report of simultaneous air and ground visual sightings of an object on the same heading as the radar target adds a further order of improbability. (Diminution of signal intensity towards the termini of both Tracks A and B does not uniquely indicate any particular cause. Track B "suddenly disappeared off the radar screen by rapidly moving out of the GCA radiation pattern" in the opinion of the operator; the description of Track A could be taken to imply a less non-linear attenuation which could for example occur due to either a] a target flying close to the bottom/top edge either of the beam or of a major lobe, or b] variation in target aspect.)

Once again, the nearest refractivity N-profile shows no evidence of more than marginal superrefraction for any level where an index can be calculated. Indeed the only feature for which there is reliable evidence is a marked subrefraction in a surface layer at Hemsby which, if characteristic of the radar site, would tend to lift the elevation of the bottom edge of the beam, thus shrinking rather than expanding the radar horizon, and so if anything reducing the likelihood of AP ground echoes. The existence of a high-level tropopausal layer might reasonably be suspected from the temperature gradient, however.

b.) Electronic Artefacts & Radio Frequency Interference

Klass [1974] argued that fast-moving tracks might be caused by faults, citing a Bendix field engineer who had "observed apparently fast-moving echoes that turned out to be equipment malfunctions." He also appeared to suggest that problems with an MTI delay line on the CPN-4 might generate spurious fast echoes. The Blue Book analysis referred to mutual RFI as a possible cause. In this case all categories of such spurious signals can be treated together in the first instance since they raise similar issues with respect to their likely display products.

Considered collectively the essentially radial motion of all three rapid targets (and the 2130 cluster of slow targets) is interesting. On a radial-timebase PPI a systemmatic relationship with the scope centre invites, at first sight, the suspicion of spurious signals generated somewhere in the amplifier or receiver circuitry, remote interference pulses, or even active jamming.

Accidental RFI can be caused by other similar radars, or sometimes a radar's own components, and even out-of-band signals such as a dissimilar radar, voice communications, navigational aids and other sources might be displayed if they are strong enough to defeat the rejection characteristics of the receiver. Typical noise/interference effects are very unlike the echoes reported, painting speckles or radial or spiral patterns across the tube face with each sweep. But it is possible that sporadic pulse-trains with a duration very small in relation to the scan rate could produce brief, isolated false targets.

Since a spot will only appear on the PPI at the position of the electron beam at the moment the pulse of interference occurs, a situation could arise whereby a source of interference with a pulse repetition frequency identical to that of the receiving set could generate an array of spots on adjacent trace radii simulating the characteristic short arc of a reflecting target. If this source of interference is also cyclic (say, another radar) and the frequency of the cycle is shorter (on the order of microseconds) than the scan rate of the receiver, then each arriving pulse train could generate a new "echo" displaced on each scan. If the input pulse length is comparable to that of the receiving radar, and the duration of each pulse-train is comparable to the dwell-time of a target within the approximately 2-degree beamwidth of the CPN-4 (about 0.02 sec.), then each "echo" could resemble that from an aerial point target and the false track thus created would describe a radius from the point of acquisition in to the centre of the scope.

In 1956 the Blue Book Officer, Capt. G. T. Gregory, suggested that "interference between two radar stations some distance apart" might be caused by inversion conditions, but the hypothesis was not pursued to a level of detail sufficient to test it. It is conceivable that discrete sampling of remote radar pulses might occur in this way, leading to a false track generated by the mechanism described above. Unfortunately, however, this hypothesis breaks down when applied to the diametric motion of all three Bentwaters targets.

The motion of such a track between opposite scope sectors requires an incredibly fortuitous relationship between two different and immediately consecutive input periodicities synchronised to the half-rotation time of the display. A pulse train simulating a radially approaching target and having therefore a period slightly less than the scope rotation period would, upon reaching scope centre on trace-group T, next display on adjacent trace-group T(1) at the scope periphery before moving once again towards the centre, repeating this cycle through trace-groups T(-2) . . . T(-3) . . . T(-4) and so on around the scope until it returned to T again.

Spurious pulses mimicking a radially approaching target on trace group T
would disappear at scope centre C, reappearing at scope periphery on adjacent trace groups T-1, T-2, et seq., unless delayed for half the rotation period. A delayed pulse train would even then progress inward along the opposite radius, not outwards.

