Analysis of the Bentwaters Slow Targets - Part I

An Opinion by Martin Shough

At about the same time that the first rapid target, Track A, was detected by A/2C Vaccare crossing a surveillance scope of the Bentwaters GCA radar on a WNW heading towards RAF Lakenheath, a cluster of targets was also seen moving at much slower speed on a different heading. Another operator gave his report of the track (which we will identify here as Track D) as it appeared on his scope. IR-1-56 from Bentwaters summarises the events as follows:

T/Sgt Elmer L. Whenry, GCA Operator, 1264th AACS Squadron, RAF Bentwaters . . . stated that 12 to 15 unidentified objects were tracked by the Bentwers GCA (AN-MPN-11A) between 2130Z and 2155Z, 13 August 1956. This group was picked up approximately 8 miles southwest of RAF Station Bentwaters and were tracked on the radar scope clearly until the objects were approximately 14 miles northeast of Bentwaters. At the latter point on the course of these objects, they faded considerably on the radar scope. However, the 12 to 15 objects were tracked to a point about 40 miles N.E. of Bentwaters. At the approximate 40-mile range [the] individual objects in this group appeared to converge into one very large object which appeared to be several times larger than B-36 aircraft due to the size of the blip on the radar scope. At the time that the individual objects seemed to converge into one large object, the large object appeared to remain stationary for 10 to 15 minutes. The large object then moved N.E. approximately 5 or 6 miles then stopped its movement for 3 to 5 minutes then moved north disappearing off the radar scope [maximum CPN-4 range = 60 miles].

Sgt Whenry stated that the 12 to 15 unidentified objects were preceded by 3 objects which were in a triangular formation with an estimated 1800 feet separating each object in this formation. The other objects were scattered behind the lead formation of 3 at irregular intervals with the whole group simultaneously covering a 6 to 7 mile area. Prior to consolidation into one object 40 miles N.E. of Bentwaters. Course flown by this group of objects had slight deviations from S.W. to N.E.

Sgt Whenry added that these objects appeared as normal targets on the GCA scope and that normal checks made to determine possible malfunction of the GCA radar failed to indicate anything was technically wrong. Sgt Whenry estimated that the unidentified objects in this group moved at the rate of between 80 and 125 miles per hour. He computed this speed by using the range margins on the GCA scope.

The track description here leads to a total length of 73 miles, and it is immediately apparent that this computed speed cannot be the bracketed mean speed over the whole track, since even at a constant 125 mph (to use the highest value) the targets would take some 35 minutes to cover 73 miles, whereas the observation reportedly lasted only 25 minutes. And this is not to take account of the reported periods of stationarity, which alone occupied a total bracketed duration of between 13 and 20 minutes. Adopting the lower of these estimates leaves only 12 minutes of mobility, which implies a mean speed of 365 mph. Therefore it is necessary to assume a) at least one quantitative error somewhere in the chain of observation, reporting and transcription, or b) that the cited range of speeds was measured by Whenry only for one limited portion of the track. As will be seen later, this inconsistency can probably be resolved by correcting a single-digit clerical error for which there is internal evidence elsewhere in IR-1-56.

Track D, RAF Bentwaters, 2130* - 2155Z
[*Note: Internal and independent evidence discussed in the text indicates that this start time given in IR-1-56 is in error and should read '2100'.]

Track D (above) has been regarded by some analysts as due to ground returns via partial reflection from moving waves on an inversion surface. This hypothesis was originally developed by Borden and Vickers [1953] to interpret the Washington National Airport observations of July 1952. Thayer [1969, 1971] considered its applicability to the Bentwaters report, concluding that although certain features (such as the two periods of reported stationarity) were inconsistent with this mechanism a general direction of movement related to that of the wind and roughly appropriate speeds were suggestive of this cause. McDonald on the other hand remarked [1970, 1972]:

The strength of the individual echoes (stated as comparable to normal aircraft returns), the merging of the 15 or so into a single echo, the two intervals of stationarity, and final offscope motion at a direction about 45° from the initial motion, are all wholely unexplainable in terms of [partial reflection].

Klass [1974] adopted Thayer's early partial reflection model without addressing the inconsistencies to which Thayer had drawn attention, resting his argument on the approximate concordance of the average target motion with winds-aloft conditions. More recently Thayer [personal communications] concluded that a refined model involving forward scatter from regions of clear air turbulence associated with a wind shear across an inversion boundary offered the best meteorological fit to these observations.

