Intelligence fusion has been defined in doctrine as the process of collecting and examining information from all sources and intelligence disciplines to derive an assessment of detected activity that is as complete as possible. It is a subject that is well worth considering in the context of Allied operations against Nazi Germany's V1 flying bomb in 1943-44. Posted in memory of Anthea Laing, who first sparked my interest in this subject.
The effectiveness of the V1 campaign was undermined both by Allied countermeasures and the inherent limitations of the weapon. Many were shot down by Allied fighters or anti-aircraft guns. The V1 also proved inaccurate and numerous weapons failed at the point of launch or in flight. Consequently, far fewer missiles reached London than German plans originally envisaged. Out of a total of 8,081 missiles plotted by UK air defences between 13 June and 1 September 1944 (inclusive), 3,765 were intercepted and just 2,340 reached the Greater London area.
Needless to say, in addition to their defensive measures, the Allies also mounted a significant offensive air campaign against the V1 and its launch sites. Curiously, however, this is often presented as something of a failure. The Allied air forces inflicted such serious damage on the original launch sites - some 96 so-called ski sites - that they were abandoned; but the Germans switched to more tactical ‘modified’ sites that were extremely difficult to target from the air.
 |
| A 'modified' V1 launch site, difficult to spot and equally difficult to bomb; note also the square non-magnetic building sited parallel to the launch ramp. |
Hardly any damage was inflicted on the modified sites before the V1 campaign began, and the Allies instead maintained a substantial and costly effort targeting ski sites that the Germans had no intention of using. The failure of offensive air power is then typically implied by the very fact that the V1 campaign was mounted and sustained until the beginning of September 1944.
This is misleading. The cost involved in abandoning the ski sites in favour of the modified sites was significant. It was always clear to the Germans that it would be impossible to keep numerous launchers scattered all along the French channel coast supplied after the V1 campaign began. Their fixed facilities and especially the ski-shaped storage bunkers that gave the launch sites their nickname were designed to maintain a reserve of up to 20 missiles near the launch ramps as a basis for the continuous bombardment of England. There was no provision for any similar reserve at the modified sites. They were thus depended on constant replenishment via overland transportation from forward weapon storage depots. Both the communications links and the storage depots proved vulnerable to air interdiction, even if the modified launchers were very difficult to hit.
 |
| Pre-strike imagery for the V1 storage depot at St Leu D'Esserent, which was heavily and successfully targeted by the Allied air forces. |
 |
| A V1 train destroyed by the RAF. |
Beyond this, operations from modified launchers were hand-to-mouth, intermittent and often rushed. Resupply of the launch sites became increasingly difficult due to Allied air activity. As we have noted, the launch failure rate was high. National Socialism was not exactly known for its tolerance of human failure, so the launch crews invariably blamed misfires on defective components. Therefore, there is no precise record of the proportion of V1s that were misfired in haste due to the ever-present threat of air attack, due to the difficulties of launching missiles in darkness (to avoid air attack), because the launch site was covered in bomb craters, as many soon were, or because of even superficial damage to launch ramps or the non-magnetic buildings where the V1’s wings were fitted and its compass mechanism was set up. It is equally impossible to calculate how far launches carried out in such circumstances contributed to the V1’s inaccuracy. Nevertheless, there can be no doubt that the effect was significant.
 |
| The reality of operations from modified sites: a constant threat from the air, operating highly sensitive equipment in a lunar landscape of bomb craters, and multiple missfires (B, C, D). Note the crater at the end of the ramp (A). |
 |
| The modified site at Fressin; left, the launch ramp appears intact although there are craters immediately next to it; right, the non-magnetic building has clearly been damaged. |
 |
| While the servicing building was concealed in a barn at this modified site, the non-magnetic building stood in the open. |
 |
| While the launch ramp appears intact despite several near misses, the non- magnetic building has been destroyed, rendering the site non-operational. |
‘From the night of 16 June, the attack continued on an average scale, in round figures, of one hundred bombs every twenty-four hours. There were considerable daily and weekly variations in the weight of attack. The worst day was the twenty-four hours following dusk on 2 July, when 161 flying bombs came close to the coast or passed overland. On the lightest days, 13 and 15 July, the comparable figure was 42 in each case. The week ending 8 July, when 820 flying bombs were plotted, was the heaviest of the whole attack. The following week, however, the total fell by more than a third to 535.’