If, having reached scope centre C on trace-group T, the signal were instead immediately to begin painting in the diametrically opposite sector, on trace TA, it would have to be delayed by the half-rotation period (approximately 2 seconds) and would even then progress inward towards scope-centre once again from TA to C as the mirror image of its previous movement on trace-group T.

In order to appear moving outwards from C to TA, maintaining a diametric heading, the signal phase relationship with the rotation period would also have to change so that the input period now became slightly longer than the scan rate. In addition, in order for the displayed speed of the target to be constant through the crossover between sectors the negative difference between the input period and the rotation period on trace-group T would have to equal the immediately consecutive positive difference on TA. The onset of this new period would be required to occur precisely at the starting point (scope-centre) of the opposite trace-group. That the microsecond synchrony required to collectively satisfy these several conditions could occur by chance, even given the appropriate input source(s), is negligible even once, and on three separate occasions is essentially nil.

These circumstances raise the question of active jamming or 'spoofing'. It cannot be ruled out that targets of the type described at Bentwaters, detected on a single radar with (except in one case, Track C) no visual corroboration, could be generated by spoofing. However it seems unlikely given that the antenna azimuths at the time of signal reception were divergent from track to track and in each case were reciprocal at the start and end of the track. Active jamming could simulate the necessary time-domain synchrony for spoof targets to cross the scope centre, as described above, but the problem is how the signal gets to the receiver from an external transmitter.

If the input is from a static transmitter through the antenna link then in all these cases the reciprocal antenna azimuths would imply a signal strong enough to utilise the extremely low back lobe gain of the antenna. This means a powerful or close-range static transmitter with the output signal strength controlled to compensate the 104 or greater variation from main-lobe to back-lobe antenna gain. Alternatively a mobile transmitter could be imagined (though not easily), perhaps a helicopter platform hopping from one side of the radar site to the other. But such a platform would be detected by radar and/or visually/aurally from the Tower. Or the spoof could be arranged by having equipment secretly hard-wired into the receiver circuitry of the CPN-4. None of these scenarios is impossible, but presumably they are somewhat unlikely and no such spoofing programme is known to have been developed in 1956. (A discussion of CIA's later 'Project Palladium' deception jamming exploits is given in the context of the first intercepted Lakenheath target, Track E, Part 1, sect. h.)

c.) Ghost Echoes

There are rare circumstances in which signals returned from windborne objects or weather could create targets travelling apparently against the winds. A very efficient ground reflector (i.e., a large chimney or building) at a diametrically opposite azimuth could return secondary reflections which would appear to close range from the opposite sector as the weather target closes range at windspeed. But in the present case the speeds are totally inappropriate, and such a secondary reflection could not appear to cross the centre of the scope into the opposite sector. The ghost reflection would furthermore have a scope presentation inferior to both the primary and secondary reflectors themselves, and even though the ground reflector might be masked by permanent clutter the airborne target (which, be it weather, a clear-air structure or a solid object, is required to be an efficent radar reflector inside the beam pattern) would itself present a strong echo.

Ghosts due to secondary ground reflections from aircraft can attain quite high angular speeds, displaying beyond the primary (airborne) reflector at ranges proportional to the added round-trip time to the secondary (ground) reflector and always on the same azimuth as the primary reflector. Thus the angular rate is preserved but at greater displayed range, leading to high displayed speed. But high-speed radial ghosts are a special case. The highest speeds require a fortuitous geometry with the aircraft flying directly towards the antenna with a secondary reflector lying on the same radius, and even then the ghost cannot exceed twice the speed of closure between the aircraft and the reflector (i.e., twice the aircraft's ground speed). This maximum represents the case where the secondary reflector is at or close to the location of the antenna, so that the the ghost's displayed range is always twice that of the aircraft. (Note: if the ground reflector is primary the ghost's movement will be radial on its azimuth but cannot exceed the true rate of closure of the aircraft.) In this case, if the secondary ground reflector is equally efficient through 180 degrees then, as the aircraft flies over the reflector and the antenna, its ghost will move ahead of the aircraft on the reciprocal heading, again at a maximum of twice the aircraft's ground speed. Thus a ghost on a diametric heading could conceivably be generated moving (given typical jet speeds circa 1956) at displayed speeds on the order of Mach 2 or less. However this is a highly improbable situation and is still less likely to occur three times on two azimuths 25 degrees apart. The achievable ghost speed is still too low by at least a factor of two. And further, the 3 aircraft responsible would be far stronger targets than their ghosts, which in each case would appear to pursue and overtake them - a very singular circumstance which could hardly escape mention.