This class of hypotheses is considered in Part 2, along with the more common types of anomalous propagation echo, and in detail in Part 3. Part 2 also covers possible electronic countermeasures "spoofing" techniques such as active jamming, and the related problems of radio frequency interference or spurious internal noise signals, together with sidelobe ghosts and multiple reflection effects.

The class of hypotheses considered here in Part 1 covers direct returns from radar-reflective targets (with the proviso that such targets are not necessarily exclusive of complicating factors such as sidelobe echoes, ghost reflections and multiple-trip echoes):

a) Aircraft
b) Birds
c) Insects
d) Balloons
e) Meteors, Orbital/Suborbital Bodies
f) Precipitation
g) Ionisation Phenomena

Track D

a.) Aircraft

One might wonder if the reported "fading" of the receeding targets at 14 miles is symptomatic generally of disturbed propagation or other causes of phantom echoes. The answer is: not uniquely. A tangible radar-reflective target such as an aircraft can suffer attenuation proportional to displayed range for three reasons: one electromechanical; one electromagnetic; and one physical.

Firstly, due to the convergence of the radial trace lines of the PPI towards the centre of the tube face, spot-brightness per unit area decreases from the centre towards the perimeter of the display, so that echoes of the same given strength exhibit a dimmer presentation with increasing displayed range. Secondly, reflected signal intensity from point targets varies inversely with the 4th power of the range from the antenna. And thirdly, variation in aspect imposes a further unpredictable fluctuation which in actual practice tends to result in decreasing cross-section with increasing range due to the typical configurations of aerodynamic bodies. Since many aeroforms (aircraft and birds, for example) have plan sections very much larger than tail-on sections, as elevation and range vary so the radar beam will intercept the target at an angle which illuminates more or less of its underside, so that a receeding aircraft which is nearly in plan at short range and high elevation will approximate more closely to its tail-on section as it gains range, simultaneous with a diminishing incidence angle on the undersurface tending to specularly reflect energy away from the antenna. (It is diverting to note en passant that this effect will be partly proportional to wing aspect ratio, and would be especially significant in a planform of very low aspect ratio - for example, a circular wing aircraft.)

Having no information on the form of any hypothetical "objects" in this case it is not possible to address the fading of the targets in terms of changing aspect, except to say that the superimposition of these three curves of echo diminution could cause highly non-linear fading to occur, and it is possible that a catastrophe effect might ensue when cross-section due to aspect dropped through some threshold figure. This of course ignores possible changes of configuration or flight attitude on the part of the "objects".

The echoes reported are in this respect not necessarily inconsistent with returns from aircraft. Arguments in favour of this hypothesis are: The apparent "formation" of the lead targets - a loose cluster preceded by three targets spaced uniformly in a triangular pattern; the reported order of speed which could equate with light piston-engine aircraft or helicopters; the change of heading from NE to N, which is less easy to equate with meteorological phenomena; the presentation of the targets, which appeared comparable to "normal" aircraft targets. On the other hand, apart from the circumstantial improbability of a flight of some 15 or more aircraft covertly overflying a military airfield without identification or authority, there are several arguments which weigh heavily against this hypothesis.

The implied altitude of the targets at acquisition, range 8 miles SW of the field, would be on the order of 1000 feet or less for the CPN-4 pattern, which suggests that the altitude of aircraft when over the field would be quite low (assuming typical constraints on helicopter/light aircraft performance in climb). This conclusion would be supported by the assigned mean T-33 search altitude of about 3,500' in the opposite sector, an estimate presumably derived from the slant range of the targets entering and exiting the radar shadow (see later). So many aircraft at low altitude would certainly be seen and/or heard by ground personnel.

The negative result of the T-33 search is inconclusive in respect of some hypotheses, but in view of the required lighting of helicopters and light aircraft which might have inadvertently strayed over the base without flight plan it could be held to be quite damaging to this hypothesis. (The same objection presumably need not apply to a covert action - perhaps a test of defence readiness - in which neither interception nor identification might be desired. The scenario seems inherently improbable, however.)

The strongest internal reason for doubting the aircraft hypothesis is the merging of the targets into an echo which remained motionless for significant periods of time. Considering first the merging, the question arises whether this should be interpreted as a real closure or an effect of the coarsening of resolution with range.