Over the whole period, the UK air defences recorded 2,934 V1 launches; over the next five weeks this figure fell slightly to 2,667. Finally, in the period from 16 August to 1 September, the defences reported 1,115 V1 launches. Substantial variations continued throughout, from highs of over 100 per day to lows of 14 or 15 per day. Some of the highest launch rates were achieved in periods of bad weather, when there was less risk of air attack, and lower rates often coincided with clearer conditions. However, the main reason for the fluctuating launch figures was the impossibility of maintaining missile supplies to forward locations at constant levels. The higher launch figures were achieved by depleting forward stocks, after which the bombardment invariably slackened while more flying bombs were laboriously deployed to the launch areas.
In summary, there was a yawning gulf between the original German V1 bombardment plans, based on the ski sites, and the campaign ultimately mounted from the modified sites in the summer of 1944. How, then, did the Allies initially identify the significance of the ski sites and the threat that they posed?
The answer lies in one of the Second World War’s greatest intelligence sagas. It is a story that enjoys an almost legendary status in the annals of intelligence history, yet it also has enduring long-term implications for intelligence practitioners. Recent years have witnessed considerable argument between those who contend that intelligence collection assets should be spread thinly to ensure that as many customers as possible obtain at least some support, and those who maintain that collection platforms and capabilities should be massed, so that individual assets can cue one another, and so that their products can be ‘fused’. The proponents of ‘mass, soak, layer and cross-cue’ argue that this approach produces far more detailed, informative and valuable intelligence than than the alternative of ‘equitable sharing’.
Of course, much depends on the difficulty of the collection task, and this is clearly illustrated by Allied efforts to unearth the secrets of Germany’s second vengeance weapon - the V2 rocket. In this instance, Human Intelligence (HUMINT) was enough to guide or ‘cue’ Imagery Intelligence (IMINT) to the German Army’s experimental establishment at Peenemünde, on the Baltic island of Usedom - a location where the V2 could be photographed, confirming that the Germans possessed an operational rocket.
The V2’s visibility from the air resulted from the elaborate and protracted launch processes associated with trial rockets in 1943 and from the high visibility of the German launch pads and launch equipment. Moreover, rockets were left in exposed locations for too long. The German Army was not sufficiently air-minded, and could easily have done more to conceal V2s near the launch pads by erecting quite simple roofed buildings. For all these reasons, after the Allies noted an upsurge of HUMINT on the V2 in the spring of 1943, they succeeded in acquiring imagery confirming its existence by 12 June - a remarkably rapid time-line.
 |
| A V2 rocket at Peenemünde, 12 June 1943. |
 |
| The much clearer photograph of a V2 taken on 23 June 1943. |
By contrast, the V1 collection task proved far more difficult. The Germans were testing the V1 from sites near the Luftwaffe’s experimental establishment at Peenemünde West from the first months of 1943. In July, Lehr und Erprobungskommando Wachtel formed under Oberst Max Wachtel to bring the V1 through service trials and generate the framework of an operational regiment, Flak Regiment 155, to launch it. Wachtel’s headquarters was at Zempin, along the coast to the south-east of the German Army’s experimental facility.
The growing volume of HUMINT on German secret weapons reaching the Allies in the first months of 1943 dealt largely with a weapon described as a rocket rather than a flying bomb, and there was no little confusion between the two. However, in August 1943, a HUMINT source reported that both a rocket and a flying bomb (or pilotless aircraft) were being developed, and the probability of a double threat was similarly suggested by Ultra in September. A second HUMINT source, the French agent Jennie Rousseau, subsequently produced the so-called Wachtel Report, which linked the development of secret weapons to Usedom, notably to Peenemünde and Zempin, where they were allegedly being trialled and where a launch regiment was being trained. The weapons were to be fired at England from 108 catapults deployed on the French coast.