Ghosts due to reflections between two aircraft can occur, but such tracks are not preferentially radial and the orientation which would be optimum in terms of ghost velocity (closing head-on, separating tail-on) offers an exceedingly poor reflection aspect. In addition the symmetry of this situation would be expected to generate two ghosts, one on the azimuth of each aircraft, and the aircraft would still have to attain ground speeds of at least Mach 2 to yield ghost velocities of around 4000 mph (closing speed of the two aircraft plus speed of closure between the first reflector and the antenna). It will be evident that the reflection geometry alone is so improbable as to be effectively ruled out as an explanation of multiple rapid targets, and if jets performing at Mach 2 had existed in 1956 at Bentwaters then their own bizarre antics over the airfield would certainly have attracted notice, not only by their own direct radar echoes on-scope but also aurally and visually by ground personnel. In short no conceivable ghost reflection is capable of generating echoes similar to the targets observed.

d.) Missiles & Orbital Bodies

History records no artificial orbital satellites at this date. As for ballistic missile trajectories, note that tracking between diametrically opposite scope sectors implies overhead trajectories, i.e. first-trip detection of local launches rather than multiple-trip detection of remote launches at over-the-horizon distances. Short-range air-to-air missiles were being live-tested in designated rocketry range areas of the UK at this time. The de Havilland 'Firestreak' IR-guided AAM, code name 'Blue Jay', was developed and tested on modified Venom 2's and had been in service on the Royal Navy carrier-based FAW Sea Venom variant since 1954, and RAF Neatishead controlled 22 live rocketry exercises with USAF F-86Ds from Manston and Woodbridge during August 1956 alone. However, the 50-60 mile lengths of the radar tracks rule out the possibility of this type of ordnance. This leaves local surface-launched missiles, but such launches seem improbable.

The first widely deployed SAM missile in use at this date was the American Nike-Ajax, rolled out between1954 and 1958 for domestic air defence. The local US air defence in Suffolk in 1956 was the responsibility of the 60th Anti Aircraft Artillery battalion (Automatic Weapons) who are believed to have been armed with 75mm Skysweeper AA guns and associated fire-control radar at this date (the 60th became reborn as a Nike missile unit of the Army Air Defence Command at Fort Bliss, Texas, after disbanding in England in July 1957). But even if Nikes had been in use in the UK at this time, the range of the Nike (about 25-30 miles) and its speed (about 2000 mph) both fall short of the reported target behaviour by a factor of two. In addition the divergent bearings of the three tracks, implying widely separated origins over the sea, rule out any single land-based launch site, and any kind of coordinated multiple missile test over land in the populous SE of England appears highly unlikely.

e.) Multiple-Trip Echoes

Echoes from remote targets beyond the unambiguous range of the set can appear at spurious displayed ranges, speeds and courses. In AP conditions distant ground returns can be detected in this way, but the resulting stochastic distribution of any echoes remains the same as for sporadic AP at any range. The effect introduces no order. It does nothing to improve the fit with the AP hypothesis, and all the counter-indications considered in section a.) above remain unaddressed.

Echoes from unusual reflectors such as auroral ionisation or the moon also fall into this category. These need not be considered in detail.

Briefly, auroral echoes might conceivably give rise to relatively discrete spot targets by multiple trip at any range on a surveillance PPI , but only in the auroral quadrant - i.e., for N hemisphere radars, the magnetic N quadrant - where the off-perpendicular angle made by the radar line of sight with the magnetic field lines is no more then a few degrees [Leadabrand 1965]. No portion of Tracks A, B or C enters the N quadrant of the PPI. No explanation of linear diametric E-W tracks is offered by echoes of this kind. Further the detection of auroral echoes at S-band at all is highly unlikely. At sensitivities typical in 1956, auroral streamers were considered very marginal targets even at L-band [Plank 1956] and were normally only appreciable at wavelengths twice again as long. Also, records of auroral activity kept by Edinburgh University from 1952 show no visible displays on the dates in question, even at higher latitudes than East Anglia, and the auroral geomagnetic index was low.

Multiple-trip echoes from the moon are possible (see Analysis of Track E, Part 1.e. for detailed discussion) but, to be brief, they are not explanations of hypersonic targets on essentially ESE-WNW trajectories roughly normal to the SW bearing of the moon at this time, and again the diametric crossing of echoes between scope sectors is not possible for this mechanism.