The resolution cell is defined by three values, two of which are range dependent: resolution in azimuth is determined by beam width (strictly speaking by the relation r = 1.22 rad [l / D], where l = wavelength and D = antenna width) which therefore varies with range; the vertical height of the cell is determined by the vertical coverage, again varying with range; and the range resolution is determined by the 1/2-pulse length, which is independent of range. Resolution of targets in plan on the PPI is therefore dependent on the beam width and the pulse length of the set. The beam width of the CPN-4 is given as 2.3° at the half-power points, which for the sake of simplicity it will be sufficient to approximate as 2° giving an azimuth resolution at 40 miles of about 7400', and at 60 miles of about 11,100'. The range dimension of the cell (1/2-pulse length) for a 1 microsecond pulse 984' long is about 500'; for a 0.5 microsecond pulse, 250'. So at the point of closure of the targets the cell is about 1.4 miles by a few hundred feet, whereas the previously observed spacing of the targets was much greater on the range axis: The 3 lead objects alone were in a triangular formation 1800' on a side, followed by the remaining targets scattered to 6 or 7 miles on the range axis. Azimuth separation would therefore no longer be resolvable at these ranges; but a factor-of-100 difference between reported range-axis separation and PPI range resolution indicates that the merging of the targets was, as implied, a "real" closure in terms of displayed range.

It would be possible for some 15 aircraft to fly in line abreast with a close lateral separation of less than about 500' (= 1.4 miles/15), also maintaining alignment to within a few hundred feet, for some minutes (although it is presumably unlikely except for display purposes, which is hardly relevant here). And it is true that PPI discrimination would not be so fine as to require this degree of contiguity: Integration of numerous close blips on the tube phosphor might well cause targets with a much larger overall range-azimuth distribution to present as a single echo. But it would not be possible for light planes to maintain station even within the practical PPI resolution cell for many minutes, except by climbing to 30,000' and performing an aerobatic turn into the 100 mph WSW headwind (twice, for several minutes), which would be both incredible and pointless.

Helicopters, on the other hand, could present such behaviour on the PPI, and it should also be remembered that the resolution cell has a vertical dimension of several tens of k/ft. It is interesting, therefore, that the area of closure and stationarity is over the North Sea, 15-20 miles off the Norfolk coast and thus over sufficiently deep water to suggest the possibility of a naval supply and recovery exercise.

If the targets were helicopters then low altitude would be indicated because of typical limitations on hover out of ground effect and the practicalities of any such exercise. But the normal horizon of the CPN-4 for even very good targets at a range of 60 miles would probably be above 7,000', which would be approaching the approved ceiling for many helicopters. Even the closest point of stationarity (range 40 miles) implies "hovering" at an altitude above 3500' for a period approaching 15 minutes, and helicopters simply do not behave in this fashion.

It is possible that superrefractive conditions might allow targets at this range to be picked up at altitudes below the normal horizon; nevertheless it seems unlikely that a major air-sea exercise involving so many helicopters would be in progress among the shipping lanes and fishing grounds close to the ports of Lowestoft and Great Yarmouth. It should also be borne in mind that helicopters are not normally flown in very close formation due to the danger of downwash from the rotors, and so large a number of vehicles in close proximity might entail considerable risk. Also there is no meteorological evidence from balloon soundings of significant superrefraction at any level of the troposphere, and indeed surface conditions appear to have been unusually subrefractive, which would lift the radar lobes and reduce the effective horizon. A high-level tropopausal inversion suggests the possibility (unconfirmed owing to absent humidity data) of an elevated duct above 30,000' which could perhaps allow second- or third-trip echoes from remote targets many miles out to sea. But this offers no explanation of the continuity of tracking from the SW over the radar site.

In summary, fixed-wing aircraft seem improbable for several reasons, notably the absence of visual/aural detection and the stationary episodes. It is noteworthy that the lower reported ground speed of the targets, given a roughly following wind of 30 knots at 6000', would imply an airspeed as low as about 45 mph, dropping to 35 mph at 10,000' and to as little as 20 mph above 16,000'. Therefore the low altitude implied by the short slant range of the targets at acquisition, and by the inferred operator-estimate of altitude during overflight of the radar, is also implied by the demand that airspeed be above an aircraft's stalling speed.