At around the same time, the Allies began receiving intelligence on a number of new German construction sites in northern France, which included what proved to be the first ski sites. However, their purpose was initially unclear, and it was not until the end of October that the Allies received a detailed report on a ski site - Bois Carré, north-east of Abbeville. There followed an intensive air reconnaissance survey along the French channel coast to locate all similar facilities.
 |
| Oblique and vertical imagery of the Bois Carré ski site. |
Air reconnaissance was also intensified over Usedom, but IMINT could not repeat the rapid success achieved in spotting the V2. The V1 was a smaller weapon, and was easier to conceal; its launch apparatus was less obtrusive, and less time was expended in launch preparation, when the missile was exposed on the launch ramp. Moreover, the Luftwaffe were more air-minded than the German Army: their concealment doctrine was better. They built sheds near the launch ramps to reduce the missiles’ visibility; movement between the experimental site and the launch areas was never observed from the air and probably involved the use of covered wagons or else was confined to the hours of darkness. The Luftwaffe may have had radar-based warnings of approaching reconnaissance aircraft, which produced an extremely distinctive radar signature. The trial launch rate rarely exceeded two per day in 1943, so photo opportunities were fleeting. The Allies were looking for pilotless aircraft, so they naturally focused their attention around the airfield at Peenemünde West rather than the actual launch areas. They were also diverted by the counter-V2 task until the execution of Operation Hydra, Bomber Command’s attack on the German Army site at Peenemünde, on the night of 17/18 August.
For all these reasons, Allied air reconnaissance activity throughout the summer and autumn of 1943 yielded little sign of the V1. Launch ramps at the northern tip of Usedom were spotted, but (in an area of land reclamation) the Industry Section interpreters at the Central Interpretation Unit, RAF Medmenham, wrongly identified them as dredging equipment. PIs scrutinising the pair of parallel walls constructed at every ski site, orientated towards London, did not subsequently match them with similar structures on Usedom, possibly because they were incomplete. Their internal components had yet to be installed.
There were odd snippets of imagery showing possible pilotless aircraft near Peenemünde West airfield. For example, one small aircraft was spotted in an enclosure next to a building correctly thought by RAF interpreter Flight Officer Constance Babington Smith to house an engine test bench. ‘The midget aircraft had the aggravating cotton-wool look that all light-coloured or shiny objects acquire on aerial photographs, owing to the “light-spread” that blots out shadow and prevents detailed interpretation, and also makes things look deceptively larger than they are.’ The snub-nosed aircraft does not look like a V1 but might possibly have had its nose and warhead sections removed, shortening the front of the fuselage. The aircraft was christened the Peenemünde 20, as its wingspan measured about 20 feet.
Babington Smith subsequently discovered four more engine test bench sheds: the tell-tale exhaust deposits were clearly visible in the photographs. And there was something else. As she put it, ‘I checked the activity near them from cover to cover, and surprisingly, I thought, I did find one crumb of evidence to link them with the “Peenemünde 20”. On several dates there was an object resembling a midget airframe outside one of them.’ If this was correct, the object in question was the fuselage only; there is no sign in the photographs of any wings.
 |
| V1 launch ramps at the northern tip of Usedom, summer 1943, wrongly identified as dredging equipment. |
 |
| The Peenemünde 20, spotted in an enclosure outside a building later confirmed as the Argus Röhr test-bench. |
 |
| The four test-bench sheds subsequently spotted by Babington Smith; the object suspected of being a Peenemünde 20 is outside the building at the bottom of the photo. |
There was no further imagery of interest. This was because, having been guided to the general area of Peenemünde West, IMINT was being gathered in isolation, without the essential element of cuing via other collection media. Something further was required to generate a more informative and fused intelligence picture, and it was provided through an inspired exercise in lateral thinking by Dr RV Jones, Chief Scientific Adviser to the Air Ministry and MI6. It was Jones who realised that if the Germans were testing long-range weapons, they would need to track them, and that they would probably track them using radar. For this purpose, they were likely to draw on the services of a Luftwaffe formation entitled the Air Experimental Signals Regiment, particularly their 14th or 15th Companies. From earlier wartime experience, Jones knew that the Regiment’s Enigma-based signals traffic had been penetrated by Bletchley Park. ‘I therefore told [Professor Frederick] Norman at Bletchley of my line of thought, and asked him to see that Bletchley and the Y Service generally followed those two companies as closely as possible.’