Multiple-trip returns from moving aerial targets such as aircraft are ruled out by several arguments, central to which are the target speeds and the crossing of the targets into opposite scope sectors. Multiple-trip returns from targets with a tangential component of motion will move at displayed speeds slower than the true speed of the reflector, whilst radial motion will always be displayed at the true speed. There is no mechanism whereby multiple-trip effects can lead to spuriously high displayed speeds of k/mph, and such returns are restricted to the azimuth of the reflector so that the angular rate is preserved at smaller displayed range: thus, multiple-trip echoes from a moving target could not cross through 180 degrees unless the target did, implying still more fabulous velocities. (Sporadic detection of two or more remote multiple-trip targets on divergent azimuths which generated the illusion of a consistent track merely complicates an already improbable hypothesis, and there is no intelligible explanation of the systematic reference of the tracks to the scope centre.) The only bodies likely in the atmosphere over East Anglia in 1956 with velocities of an order comparable to that of the targets are meteors.

f.) Meteors

Returns from the trail ionisation of some meteors can be detected on some radars in certain circumstances. Due to the rate of ion recombination the wake can only be detected for a short distance behind the meteor so the returns will typically be displayed as point targets. Since the date of the incidents is close to the maximum of the Perseid meteor shower some attention has been given to the hypothesis that visual reports were due to meteors from this shower, and the 1956 Blue Book evaluation also considered in passing the possibility of meteor-wake returns on Bentwaters' GCA radar.

Meteor returns will very rarely paint as more than a single point echo on a surveillance radar [Blackmer et al., 1969]. But during an intense shower, where rates might be several tens of meteors per hour, an illusion of an extended 'track' might be caused by a chance allignment of a few echoes due to different meteors detected on successive sweeps, depending on the scan rate of the radar (see diagram 'Schematic diagram of tangential Perseid meteors displayed on the CPN-4' and Note below).

In the present case this seems highly unlikely. The Perseid rate at this date is believed to have been some 20 particles per hour (the maximum rate of about 50 per hour occurring at the shower peak, which in 1956 was a couple of days earlier). On a 15 rpm PPI the hypothesis requires a cluster of meteors occurring at 12 times the mean rate, evenly spaced at rather regular 4-second intervals. The probability of such spacing in time, combined with the appearance of a progressive displacement across the PPI, is an inconceivable degree of order for an essentially stochastic process. One would have to assume that the operator's report of a linear track of echoes was essentially delusional, at which point one might as well discard all considerations of physics and logic and shrug off the whole event. Yet, as shown above, the details of Track A recorded by Vaccare show evidence of rather careful measurement of scan-to-scan echo displacements consistent with the rate implied by overall time and distance. This track, like Track B, was some 10 paints long and in both cases the presentation was that of an aircraft echo.

The reported speeds are far lower than the mean true speeds of common meteors, which enter the atmosphere on direct orbits at some 20,000 mph and on retrograde orbits at some 150,000 mph. Since the apex of the earth's way (a tangent to the earth's orbit in the direction of its travel around the sun) lies to the east of the meridian, meteors on an essentially E-W trajectory will be retrograde with relative entry velocities greater than about 100,000 mph. The Perseids are such swift retrograde particles and their short-lived trails will be displayed, if at all, only for one single scan. More importantly, their high orbital inclination to the ecliptic means that for the date and time in question the celestial coordinates of the Perseid radiant differ in azimuth by about 55 and 75 degrees (N) from the (two) plotted target headings, ruling out returns from chance overflights of the radar site by meteors from this radiant. (Note: the Perseid hypothesis is also examined later in a detailed discussion of radar-visual Track C.)

Multiple-trip returns could yield spuriously slow displayed speeds in principle. Meteors (perhaps mavericks unconnected with the Perseids) travelling at relatively slow velocities on the order of 10,000 mph at slant ranges of >120 miles could be displayed as third-trip returns with apparent velocities perhaps 50% lower, but the probability of detection of a meteor, which due to rapid recombination times will be scanned as a point target subject to inverse 4th-power signal attenuation, is negligible at such ranges (120 miles being the minimum third-trip slant range corresponding to zero displayed range; true range over the rest of the track would become significantly greater) at S-band and low power. If detected at all it will almost certainly be as an isolated response on a single scan only. Even if we invoke an abnormally long-lived meteor of sufficient duration to return signals on a number of successive 4-second scans, it is practically impossible for these multiple-trip echoes (necessary, remember, to bring down the displayed speed to the meteoritically-sluggish value of just a few thousand mph) to display as a straight track crossing a diameter of the scope unless this is the true overhead course of the meteor (see below). But a meteor moving on such a radial heading is extremely unlikely to be detected at all, even within the unambiguous range, still less by multiple-trip echoes from beyond the design range.