Helicopters seem to offer a less implausible explanation and the suggestion of some sort of naval exercise can to some extent rationalise the rather singular behaviour of the targets. But in view of the secrecy implied by the absence of identification by radio, flight plan or IFF transponder, which secrecy appears inconsistent with overt flight at low altitude over a military airfield; the absence of visual/aural ground observations at Bentwaters; the negative result of the aerial search; the extreme rarity of such large helicopter formations; and the roughly 15-minute confinement within a range resolution on the order of hundreds of feet through discontinuous manoeuvres, the probability is unarguably very low.

It is certainly relevant that the first concern of a GCA radar unit is safe management of low-altitude air traffic in landing patterns over the field, and strenuous attempts to identify a large group of inbound aircraft would have been made by GCA and by Tower controllers who were "aware that Bentwaters GCA was tracking [the targets] at this time." Aircraft were considered but discarded as an explanation by Captain Edward Holt, Air Targets Officer, in IR-1-56: "Several aircraft were in the Bentwaters area at the time of these sightings but these could not have been mistaken for the Unidentified Flying Objects."

b.) Birds

Flocks of birds can be tracked on surveillance radars at ranges of tens of miles and could present unresolved targets with cross-sections comparable to aircraft at similar ranges under certain conditions, although typically such echoes are not similar to the narrow presentation characteristic of aircraft on the PPI. The cross-sections of various birds have been measured experimentally at different wavelengths [Konrad, Hicks & Dobson 1968].

At S-band the total reflectivity of a flock of 500 pigeons with broadside radar cross-sections on the order of 100 cm² would, assuming each bird to contribute without significant occultation, approximate that of a small aircraft with a tail-on section of about 5 m² at the same range. The same flock tail-on, however, would present a section of only about 0.01 m², or 0.2% of that due to a poorly aspected small aircraft. A flight of several tens of duck with tail-on sections >25 cm² (24 cm² has been measured at 400 MHz; the cross-section at S-band would be greater) distributed within a spherical volume of radius several tens of metres could present an unresolved target comparable to that of a small aircraft down to ranges as small as a few miles. Many birds maintain clusters over considerable distances and some, such as geese or ducks, exhibit formations superficially similar to the leading "arrowhead" of targets observed at Bentwaters.

Klass [1974] points to the presence of a large bird sanctuary in the area; in fact there have long been several large wildlife reserves in the East Anglian coastal region which are home to large numbers of migratory and non-migratory birds, including the major sanctuary at the mouth of the River Ore below Orford Ness, some six miles SE of Bentwaters. It is likely that there were unusual concentrations of birds in the region circa 1956.

The airspeeds of different birds have been measured by tracking radars and these can be compared with the observed target speeds [Houghton 1964]. Groundspeeds will be the resultant of the two vectors of windspeed and the bird's airspeed. Bird airspeeds vary from <5 to >40 knots. 90% of birds fly below 5000' and birds above 13,000' are very rare, although occasional bird-strikes have been reported by aircraft at around 20,000' (some 75% occur below 3000'). The median nighttime altitude of birds over SE England measured by radar over a period of a year was found [Eastwood & Rider, 1965; 1967] to be 1800', whilst the daylight value was 1000'; but at the same time there is a marked tendency for birds to fly at lower altitudes in the absence of cloud cover (weather in the Bentwaters area was clear with excellent visibility).

Birds flying at low level would have almost the full benefit of 230° tail winds at 6-12 mph; birds flying at 6000' would have less than the full benefit of winds from 260° at 34 mph. To achieve the lowest estimated target speed of 80 mph birds would need to fly at about 10,000'. To achieve the highest reported speed of 125 mph would require the fastest of common flocking birds, the duck, flying at about 20,000' with an airspeed at the upper limit of its range. These values are improbable in terms of their statistical distributions, and furthermore inconsistent with the low altitude implied by the target ranges at acquisition and inferred from the T-33's assigned search altitude.

Because of the nonlinear relation between range and power returned on a radar scope, birds detected at close range can return unexpectedly strong echoes, and an inexperienced operator who failed to mentally 'correct' for this inverse-4th power signal attenuation might be misled by the similar scope presentations of a nearby bird echo and a distant aircraft echo. However, with regard to the merger of the targets into a stationary echo for some minutes, this behaviour clearly fixes a range at 40 miles, and it will be seen that this implies a very considerable density of birds.

