Antiaircraft guns were almost as old as aircraft but had always been woefully inaccurate, especially against high-flying bombers. In general, the best that gunners could hope to do was to throw so dense a wall of shells into the air that a plane was bound to run into one sooner or later. Blackett arrived at AntiAircraft Command in August just as the first radars were being delivered to antiaircraft artillery batteries around London. The idea of using radar data to calculate the proper bearing and elevation of the guns—the process known as “gun laying,” GL for short—was eminently sensible. The only problem, as Blackett quickly found, was that no one had bothered to figure out how to connect the gun-laying radars to the guns:
Immense scientific and technical brilliance had gone into the rapid design and manufacture of the GL [radar] sets … but unfortunately, partly through a shortage of scientific and technical personnel but also partly through a certain lack of imaginative insight into operational realities, hardly any detailed attention had been paid to how actually to use the GL data to direct the guns.… Thus the first months of the A.A. battle against the night bomber were fought with highly developed radar sets and guns, but with the crudest and most improvised links between them, belonging technically to the level of the First rather than the Second World War.35
General Pile was more than willing to give Blackett a free hand to see if he could improve the situation. Pile was not a natural intellectual; the eldest son of an Anglo-Irish baronet, he had spent his entire adult life in the army, winning the Military Cross and DSO as an artillery officer in France in the First World War. But he was always interested in new things. On the advice of the strategist and armored warfare theorist Colonel J. F. C. Fuller he had transferred to the Royal Tank Corps in 1923. “They like bright ideas there,” Fuller told him. What impressed Pile now about Blackett was that not only was he “a first-class scientist,” but he understood the realities of fighting a war and the need to make do rather than pursue some theoretical state of perfection. Pile wrote in his memoirs: “Blackett was an ideal man for the job. He spoke his mind clearly, and was always ready to admit the fact that the most desirable things sometimes may be inadvisable.”36
Blackett was able to quickly assemble a small team mainly by hiring some young scientists he already knew. The first to arrive were A. V. Hill’s son David, who like his father was a physiologist studying the mechanics of muscles, and David’s close friend Andrew F. Huxley—a future Nobel Prize winner in medicine and half brother to two other famous Huxleys, Julian and Aldous. Like Blackett, Huxley had what one member of the team called “an incredible mixture of theoretical and practical expertise.” He could take apart one of the large mechanical “predictors” that guided the guns “like a born mechanic.” A third physiologist, Leonard Bayliss, from University College, London, became Blackett’s deputy. There were also two physicists, Frank Nabarro from Bristol University and Ivor Evans (Evans’s name had come from the Central Register); an astronomer, Hugh Butler; an expert from the Admiralty on gunnery predictors, Arthur Porter; an army second lieutenant, G. W. Raybould, who had been a surveyor in civilian life and, as Blackett found when he happened to meet him when inspecting an antiaircraft battery in the Midlands, had already worked out his own method for converting radar data to gun bearings; and “a girl mathematician,” a Miss Keast.
The group’s official status was utterly vague. Blackett was called Scientific Advisor to the Commander-in-Chief, AntiAircraft Command, but his salary was being paid by the Ministry of Aircraft Production. As Bayliss recalled, the scientists just decided to call themselves the AntiAircraft Command Research Group: “Our position was established in two ways; firstly we informed the messengers to deliver to us all correspondence so addressed, secondly we had a rubber stamp cut with the letters AACRG. It is not known whether any more formal authorisation or recognition was ever obtained.”37 Unofficially, the group soon had another name that stuck, Blackett’s Circus.
Aiming an AA gun, even assuming the target was flying straight and level, required a complex calculation in spherical trigonometry. The plane’s height, course, and speed had to be determined and translated into an elevation angle and azimuth of the gun; at typical bomber speeds the gun needed to be aimed roughly a mile ahead of the plane’s position at the instant the gun was fired so that the shell would intercept the target as it continued forward on its course.