Jones was quite right: Ultra soon disclosed that the 14th Company was being deployed along the Baltic coast east of Peenemünde. However, while the nature of their work was suspected, it could not at first be gauged with certainty. In fact, their tracking was a vital element in the development of the V1, and the information they collected was continuously fed back to the missile’s senior design and technical personnel - but not via the Enigma machine. The Air Ministry's intelligence narrative takes up the tale.
'The power of the planned enfilade through the German Air Force Signals Experimental Regiment lay in its being independent of the nature of the missile, providing that this missile was of long enough range to require radar plotting. In October 1943 the radar attack yielded its first result: one of the two signals companies which we were minutely watching started its own line of W/T traffic. Fortunately the Germans were indiscreet enough to use a low grade code which, though relatively uninformative in single messages, became very revealing under a complete watch. From the contents there could be no doubt that radar stations had been set up along the North German coast, and that these stations were plotting a pilotless aircraft known as the FZG 76. The plots from the radar stations were received by the Y-Service...'
The first problem for the Allies was then to locate the radar stations. This was achieved by timing the plots along their tracks; by extrapolating from one station to another, their relative positions could be calculated. As soon as one station had been firmly located, the position of the others could be fixed station by station, and it was then possible to build up a plotting framework. This wartime map was drawn to show sample tracks generated from the intercepted German radar plots and clearly demonstrated that Zempin and the northern tip of Usedom were the points of origin for the pilotless aircraft launches.
 |
And so it was that SIGINT effectively cued IMINT on to the V1. The key air reconnaissance mission was long delayed by adverse weather in November, and its parameters remain deeply mysterious. The 540 Squadron Mosquito involved, crewed by Squadron Leader John Merifield and Flying Officer William Whalley, was allegedly sent to Berlin on the 28th on a BDA task, following a heavy bombing attack the previous night. Finding the city covered by cloud, Merifield is then said to have flown north to the Baltic to conduct secondary tasking, which ultimately included the two launch locations.
This would not in itself have been an unusual sequence of events: it was common for air reconnaissance missions to be assigned multiple objectives. But Jones, in his 1943 report, stated that the Allies were attempting to photograph the two launch areas at Usedom at specific times to coincide with the typical German launch schedule. This would have been very difficult to achieve if they were secondary targets. The AHB narrator was likewise suspicious. As he put it, ‘Whether the securing of the photographs ... on this day was fortuitous must remain a matter for conjecture.’ Constance Babington Smith claimed that the sortie’s secondary task was to photograph radar equipment at Zinnowitz, near Zempin, and that the northern tip of Usedom was only imaged because the pilot had a small amount of remaining film to use up. Again, this hardly seems likely, for Jones had specifically requested imagery of both locations because they appeared to be the points of origin of the pilotless aircraft tracks.
All we can say for certain is that Merifield would have been equipped with some kind of cover story for his mission, in case he was shot down, captured and interrogated. In a mission guided by SIGINT, this was particularly important to preserve the security of the source. It seems possible that the cover story subsequently became confused with reality, and it is notable that the squadron record lists Peenemünde as an objective, although it also repeats the assertion that Berlin was the primary target.