The estimated equivalent cross-section of the integrated echo (about 200 m²; see below) would correspond to several tens of thousands of ducks. This order of magnitude checks with the value required for the 15-or-more "normal targets" before integration: If these targets had a presentation similar to the smallest tail-on aircraft aspect likely to have been regarded as "normal" by a military operator - 2 m², the bottom end of the typical range of values for a poorly-aspected small fighter - then a total equivalent echoing area of >30 m² is implied, corresponding again to >10,000 tail-on ducks. Both of these figures are minima since no account has been taken of occultation. There is obviously a great deal of approximation here, but it is clear that we require on the order of tens of thousands of large birds occupying a volume (defined by the resolution cell, with some allowance for imperfect PPI discrimination) equivalent to a sphere of, say 500m radius or larger. This density is actually not as unreasonable as it may sound, giving about 10,000 m³ per bird for a flock of 50,000, or an average separation of about 15 metres. Nevertheless, the absolute number of birds required seems impossibly large.

The seaward motion of the targets might suggest seabirds, perhaps flocking around a night fishing fleet and detected at unusually low altitude due to superrefractive conditions. But gulls do not typically fly in formations of discrete clusters over long distances, or congregate in such enormous numbers, and their low airspeeds of around 20 knots would not yield even the lowest estimated target speed below altitudes of >16,000' (winds 260°, 55 knots).

The lead "formation" of 3 targets in a triangular pattern might conceivably be produced by large flocks of birds; however 15+ such flocks maintaining compaction and separation over a distance of some 50 miles would be extraordinary. The stop-go motion of the integrated target after convergence with stationary episodes of many minutes within the limits of PPI resolution are also features difficult to account for in terms of birds. Individual echo fluctuations due to wing motion might be masked in large flocks, but fluctuations due to overall movements and occultations within the flock would be expected to distinguish the PPI spot from that due to a "normal" aircraft target over the space of many sweeps.

In summary, the altitude/speed performance of birds, the movements characteristic of birds, the numbers typical of flocks of birds (in a UK context) and the PPI presentation of birds all appear to be inconsistent with the reported behaviour of the targets.

c.) Insects

At first sight, insects in any conceivable quantity would seem most unlikely to reflect signals of the strength indicated. Something in the region of 30-50 million moths, for example, would be required to produce a signal of the strength indicated at 40 miles. But migrant butterflies such as the cabbage white, for example, have been known to swarm in comparably huge numbers over southern England and might conceivably present as unusual, even prominent, targets on S-band radar scopes. However, neither the speed nor pattern of motion of these targets is characteristic of the behaviour of insects.

Insects, as well as parachuting spiders, can be lofted and carried in concentrations by atmospheric movements. The occasional radar detection of convective thermals and advective breeze fronts, for example, is thought sometimes to be in part due to insects caught up by them as well as to birds riding them (possibly to feed on the insects). However, not only does the ordered disposition of the targets (in particular the lead "formation") conflict with the likely dispersal of wind-borne insects, the reported speeds are also inconsistent with insects whose speeds would be essentially that of the wind at their altitude. Insects are only rarely encountered above a few thousand feet, with most occurring in the first thousand feet. Between 1000 and 4000', typical densities have been measured (over the southern United States) of around 1 per 2 km³, falling significantly at higher altitudes [Glick 1939, cited in: Blackmer et al., 1969], and one would certainly not expect to find huge migrant swarms in the stratosphere. Therefore insects of any kind offer a poor hypothesis in this case since even the lowest reported target speed could not be attained by insects below about 30,000'.

d.) Balloons

A radiosonde balloon launched from the area of Ipswich airport (10 miles SW of Bentwaters) could, climbing at a typical rate of 1000-1200 fpm [Lally 1969] and propelled by 5-10 knot near-surface winds on a 50° heading, climb above the lower limit of Bentwaters' CPN-4 coverage in something less than a minute at a range of about 8 miles, consistent with the location of the targets at acquisition. In the first 5 minutes' climb to 6000' (at 1200 fpm) the balloon would travel a little over 1 mile across the ground. Between 6-10,000' it would travel about 2.2 miles in about 3.5 minutes. About 5 minutes later at 16,000' it would have covered a further 4.5 miles, and at 30,000', after another 12 minutes, it would be some 25 miles further on. At this point the balloon would be travelling NE at about 100 mph, range 25 miles from the antenna and increasing.