The Sperry Predictor was a half-ton mechanical computer designed to automate the task. A spotter would track the target, turning two dials to keep the airplane centered in a sight; the predictor, employing an array of gears, motors, and dials, would then transform the rate of change in those inputs into a continuously updated gun bearing. But the whole system was designed for visual spotting by day. There was no way to feed the radar data in directly. Blackett also found that the radar data at any given moment had a significant margin of error, so that even if the predictor could somehow be fed the data, the errors would cause the aim point of the gun to zig and zag with each updated position. Over a series of radar measurements, however, the errors tended to cancel out, and a fairly accurate track of the bomber’s course could be obtained. Blackett’s group first worked out a paper method of plotting the radar data and smoothing it out to obtain an average track that could be used to manually work out an elevation and bearing for the battery a given number of seconds later. Later they devised a modified Sperry Predictor—dubbed the “castrated” predictor—that allowed the operators to enter the radar tracking data by turning a dial originally intended to allow for wind corrections.38
About 120 guns were deployed around London, in batteries of 4 each spaced to cover the entire area of the city with overlapping fields of fire. There were only enough radar sets when the Blitz began in September 1940 to equip half of the 30 batteries. Blackett argued that there would be little to lose by grouping them into 15 eight-gun batteries instead, since the batteries without radar were extremely unlikely to hit anything and thus were already leaving gaping holes in the perimeter. A further calculation he made established that the whole idea of “complete coverage” had been an illusion anyway. Though the spacing between each of the 30 batteries in the original deployment scheme corresponded to the effective range of the guns under daytime conditions against slow-flying targets, the effective coverage of the guns using radar against modern fast bombers at night was far more limited: unless a bomber was coming almost directly toward the battery, its change in bearing took place so rapidly that the paper-and-pencil methods necessary for handling the radar data could not keep up. In fact, even the original 30-battery deployment was riddled with gaps between each battery’s effective field of fire. Concentrating the guns into 15 batteries hardly changed the size of the holes in the defensive screen, while greatly increasing the odds that any given gun would successfully engage a target by giving them all access to the radar data.
The scientists calculated that it was taking 20,000 shells to bring down one German bomber when they started their work. By the following summer, Blackett’s methods had cut the number of “rounds per bird,” as they termed it, to 4,000. One anomaly in the data which caused much head scratching was that AA batteries deployed along the coasts were bringing down enemy bombers with half the number of rounds per bird as inland sites. “All kinds of far-fetched hypotheses were considered as possible explanations of this strange result,” Blackett recalled. Perhaps the radar worked better over water; perhaps the enemy bombers flew straighter or lower while coming in across the sea than they did over land. “Then suddenly the true explanation flashed into mind.” It was simply that the inevitably exaggerated claims of planes shot down by each battery could be verified on land by locating the crash site, but not over sea. “This explanation should have been thought of at once,” Blackett observed, “as there is plenty of evidence to show that unchecked combat claims, made in absolute good faith, are generally much too high.”39 In the absence of Blackett’s explanation, disastrous changes in the deployment of the gun
s and radars might have been ordered on the mistaken belief that units ought to be concentrated on the coasts for greatest effectiveness.
Another bit of statistical reassurance that Blackett was able to provide was an even more elementary yet more crucial one. Pile, as Blackett would describe him in a postwar lecture to the Royal Statistical Society, was “an extremely intelligent general but he was a little flighty in his emotions and if one night he brought down six aircraft and the next night only two he would think something was wrong with the method and want to alter the methods and reprimand the crews.… I circulated a table of distributions round all the staff at the Command during the blitz showing that if they expected five aircraft there would be many times they would shoot down two [or] three.”