McDonald concluded that the frequency is about 2 orders of magnitude too high "to afford even marginal likelihood" of meteor-wake returns [McDonald 1972], and this is undoubtedly true for common meteors. Meteors were first observed on radar during World War II on low frequency instruments and the ionisation trail has a very small radar cross-section even at L-band (such as the 23cm CPS-5 used by Lakenheath RATCC); the 10 cm CPN-4 is very far removed indeed from the metric (40 MHz) radars which are favoured for dedicated meteor research work. [Ayer 1969]. Auroral ionisation has been studied at different wavelengths [Leadabrand 1964], generally greater than 50 cm or so, and the frequency-dependent efficiency of scattering from plasmas is well illustrated in these studies. Low-frequency VHF search radars operating in the region of 100 MHz will "see" these plasma regions as echoes with cross-sections on the order of 10,000 m2, but for GHz (L- to S-band) frequencies the cross-section drops by a factor of 100,000 to only about 0.1m2. In general for a tenfold decrease in radar wavelength a one-hundredfold increase in electron density is required for the same reflection efficiency [Blackmer et al 1969]. In other words the cross-section is proportional to the square of the wavelength. Moreover, power returned is proportional to the cube of the wavelength [McDonald 1969 citing Hawkins; see here], as well as to the fourth power of the distance (as for all effective point targets). Thus, the S-band CPN-4 could very probably not detect the plasma trails of Perseid or any other common meteors, certainly not at extreme range by multiple trip, and especially not on approximately radial headings since radar cross-sections are extremely small indeed in this orientation [Blackmer et al, 1969]. Only tangential tracks normal to the radar line of sight (diagram below) exhibit the so-called radiant condition; at grazing incidences any energy returned is limited to scattering from turbulence in the trail [Planck 1956] and would not be detectable.

Schematic diagram of tangential Perseid meteors displayed on the CPN-4

With the radiant of the meteor stream at infinity at 34 azimuth, the lines a-a' and b-b' represent roughly the projected courses on the display of Perseid meteors passing at about 100 km slant range, just beyond the maximum unambiguous range of the CPN-4. Since the meteors will typically only survive for a single scan, their echoes could appear as single point targets, in principle anywhere on these lines but preferentially on their northern arcs close to the scope centre where the off-perpendicular angle of the radar line of sight reduces towards zero. For second-trip echoes from greater ranges out to about 120 miles the angles in the projected courses become more acute, so that the sectors A and B narrow towards the two radii y and y'. In the case shown 3rd trip echoes could in principle also display in the scope sectors defined by x and x', z and z', but the range and trail-orientation in these cases is so unfavourable that the chance can be ignored. A closest approach greater than 120 miles would permit only 3rd-trip echoes. The acuteness of the dog-leg in the projected course will also vary inversely as the angular elevation of the meteor, as will probability of detection. Meteors passing directly over the radar are unlikely to be detected at all, but in this case echoes would always appear on the diameter defined by the true azimuth of 34 regardless of slant range. In short, it is impossible for a Perseid meteor, of no matter how long a duration, to be painted as a series of echoes on an E-W course. However it should be noted that in general the statistically-favoured scope positions for the display of any Perseid echoes will lie along diameter y-y' (see Note in box below).

Because the electronics 'assumes' that a target is within the designed unambiguous range it is not necessarily the case that a radial track displayed on the PPI is due to a target on a true radial heading. However the likelihood of true radial (or rather diametric) headings in this case seems very high. If a moving target is not on a radial heading (i.e., not passing through the zenith but on a tangential trajectory at a relatively low elevation) its displayed course by multiple-trip detection will generally be a curve that approaches scope centre then recedes. It isn't quite impossible that a multiple-trip track should be straight, diametric and symmetrical about the scope centre. But this would require a special parabolic or hyperbolic trajectory with the parabola in the plane of detection and the antenna at the focus, the true range to the furthest point on the curve coinciding with some whole multiple of the unambiguous trip distance (i.e., 60 miles, 120 miles etc.). This is clearly improbable for any target. In the case of a ballistic body such as a meteor it seems impossible - even once, let alone three times - on the grounds of extreme range, non-ballistic lateral motion, and fortuitous geometry. Furthermore it is geometrically impossible for such a straight E-W or ESE-WNW track to be produced either by first-trip or by multiple-trip from a Perseid meteor (see diagram above, also Note below).