However the initial speed at acquisition would be an order of magnitude lower than the mean of the range of speeds reported and the balloon would not achieve that mean for about 25 minutes. In addition, the displayed initial speed would be lower than the balloon's true airspeed since the rate of increase in elevation due to ascent at about 11 knots is balanced against closure due to a ground speed of about 10 knots, so that slant range decreases more slowly than ground range; indeed, at first detection the balloon would probably appear stationary on the PPI. (It is even possible that such a balloon might be displayed moving SW very briefly - in the opposite direction - if radar contact were made low enough.)

Alternatively a balloon launched some 30 minutes earlier from a location about 40 miles SW of Bentwaters could be travelling at about 100 mph at 30,000' at the point of acquisition (assuming that the GCA surveillance scope was not being monitored, or was not switched on before this time). The balloon would normally continue to climb until it shattered. Most neoprene or rubber radiosondes of the period would burst between 80,000' and 100,000' but very occasionally they have been known to climb to 150,000' [Lally 1969]. Such a balloon could therefore possibly survive as long as 80 minutes after acquisition, and after slowing to 40 knots at 50,000' would be expected to accelerate again as it encountered winds which (statistically) tend to increase with altitude. During the approach from the SW the gain in elevation would decrease the displayed rate of closure below the wind speed at the balloon's altitude (minimally in this case due to the smaller effect on slant range contributed by a lower angular rate of climb) and during recession to the NE the same effect would increase the displayed speed.

It is not possible that such a balloon would display the range of speeds reported and achieve the required range NE from the antenna in the 25 minutes of observation indicated by Captain Holt in IR-1-56; but it is possible in the 55 minutes which emerges from correction of an evident typographical error in the start time of Track D (see Part 2.).

However a radiosonde balloon and its small instrument package, even if fitted with a foil or mesh corner reflector (perhaps < about 1 m²), is not a very large target at the high altitudes/slant ranges implied. The performance minima quoted for the CPN-4 against a small jet and a cosecant² vertical coverage optimised for its airfield control function (i.e., minimum power wasted at high altitudes above local traffic patterns) suggest that at any slant range a jet fighter would be a marginal target at altitudes of 40,000' or so. For the the two models described above the balloons would reach altitudes during 55 minutes of about 60,000 and 90,000' respectively, and their detectability over major portions of the track would be in very considerable doubt.

Much larger polyethylene research balloons have been flown regularly since the late 'forties. These attain maximum diameters of 200' or more at typical float altitudes of 120-140,000' and are several tens of feet across even at launch. Such balloons could present strong targets at moderate slant ranges in view of their often large instrument payloads of as much as several thousand pounds [Lally 1969]. In terms of balloon motion the fading of the targets at 14 miles could correspond to a gain in altitude: 14 miles on the PPI is a slant range of 74,000' of which altitude is an unknown but possibly significant component, and the fading could correspond to the rise of balloons into the upper edge of the beam where signal strength is beginning to fall sharply. Slant range continued to increase for a further 200,000' or so, however, which would imply that the balloons at this time fortuitously levelled off at a less-than-typical float altitude to remain within the beam.

Some consistency can therefore be argued between some features of balloons and some features of the target behaviour. Arguments against balloons are rather strong however. One obvious difficulty with the balloon hypothesis is the simultaneous release of so many balloons.* Radiosondes may occasionally fail and leaks are not rare, so that a second balloon might be released after a few minutes if, for example, a reduced rate of climb was noticed by the tracker. But the rapidly widening altitude gap between the leaking and the sound balloons would mean that they were soon separated by wind shear and dispersed. Fifteen launches would be entirely without any normal practical purpose. Small clusters of balloons might be employed to lift heavy instrument packages, and it is perhaps conceivable that they might be prematurely cut adrift from the payload at low altitude without rupturing; but the redundant balloons would of themselves offer very poor radar targets. (Toy balloons are sometimes released in large numbers, but of course would offer the very poorest of radar targets.) The launch of large numbers of big polyethylene research balloons is so intrinsically unlikely as to be ruled out. In general the problem with any 6-7 mile cluster of balloons climbing through varying winds is the dispersal that would take place over a period of tens of minutes.