Historical attention has chiefly focused on the imagery of northern Usedom in which Babington Smith spotted a V1 on one of the launch ramps formerly misidentified as dredging equipment. She claimed she had been encouraged by the test bench imagery to look further away from the airfield. However, Jones recorded in his report that it was the identification of launch ramps in photographs of Zempin that caused the ‘dredging equipment’ to be re-examined. In his words, ‘The Zempin photographs recalled to CIU certain unidentified structures at Peenemünde (near the rocket site) and on the new photographs the connection was confirmed because, as hoped, we caught a small aircraft on one of the Peenemünde ramps.’
 |
| The photograph that captured a V1 on one of the launch ramps at the northern tip of Usedom, 28 November 1943 - a truly brilliant spot by any standards. |
By contrast, the photos of Zempin have attracted far less interest and perhaps deserve closer scrutiny. Jones commented on the presence of ramps (‘catapults’) in the photos, and also buildings that resembled those under construction along the northern French coast, but another contemporary report declared that there were no actual ski buildings at Zempin, and this statement has been faithfully repeated by historians ever since.
There was certainly at least one square building - the all-important non-magnetic building constructed at every ski site - in the forest near a launch ramp, and a rectangular building resembling the longer rectangular launch preparation sheds at the ski sites. The RAF PIs would have checked the dimensions of these buildings against those of the equivalent ski site structures and would probably have found that they matched. However, close inspection of later imagery of the area reveals a second ramp at the edge of the forest, by the sea, a second square building, a second rectangular building, and a track leading round to a ski-shaped building.
 |
| Zempin: the ski-shaped building is marked A, ramps are B, square non-magnetic buildings, parallel with the ramps, are C, and the rectangular buildings are D. |
 |
| An enlargement of the launch area by the sea at Zempin. Note the scarring on the beach aligned with the ramp, which has been created by V1 misfires or the fall of the pistons used to propel missiles up the ramp. |
 |
| An oblique shot from the opposite direction; the aircraft was flying over the sea and filming south towards the beach; a second ramp was present by the time this photograph was taken. |
Jones was adamant that the Zempin photos played a vital part in establishing a direct link between the flying bombs and the ski sites in France, and this would have been a somewhat hopeful connection to draw purely on the basis of the rather nondescript buildings in the forest. He also referred to 'catapults' - plural - and even the later photographs only show two. So, is it possible that the beach-side launch area was also spotted in November 1943?
 |
| The arrow on the right shows scarring on the beach; the arrow on the left points to the ramp. |
The photograph taken on 28 November 1943 is reproduced above. A cursory inspection might suggest that the structures by the sea were entirely absent at this time. However, the imagery did not extend so far east as the later photos, it was of lower quality (perhaps because of light conditions), and there was more tree cover. The clearance of trees that had certainly occurred from the beach-side launch area by the spring of 1944 was also evident in later imagery of the forest launch facilities. The most compelling evidence for the presence of the beach-side launch ramp in November 1943 is scarring on the beach almost certainly created by V1 misfires or the fall of the pistons used to propel missiles up the ramp. Projection of these marks back into the woodland suggests the presence of a ramp at exactly the correct location. This would also imply that a non-magnetic building was present under tree cover. Yet we cannot know for certain that the Allies spotted a second ramp at this location in 1943, and it may be that the other buildings had yet to be constructed.
In the summer, the acquisition of imagery confirming the presence of V2 rockets at Peenemünde had led directly to the launch of Operation Hydra; but there was no Operation Hydra 2 after the V1 and its associated launch facilities were photographed in November. Target selection could well have been difficult: there was still very little information about precisely where the V1 was being developed at the very large Peenemünde West site. But it is likely that broader intelligence considerations were also important. After Hydra, the supply of intelligence on the V2 declined considerably as the Germans dispersed development and production and tightened security. A similar response to an attack on Peenemünde West might have denied the Allies the wealth of information on V1 testing being gleaned via SIGINT from the Air Experimental Signals Regiment and its component companies. There was far more to be gained from allowing the Germans to continue their trial launches unmolested.
So the Allies turned their attention to the ski sites. The ski site concept was developed before the RAF and the USAAF secured the almost unchallenged air superiority over northern France that they possessed by the end of 1943. It thus assumed a degree of air protection that the Luftwaffe could not provide by that time. The more obvious vulnerabilities of each site included the ski-shaped storage buildings, which were large and particularly distinctive from the air, and the two parallel walls built to accommodate the V1 launch ramps, which were normally sited in the open. The single square non-magnetic building at each site was also a key target because of its critical role in the V1 launch process. By contrast with some other German military structures, such as the Atlantic Wall and the larger V2 launch facilities in France, the ski sites did not make extensive use of reinforced concrete.