Note: The near-coincidence of the diameter y-y' (see diagram above) with the WNW heading of Track A might seem to invite suspicion for the following reason: Multiple echoes might appear sequentially or simultaneously on the scope, either by 1st-, 2nd-, 3rd-, nth trip from the same Perseid meteor or from different Perseids on parallel trajectories, and it can be seen that the distribution of possible scope positions will tend towards the line y-y' as n increases. The distribution only becomes perfectly linear in the unrealistic limit n = infinity, of course, but in general the result of any number of Perseid echoes will tend towards a linear scatter of echoes on this diameter.

However, in the truly extraordinary case that a succession of Perseid fireballs happened to generate echoes on many successive sweeps, there is no mechanism in the process of detection and display that would introduce an artefactual relationship between time and displayed range. The scatter of echoes would be arbitrary on the range axis unless the meteors themselves occurred with an improbable degree of organisation, spaced regularly in distance and at a rate correlated with the scan rate of the radar. If an illusion of an approaching track were to occur once by chance, of course, one would still require another such chance illusion - due to a different series of meteors, inversely coordinated on the range domain - occurring immediately afterwards at the diametrically opposite scope azimuth in order to simulate a track crossing between opposite scope sectors. As mentioned in the text, this also requires bursts of meteors occurring at regular intervals at about 12 times the expected mean rate for the date in question. In short, the similarity of the orientation of y-y' and Track A appears to be entirely without significance.

If the targets were meteors, therefore, they would almost certainly have to be unconnected with the Perseids and on approximately radial true headings (coincidentally) over the site. Only very large, slow fireballs would have any chance whatever of being detected for more than a single scan on the CPN-4 in this orientation.

Very rarely meteorites will have sufficient mass to survive even retrograde entry without complete ablation, creating spectacular visual fireballs. Such a fireball, experiencing significant aero-braking as it penetrates into the region of the tropopause (about 7-8 miles altitude), may decelerate to speeds of a few thousand mph and have a good chance of impacting the earth. The incandescent envelope of such a fireball may be as much as several hundred metres across for a time [Lancaster Brown 1973] and its "head echo" could be detected by sensitive search radars at almost any orientation. Thus, although Perseid meteors are ruled out, although the short range airfield surveillance CPN-4 is very far from the optimum frequency and power for meteir detection, and although the track headings imply very unfavourable trajectories, slow, low-altitude fireballs of some considerable mass still should not be ruled out without examination.

Firstly, if the targets were meteors then detection by multiple-trip returns would be suggested by the rough symmetry of the tracks about the scope centre. In the case of both Tracks B and C the range at signal acquisition is similar or identical to the range at signal loss, and in the case of Track A the difference between terminal range estimates is as low as 5 miles (ratio of means 1:1.58). That these tracks could represent the periods of radar-reflective ionisation of three distinct fireballs, occurring by happenstance inside the CPN-4 maximum drum on two different headings separated by 30 and 55 minutes, is incredible. The range-symmetry would be more plausibly explained by the symmetry of the radiation pattern itself, which suggests bolides entering and leaving the coverage. Since flat 60-mile trajectories at altitudes of a few thousand feet (cutting the bottom edge of the beam at 30 miles slant range, terminal elevation on the order of 1 degree) have no known precedent in meteoritics, these would be bolides entering and leaving the high altitude coverage, and their reference to the scope centre independent of heading suggests multiple-trip detection of high-altitude meteors on near-zenithal trajectories displayed at spuriously small ranges and speeds.

The displayed proximity to the antenna also requires multiple-trip detection: Tracks A and B have no specified close-in signal loss at all, and Track C is displayed as close as 2 miles slant range which (in terms of first trip returns) would correspond to an altitude of again only a few thousand feet. Meteors are not known to behave like this. On the other hand second-trip returns would permit a true slant range of about 62 miles and therefore a more reasonable altitude of several tens of miles.