Nevertheless a large group of very efficiently radar-reflecting balloons, however improbable, is presumably possible in the context of some classified exercise, hoax or extraordinary mishap; and it is also possible, if highly improbable, that they might remain together - even simulate a stable "formation" of targets. It should therefore be noted that a balloon at a displayed range of 8 miles and at an altitude of 30,000' consonant with the required wind speed would be at a ground range of 5.7 miles and an elevation of 45°. Thus at the time of acquisition the balloon would already be a very marginal target for beam-angle reasons, and would increase this angle at the rate of about 2°/minute whilst concurrently climbing to less and less favourable altitude. It would be detectable only briefly, and would then remain invisible to radar for some 75 rotations of the antenna. Yet the targets were "picked up approximately 8 miles southwest . . . and were tracked on the radar scope clearly until the objects were approximately 14 miles northeast of Bentwaters", with no indication of the >5-minute signal loss that would occur due to the passage of a 100 mph target through the radar shadow at 30,000'. Moreover, an altitude on the order of 1/10 this height would be consistent with the T-33 search altitude and the inferred operator-estimate made from a brief (matter of seconds) signal loss during the targets' overflight of the antenna.

Eliding for the moment the fact that balloon speeds at low altitude would be 1/10 of the target speeds at acquisition, as has been shown, it is further the case that balloons would still be subject to a long period of radar-invisibility even if released so as to climb into the beam at 8 miles, because of their lower speeds and a higher ratio of wind-speed to rate-of-climb: after some 10 minutes they would have accelerated from near-zero to about 45 knots, disappearing from the scope at a slant range of about 3.5 miles (altitude 2.5 miles, elevation 45°) and remaining undetectable for at least 6 minutes before re-emerging into the beam at a slant range of about 5 miles travelling at 65 knots. A further 10 miles of travel at a mean 70 knots would bring them to the point of fading (14 miles displayed, altitude about 25,000') in about 7.5 minutes. Thus for at least some 25% (>90 scans) of the period during which IR-1-56 states that the targets were "tracked clearly" before fading, such balloons would have been quite invisible to the radar, whereas a target on a typical flat trajectory at the mean estimated speed and at the highest altitude consonant with the T-33's assigned search altitude (5000') would have been expected to cross the zenithal shadow in no more than about 60 seconds. This quite ordinary duration of signal loss would very likely not be reported; but a duration six times as long, following initial displayed speeds climbing slowly from about zero mph to little more than about half Whenry's minimum estimate, would be highly significant features of the track whose absence from the report is without explanation.

The subsequent stationarity of the merged echoes is at first sight impossible to reconcile with balloons. But it is worth emphasising that there are other circumstances (in addition to high angular rates of climb at short ranges in light winds) in which a balloon could present a stationary return on the PPI, or even appear to reverse direction into the wind, even though the hope of explaining even one stationary episode of many minutes duration remains small.

Research balloons would usually be cut down by a squib remotely detonated from the tracking aircraft in order that instrument payloads may be retrieved. Since the PPI displays slant range to the target, an object with minimal buoyancy losing altitude from a high elevation could conceivably be displayed at reducing range even though the ground range is actually increasing. It is therefore possible (although the geometry of the situation is improbable) that a partially deflated cut-down balloon, or its chute-borne payload, could present a stationary echo on the PPI throughout a radar-visible descent of perhaps several minutes.

For example, a balloon at 130,000' and a slant range of 40 miles subtending an elevation of 37° has a ground range of 33 miles, so that to maintain roughly constant displayed range the ruptured balloon/parachute and/or its payload target would have to travel a ground track of 7 miles in about 10 minutes, or about 40 mph mean ground speed. The likelihood of this can be roughly estimated. It will be apparent that the component of PPI velocity due to ascent before rupture would be small even at high elevations (about 10 mph along the line of sight) and would be negligible at 37°, so that displayed velocity is essentially that of the wind at the altitude and thus an approximate true ground speed for the intact balloon. If windspeed at 130,000' were, say, 130 knots (150 mph), reducing fairly constantly towards near-zero at the surface with a mean of about 70 knots (80 mph) it is apparent that a partially buoyant sinking balloon with a payload of a thousand pounds might well have a mean groundspeed during descent of around half this mean figure.