Nevertheless, the challenges that confronted the Allied Air Forces during these operations should not be underestimated. Ski sites were typically built on an area of 18 to 20 acres, only a small proportion of which was occupied by the essential constructions. Accurate attack was thus substantially dependent on clear visibility conditions, which were by no means always forthcoming over the winter months. Oboe provided the only effective blind bombing alternative.
The hazards involved in attacking V-weapons sites in northern France became more pronounced over time. Along the ski site belt between St Omer and Dieppe, there were about 60 heavy and 60 light anti-aircraft guns in December 1943. By the end of the following May, there were 520 heavy and 730 light guns. Anti-aircraft defences were also heavily reinforced along other parts of the coast, such as the Cherbourg Peninsula.
The potency of these defences imposed considerable tactical and planning constraints on the Allied Air Forces as the campaign progressed. Hence, although low-level strikes and dive bombing were successfully employed against the ski sites on a number of occasions, they proved more costly than medium-level attacks. The quest for medium-level options initially led to the use of AEAF medium bombers from Second Tactical Air Force and the US IX Air Force against the ski sites, but the strategic bombers of the VIII Air Force offered a more powerful capability and the scope to achieve more rapid results. However, the prospect of their diversion from strategic targets in Germany was inevitably controversial and generated a series of protracted high-level arguments. In the end, it was agreed that the Americans should employ some of their effort against V-weapons sites ‘whenever weather conditions over Germany do not allow major attacks there but permit precision bombing over northern France’. Operations proceeded on this basis, but the continuing priority assigned to the offensive against Germany not only limited the number of bombers available for counter-V1 strikes but also the number of escort fighters.
In summary, it would be wrong to suggest that the Allied bombing campaign against the ski sites was easily planned and executed. Yet the results achieved were, ultimately, enough. By the end of January, around one third of the sites had been neutralised, while others had been damaged. Two months later, 65 sites had been severely damaged and 20 others had sustained more moderate damage. The largely French workforce raised to construct the ski sites had scattered, and the Germans were having to use their own pioneer companies for bomb damage repair. Such was the context for their final decision to abandon the ski sites altogether.
Bois Carré Pre-Strike and Post-Strike Imagery
The story of how the Allies unravelled the secrets of the V1, from development and trials at Peenemünde West and Zempin through to launch provisions in northern France, clearly adds weight to the argument that the most challenging intelligence operations require collection assets to be massed and cross-cued. Individually, none of the collection capabilities would have sufficed. HUMINT provided contradictory information and little more than a general area for the Allies to search; it lacked the precision necessary to cue IMINT, which is why an abundance of imagery from across Usedom yielded virtually no intelligence of value over several months. High-grade SIGINT - Ultra - could confirm the presence of the 14th Company radar units along the Baltic coast, but it was lower-grade SIGINT that provided the vital plotting information revealing the V1 tracks. SIGINT in general lacked the degree of certainty that IMINT provided - the incontestable visual evidence critical to establishing a direct link between the development of pilotless aircraft at Peenemünde West and the ski sites in France.
Only by mobilising a combination of HUMINT, SIGINT and IMINT did the Allies succeed in breaking through the array of security barriers that protected the V1 project. In the absence of any one of the three collection media, they would probably have failed to achieve the intelligence coup completed by Squadron Leader Merifield's air reconnaissance mission on 28 November 1943, and they might not have spotted V1s being launched from Peenemünde West until the spring of 1944, when the number of trials markedly increased. The case for massing collection assets and for the fusion of their products to provide maximum support to operational priorities can rarely have been demonstrated more emphatically.
 |
| As the frequency of V1 trials increased in 1944, they were spotted more regularly in air reconnaissance imagery; two are visible in the area marked B in this photograph. |