This conclusion is also consistent with the minimal information available concerning the visual reports concurrent with Track C. Statistically speaking the colour of a fireball is a useful measure of its speed due to the selective thermal excitation of atmospheric gases at different energies, and white light emission characterises very fast bolides with the energy to excite a wide range of gases; monochromatic emissions of various colours tend to relate to lower-energy excitation of specific gases with lower ionisation potential. No colouration is mentioned in connection with any of the visual reports from Bentwaters or Lakenheath; indeed the Lakenheath objects were "white", and BOI-485 specifically states that "all ground observers and reports from observers at Bentwaters agree on colour", favouring a fast, high-altitude meteor rather than a fireball which has been slowed to the observed very low speed (meteoritically speaking) of about 3000 mph by aero-braking in the troposphere. This is weak evidence but consistent with inferences from the radar evidence indicating multiple-trip slant ranges.

But the range performance of the CPN-4's cosecant2 surveillance pattern is by design poor at high altitude/elevation, being optimised to fill the 60-mile-radius toroidal volume of space in which aircraft under its control will be flying (up to a few tens of k/ft) rather than waste power in the stratosphere. Even excellent targets at multiple-trip slant ranges ( >60 miles + about 30 miles track length for 2nd trip; >120 miles for 3rd trip) and altitudes of tens of miles would be extremely unlikely to return a good signal, and meteor-wake ionisation, as we have seen, constitutes a poor target at S-band.

It is also noteworthy that fireballs with the durations stated or implied by IR-1-56 and BOI-485 (up to well over 1 minute in the case of Track C) are very rare and spectacular events, to say the least, which would typically generate a great many reports. In the present case, from 2120 Bentwaters Control Tower Shift Chief, S/Sgt. Lawrence S. Wright, began observing an object that attracted his curosity because it was unusual (the planet Mars was rising on the SE horizon; see astronomical data). He was "aware that the Bentwaters GCA was tracking Unidentified Flying Objects by radar" and was on the look out for same until at least 2220 (20 minutes after the detection of Track B) but he saw nothing. During this time conditions were "dusk to night", sky "clear with unlimited visibility" (notwithstanding the mention elsewhere of a 25,000' ceiling in IR-1-56, remnant high cirrus that dissolved during the evening - see here). Of course later, at 2255 concurrently with radar Track C, an object was seen (either at Bentwaters or, just possibly, at Sculthorpe), which might be construed as a fireball; but the fact remains that during the much earlier time-frame when Tower personnel might be expected to have been most vigilant, and when fast radar tracks were observed at 2130 and 2200, no fireball meteor was seen.

Moreover during this same time frame two airmen were actively hunting UFOs. According to IR-1-56 from 2130 to approximately 2215 an aerial search of the area undertaken by 1st Lt. Charles V. Metz and 1st Lt. Andrew C. Rowe in a T-33 jet trainer resulted in no report of any visible aerial activity: All that these observers reported seeing during their 45-minute search was a bright "star" (evidently also Mars) and the flashing beacon of Orford Ness lighthouse.

Further, the widespread visibility of spectacular fireballs - often reported as UFOs - in the clear sky of a summer's evening would be inconsistent with the fact that that not one report of such a fireball appears to have been generated elsewhere in the UK. Aside from the several other local observations directly associated with this case, at Lakenheath and at Ely, only one report has been discovered for the night of August 13/14 1956, which involved a totally unrelated sighting in Otterburn, Northumberland of what was probably the planet Venus.

The meteor hypothesis is further addressed in a discussion of the visual observations reported from Bentwaters concurrent with the radar detection of Track C (modelled in detail here) at 2255.

g.) Shock waves

Shock waves can be considered in two ways - either as a direct cause of changes in radio refractivity, or as a secondary cause of high-velocity disturbances in a pre-existing layer of abnormal refractivity-gradient. The former mechanism has been proposed in earlier radar-UFO work and will be dealt with first.

Sudden changes in N-gradient over very narrow layers due to shock waves have been considered by Plank [1956, 1959] and McDonald [1970] as possible explanations of unidentified radar targets. Plank remarks:

Very rapid and or erratically moving angels have also been reported. There is a type of nonaircraft echo that suddenly appears, moves for a matter of minutes in a semi-straight line path at velocities of some 600-2,000mph, and then disappears. [One] source might be shock waves, echo being the product of direct back scatter or diversion of energy to ground. Shock waves are very thin, on the order of 10-5 inches; their attenuation rate is only moderate, and the refractive index differences across them can be several hundred N units.