This having been said, the probability of such a 25-mile descent through changing winds achieving this result must be extremely small, and fifteen simultaneous ruptures or cut-downs would be even more unlikely than fifteen separate launches, especially considering that the range resolution of the CPN-4 (1/2 microsecond, or about 1/70 the range differential during descent) places great strain on the plausibility of even one such event. In addition cut downs are normally accomplished at or just before sunset (approximately 1930 GMT for lat. 52°N, long 1°E, August 13) or after dawn when visibility of the sunlit balloon is optimum. That there were two such stationary episodes has no credible explanation in terms of balloons or other windborne objects, nor can the real closure of the targets be understood. In summary the balloon hypothesis appears extremely improbable.

e.) Meteors, Orbital/Suborbital Bodies

Satellite re-entry debris is ruled out on the basis of date alone, since the first artificial satellite, Sputnick I, was not orbited until October 1957. Earlier classified tests of prototype launch vehicles, or of guided/ballistic missile systems are presumably not impossible, but the observed speed and duration of the track are inconsistent.

Multiple-trip echoes from hypothetical orbital or suborbital ballistic projectiles far beyond unambiguous range could theoretically be detected at low elevations on long-range surveillance scopes, but the hyperbolic track geometry - restricted to a narrow scope sector - and the displayed speeds in such a case would still bear no similarity whatever to the (essentially diametric, linear) behaviour reported, and the large number of targets could not be explained even by invoking multiple reflection effects of baroque complexity.

Meteorite fragmentation is ruled out by several obvious arguments, in particular that there is no known or hypothetical mechanism whereby even the slowest bolides could glide for several tens of minutes within range of an airfield surveillance radar, multiple-trip echoes or not.

f.) Precipitation

Precipitation echoes are as a rule diffuse areas of considerable extent and would not resemble point targets. The detection of such echoes depends on wavelength and antenna polarisation, and they are commonly minimised or eliminated in modern ATC radars by the use of non-linear polarisations (even though efficient weather-rejection is not always considered desirable by aircrews, who might be routed into the middle of rainstorms!). If appropriate rejection methods are not used it is possible for shorter-wavelength surveillance radars to detect rain and hail, and possible also that in mixed showers hail alone would be detected. Hail does tend to form in subcells on the order of 100m in size and shower durations peak at about 20-25 minutes [Geotis 1963]. Hail showers could therefore conceivably present as a group of discrete echoes (associated rain possibly being below threshold droplet size for detection at S-band) moving across the scope with roughly the speed of the wind at moderate altitudes (probably being lofted intermittently by updraughts to quite high altitudes).

This partial match with the targets is, however, outweighed by a number of countervailing arguments. The convergence and stationary episodes cannot be accounted for by hail cells; hail is associated with frontal systems and the updraught cells in which it is produced would tend to be scattered transversely, not parallel, to the direction of movement as well as covering a rather wider area; and it is unlikely that even remarkably discrete hail cells could present echoes indistinguishable from normal aircraft targets over several tens of minutes, especially in view of the wind shear inevitably encountered during the vertical circulation necessary to repeatedly loft the hail to altitudes at which wind speeds could account for the range of target velocities.

Finally, according to intelligence reports IR-1-56 and BOI-485, although 1/10 scattered cloud appeared about 5 hours later at 0300Z August 14 the sky at the time (Bentwaters Weather Detachment report for 132100-132200Z) was clear with some coastal haze, excellent visibility and no precipitation. Such a summer's night is not remotely suggestive of the conditions for the formation of hail.

g.) Ionisation Phenomena

Echoes from lightning channels and sferics are totally dissimilar to the targets observed, and are subject also to the objection that meteorological conditions were inappropriate. IR-1-56 confirms: "No thunderstorms were located in the area of the sightings", and the later time-frame covered by the BOI-485 weather report to 0400 also gives thunderstorm activity as "Negative".

Auroral ionisation can be detected in the N sectors on surveillance radars at moderate and high latitudes, but the S-band wavelength of the CPN-4 is remote from the optimum metric (VHF) wavelengths at which auroral cross-sections can be large, and the motion, the disposition,the presentation and the initial azimuth of the targets are all entirely inappropriate.

Ball lightning, although statistically related to the passage of electrical storms, has occasionally been reported in clear weather but very rarely with a duration exceeding a few seconds. The 15+ steadily moving targets in this case appear to have no similarity whatever to canonical ball lightning.

* * *

In summary, there appears to be no plausible explanation of these targets in terms of known radar-reflective bodies in the atmosphere. Analysis of the Bentwaters Slow Targets - Part 2 considers other explanations such as anomalous propagation, ECM simulation, radio interference, system noise, sidelobes and multiple reflections.