He encountered mixed results throughout the war trying to teach military officers such basic realities of probability and statistics. Another time, when he tried to explain that success in most operations is the sum of many attempts whose individual probability of success may be small, he was accosted a few days later by the pleased officer. “I say, Blackett,” the officer began, “I am so glad you explained to me all about probability. As soon as the war is over I am going straight to Monte Carlo and then I really will win.”40
THROUGHOUT THE FALL of 1940 the U-boat onslaught continued. On September 20, Kapitänleutnant Prien intercepted HX 72, a convoy of forty-one ships steaming from Halifax. BdU ordered in reinforcements, and over two nights the assembled wolf pack sank eleven ships in wave after wave of relentless torpedo attacks, many delivered at point-blank range. On October 17, SC 7, a convoy from Sydney, Cape Breton, consisting of thirty-five slow freighters barely able to make 7 knots, was beset by a pack of seven U-boats that proceeded to sink twenty of them, a shocking and demoralizing toll. “The danger of Great Britain being destroyed by a swift, overwhelming blow has for the time being very greatly receded,” Churchill wrote Roosevelt in early December. “In its place there is a long, gradually maturing danger, less sudden and less spectacular, but equally deadly. This mortal danger is the steady and increasing diminution of sea tonnage.” He warned that “the crunch of the whole war” was the Atlantic Ocean: “The decision for 1941 lies upon the seas.”41
In an attempt to breathe some life into the foundering anti-U-boat campaign, the War Cabinet decided to transfer operational control of RAF Coastal Command to the Admiralty. The new commander was Air Marshal Sir Philip Joubert de la Ferté, the assistant chief of the air staff, who was keenly aware of the success of radar in the Battle of Britain—and equally, the lack of any truly effective means yet for attacking U-boats from the air. He asked if Blackett could be transferred to his new command to tackle these urgent problems. In March 1941, with a salary of £1,000 a year that the Treasury deemed “appropriate to outside scientists of distinction employed in special posts,” Blackett became head of the new Operational Research Section, Coastal Command. “They have stolen my magician,” lamented Pile.42
The Real War
ONE OF THE FIRST SCIENTISTS Blackett brought in to work with him at Coastal Command was E. J. Williams, a thirty-eight-year-old Welshman whose fierce bushy eyebrows and short, stocky, powerful build made his colleagues agree that he looked more like a wrestler than a physicist.
Growing up in the village of Cwmsychpant in the heart of rural Wales, he had spoken Welsh at home and did not learn English until age five. His father, a master stonemason, “possessed in full degree the passion for education which is so marked a feature of the Welsh people,” Blackett would later write. On the wall of the Williams home hung a sampler embroidered by his mother with the words of a Welsh proverb, Gwell Dysg Golud, “Better Learning Than Riches.” One evening in 1919, shortly after graduating from high school, Williams saw a notice in the local newspaper announcing four open scholarships at Swansea Technical College. The examination was the next day and the last train was gone, but Williams talked his brother into driving him the fifty miles on his motorcycle. He won the scholarship. Around this time he announced to a friend that he intended to win an 1851 Exhibition fellowship just as Rutherford had, attain a doctorate in physics, and be elected a Fellow of the Royal Society while still in his thirties. All had subsequently happened. Williams spent 1933 in Copenhagen working with Niels Bohr. At Manchester and the Cavendish Laboratory he carried out a series of theoretical and experimental studies on cloud chamber tracks that documented the discrepancies between classical and quantum mechanics in atomic collisions.
Suffering, Blackett thought, from “crippling inhibitions” that prevented him from forming any truly close friendships, Williams tended to veer from shyness to argumentativeness; he had a “wildness and unexpectedness in his behavior” that would burst out in startling ways with those he got to know. He drove a car, Blackett said, “with a complete disregard of the laws of dynamics.” Another friend recalled that his sense of time “was practically non-existent.” Williams would not uncommonly work until three or four in the morning and then phone a friend excitedly, only to be overwhelmed with remorse upon discovering he had awakened his victim.1
Blackett had brought Williams to the Royal Aircraft Establishment in early 1940 to work on a magnetic detection device for submarines, an idea that had been discussed at the Tizard Committee a few months earlier. In a short time Williams built a prototype airborne sensor, and for the first test Blackett and Williams loaded it aboard an Anson patrol plane and had the pilot buzz the large metal airplane sheds at Farnborough at a few hundred feet to see if it could detect their magnetic gradient. The device worked in principle, and the advantage of magnetic detection of a submerged submarine was that magnetic fields were unaffected by water—unlike radio waves, which was why radar could not be used beneath the surface. The trouble was that magnetic gradients dropped off with distance to the fifth power, so for practical purposes a metallic object the size of a submarine could not be detected more than 200 or 300 feet away. (American scientists were at the same time developing a different approach to magnetic detection that would prove more feasible and lead to the device known as the magnetic anomaly detector, which would play a role in the fight against the U-boats later in the war.)