McDonald questions how such shock waves might be a cause of "erratically moving" targets, and further objects that Plank

mentions N-changes of several hundred units; but this is quite unreasonable. First, only temperature jumps and not humidity jumps could accompany shockwave passage. Secondly, in the lower atmosphere, one N-unit change is associated with approximately 1 degree C of temperature change. Third, the Rankine-Hugoniot equations permit one to relate shock-front temperature changes to concomitant peak overpressures; and an over-pressure of, say, 5 psi, is found to lead to a transient shock-heating of only about 30 degrees C (hence about 30 N-units jump across shock-front), yet this is an overpressure not only great enough to take out all nearby windows but to level weak structures and collapse roofs. In brief, the only shock waves capable of giving significant radar-reflecting characteristics would be of rather severely damaging nature, would leave unmistakable after-effects, and yet could influence a radar beam for only fractions of a second.

McDonald also points out that a puzzling feature of many rapid targets in cases where visual observers have also made reports is that there have been no reports of sonic booms.

If this argument is valid, as it appears to be, then shock waves are an unlikely direct cause of hypersonic radar echoes in the Bentwaters case. However it remains possible that they could be a secondary indirect cause.

It is obvious that a wave propagating across an atmospheric layer carries energy proportional to amplitude, frequency and velocity. The energy potential difference of ordinary windshear is equally obviously insufficient to impart the energy of a wave travelling at several kmph, for any realistic frequency and amplitude. That is, the wavelength must be so long and/or the amplitude so vanishingly small that it is hard to imagine it offering the required 'glint' aspect of a highly discrete region of efficient reflectivity to an airfield radar. But other sources of energy are possible.

Electrical storms appear to be ruled out by the general weather picture and specific reports of the absence of thunderstorm activity. But one possible source is shock waves due to meteorite fragmentation, or from the sudden thermal compression/expansion due to the ram heating in front of the incandescent meteor head. It seems conceivable that in ideal conditions such pressure waves might carry sufficient energy groundwards from a sizeable meteor at an altitude around 100 kilometers to generate hypersonic waves in a stable tropospheric layer. Whether such waves could then propagate with detectable amplitude for distances of around 60 miles is a different question.

I have not developed this hypothesis any further because the altititude of the lowest suitable layer for which there is any aerological evidence (i.e., discounting a possible surface layer at a few tens of feet) would be near the tropopause. One may of course hypothesize undetected layers at any altitude between the radiosonde data points; but it also seems likely that any energy remaining undissipated in passage through inversion layers at the tropopause and above would be too attenuated to cause disturbance of a hypothetical layer in the first few thousand feet.

So we are to consider a layer at 31,000' or above. Even direct specular reflection of radar energy back to the antenna from such a layer, at the highest possible 45 elevation of the CPN-4 main beam coverage (an effect lying a long way outside any understood abilities of the atmosphere at S-band and low peak powers), could not be displayed on the scope at a range less than about 8 miles. Yet there is no mention at all of any unusual close-in disappearance of two of the targets, and the only close-in range data quoted (for Track C) show a target displayed to within 2 miles of the radar.

On the other hand, if the radar beam intercepts a tropopausal layer at any realistic grazing angle for partial forward scatter then the slant range to the layer must be so large that returns by multiple trip become likely, in which case spuriously short displayed ranges of two or three miles become possible. However it is then not possible for the multiple-trip echo to cross "immediately", as reported, from one scope sector to the opposite sector under these conditions. That is, it would not appear as the track of a high-speed object flying continuously overhead. The minimum slant range to the interception-point on the layer, being on the order of 30 miles for second trip, implies a blind zone which could not be crossed by a shock wave propagating at 1500 mph (half the reported speed of the echo displayed on the scope, as required by the reflection geometry) in less than about a two and a half minutes, or around 40 scans of the radar.

It is possible that the report is incorrect, or that pulses were scattered from pairs of different shockwave packets in such a way that they happened to give the illusion, on as many as three occasions, of continuous zenithal tracks on straight headings. The probability of these suggestions should be weighed against the evidence of internal consistency in the range/speed/altitude data for Track C discussed here.

Martin Shough 2003

Archive Index