Under Blackett’s direction at Coastal Command, Williams now began work on what would subsequently become the single most cited example of operational research applied to war. When Blackett arrived at the command in March 1941 it was just beginning to receive aircraft and weapons that had a prayer of doing actual damage to a U-boat. At the start of the war the standard antisubmarine aircraft was the obsolete Anson, a forty-two-foot-long light transport plane whose only tangible contribution to the war effort would be as a trainer for bomber crews. The Anson could carry two 100-pound bombs and had a useful range from shore of roughly 250 miles at best. “Indeed the primary role of Coastal Command was reconnaissance and squadrons engaged in antisubmarine war were mainly intended to shadow and report the presence of U-boats to base and the Naval forces in the area,” a 1944 Coastal Command lecture acknowledged.2 The situation was so desperate at the start of the war that in December 1939 the command took to sending unarmed “scarecrow” patrols of open-cockpit Tiger Moth biplane trainers and Hornet Moth touring planes over the coasts for a few months in the hopes of forcing U-boats to waste time by submerging. Coastal Command had since replaced the Ansons with Hudson, Whitley, and Wellington twin-engine medium bombers and a small number of Sunderland flying boats, and was now, in 1941, receiving the first of the promised help from America of longer range Catalina flying boats and four-engine B-24 Liberators.
Frustrations with the ineffective 100-pound antisubmarine bombs had continued. Someone finally had the wit to consult British submariners about their own experiences on the receiving end of German air attacks. An August 18, 1940, memorandum noted that even near misses from bombs failed to do any serious damage; only in the case of a direct hit on a surfaced submarine could an attacker hope to inflict a lethal blow with an air-dropped bomb. “This itself is unlikely,” the memoran
dum noted, “as any well trained submarine will be at least at 30’ when the bomb is dropped”—having initiated a crash dive as soon as the approaching aircraft was spotted.
By contrast, a series of “disastrous” German air attacks on British submarines off Norway were almost certainly due to air-dropped depth charges, “of which our submarines have almost positive evidence.”3 Because of the dangers of the aircraft being hit by the blast from its own bomb, antisubmarine bombs had to be dropped from at least several hundred feet. There were no low-level bombsights, and accuracy was poor. A depth charge, by contrast, could be dropped from a plane just skimming the surface. That month Coastal Command acquired 700 of the navy’s Mark VII 450-pound depth charges, hastily adapted with fins and a rounded nose fairing for aircraft use.
Still, as the British aviation historian Alfred Price concluded, “During 1940 aircraft had caused no more than a mild harassment to enemy submarines.”4 Pressure was mounting to produce results. U-boats, mines, surface raiders, and the Luftwaffe’s four-engine Focke-Wulf 200 Condor bombers, based in France, together destroyed a half million tons of shipping in March 1941. The average number of U-boats at sea in the Atlantic and Mediterranean on any one day continued to mount steadily, reaching almost twenty in April and nearly double that by summer 1941. Since the start of the war, losses to Allied and neutral shipping had consistently outstripped the pace of new construction; the 40 million tons of shipping capacity available in 1939 had been whittled down to 35 million. “My thought had rested day and night upon this awe-striking problem,” Churchill later wrote. “This mortal danger to our lifelines gnawed my bowels.” On March 6 he issued a “Battle of the Atlantic Directive” ordering commanders “to take the offensive against the U-boat” and began convening weekly meetings of a cabinet-level Battle of the Atlantic Committee. The transfer of Coastal Command to Admiralty control was part of the shake-up, as was the relocation of Western Approaches Command from Plymouth to Liverpool, where it would be closer to the center of action in the war between the convoys and their attackers.
Blackett's War Page 19
