It would also give them a chance to be at the scene where the action was now shifting, with the withdrawal of the U-boats from the American coast, and to play a key part in managing the introduction of the new centimeter-wave radar. Shockley already knew Roberts, having visited Langley several times; both Morse and Shockley were keenly aware of the potential of the new radar in the war against the U-boats if all of the inevitable kinks could be ironed out in its introduction to operational use.
Their trip over was by a long and roundabout route. Boarding a Pan Am flying boat in New York, they flew first to Bermuda, then to the Azores, then to Lisbon, where they arrived in the early evening, the plane circling the harbor for half an hour waiting for the fishing boats to clear a path for them in the harbor. A little after midnight they continued on to Ireland, touching down at the seaplane port in Foyne on the west coast just after dawn, then transferring to a flight departing from nearby Shannon Airport at noon for Bristol. They arrived in London that night. It was Thursday, November 19, three full days after they had left New York. In the still blacked-out city they were driven to their appointments over the next few weeks by “cheerful, iron-nerved girls in uniform” who sped through the dark and often fog-obscured streets.
The following Monday the Americans met Blackett at the headquarters of the U.S. Eighth Air Force, which had been established at High Wycombe, about thirty miles northwest of London. Blackett was “even leaner and quieter than when I had known him in 1931,” Morse thought. Morse and Shockley spent the ensuing days steeped in conferences with Blackett, Williams, Gordon, Whitehead, Baughan, and other members of the British operational research groups at the Admiralty and Coastal Command headquarters and visiting an asdic training center in Scotland and a Coastal Command group in Plymouth. Morse did not see Tizard but he did have a session with Sir Stafford Cripps, the new minister for aircraft production who effectively headed the Anti-U-boat Warfare Committee, and proposed that an American representative be added to the committee.
Morse also took every opportunity to press for having some of the British civilian scientists come “to the States to give us help.” Blackett on more than one occasion mentioned his concerns about the allocation of the heavy bomber force in the antisubmarine fight. In the discussion at VIII Bomber Command, Blackett noted that the entire premise of the B-17 Flying Fortresses, which were the backbone of the growing American bombardment force in England, was that they would be able to carry out high-altitude, precision daylight attacks. But most of Europe was so frequently blanketed in low clouds that this was unlikely to be possible very often in practice. A better use for the planes, Blackett suggested, might be to carry out ASV radar attacks on submarines in the bay.60 Morse also noted that while virtually all of the American officers at the bomber commands were convinced that the best way to deal with the submarine menace was to bomb the German U-boat bases along the French coast, “Prof. Blackett, who is the only person I met having quantitative data concerning both operations, is of the opinion that it would be better to attack the submarine after it had left the base.” Blackett’s preliminary figures that he showed Morse suggested that every 100 hours of ASV patrolling in the bay destroyed three times as many U-boats as the same amount of time spent on missions to bomb the submarine pens. The main problem was that the heavy concrete roofs of the pens were proving virtually impervious to bombs. (“Concrete is quite cheap,” Morse noted.)61
On December 2, Morse had a long talk with Commander Winn of the Operational Intelligence Centre, and Winn suggested he might find it interesting to come and visit the submarine plotting room, if the required high-level permission could be obtained. A week later Morse was again at the Admiralty and mentioned it to Blackett, who, to Morse’s amazement, arranged the visit “in about three minutes.” Winn pointed out to Morse on the plot a convoy currently under attack; the Americans, he complained, were still reluctant to reroute convoys and had ignored his plea of a few days earlier to do so, which might have saved the convoy. “Possibly they do not yet believe the plot predictions!” Morse recorded in his diary.
Shockley had meanwhile left for a visit to St. Eval on November 29 and would later return there and end up staying on for an additional three months to help Roberts’s squadrons learn to use their radars and refine operational procedures. Riding along on air patrols over the coast of Virginia was one thing, but Shockley was now in a war zone and just how a civilian scientist might be treated if a military plane he was riding in went down and he was captured by the Germans had already raised some qualms. Admiral Furer of the U.S. Navy’s R&D department had noted that under international law, an enemy national civilian aboard a naval aircraft had the same status as on a ship at sea; he could be held as a prisoner of war—but not shot as a spy. Furer added, however, that the navy “has a mild doubt that the letter of International Law will be observed” by the Germans. The legal situation was murkier with respect to a civilian on an army plane. In any event, the RAF had adopted the precaution of issuing civilian personnel engaged on scientific missions that took them into a war zone a temporary “honorary commission” as an officer, which allowed them to wear a uniform and presumably be treated as a normal combatant if they fell into enemy hands.62
Morse sailed home on the Queen Elizabeth on December 21 (the available planes were all booked full of high-ranking officers flying home for Christmas). The luxury cruise liner had been pressed into service as a troop transport and Morse was assigned the bottom tier of a triple-decker bunk hastily assembled of two-by-fours “brutally nailed” into the beautiful wood paneling of what had been a first-class cabin.63
A couple of weeks after arriving back in Washington, Morse heard from Shockley of a jarring but revealing incident that had happened at St. Eval while Shockley was there. It was a sobering allegory of the critical importance of the work of the operational research groups in the deadly business of war. One of the Liberator crews had successfully sighted a surfaced U-boat, but when they tried to drop their depth bombs the shackles holding the bombs froze up and would not release. On their return to base they discovered that the shackles had rusted tight in the damp Cornwall air. On their next patrol, January 22, 1943, the crew ran into a pea soup fog; trying to find their way back to St. Eval over the trackless sea at zero visibility, the plane slammed into the Cornwall cliffs, killing all aboard.64 Morse was prompted to look at the statistics on flying hours per successful attack, and realized with a certain shock that even if the crew of Aircraft S of 2nd Squadron had not suffered that tragic end, their experience was in one sense completely typical: the crew of a sub-hunting aircraft on average encountered a single chance to attack a U-boat during all of its operational patrols. “There was no such thing as an experienced antisubmarine air crew,” Morse wrote:
Fighter pilots and ground troops usually fought several battles during their active life; if they were lucky enough during the first few “lessons,” they learned how to fight by fighting. But if ASW fliers didn’t do things right the first time, they usually had no other chance.… Therefore the only way to determine the best antisubmarine tactics was for a group such as ours to study all the reports of battles with U-boats, made by all the crews, and then to determine … what to do and what not to do.65
It was, Morse realized, the strongest argument yet for the importance of their work. It was also a disquieting reminder of all that was riding on it.
A Very Scientific Victory
AT THE FIRST MEETING of the Cabinet Anti-U-boat Warfare Committee back in November 1942, Cherwell was once again at the prime minister’s side and a thorn in Blackett’s. Cherwell as usual commandeered the technical discussion, peremptorily demanding a study on the effectiveness of air and surface escorts in protecting merchant ships from U-boat attack. The committee assigned Blackett the job of carrying out Cherwell’s bidding and was told to prepare “as soon as possible, an agreed statistical analysis” of the data. By “agreed,” Churchill made clear, he meant that Blackett was to submit hi
s numbers to Cherwell for approval before forwarding his findings to the full committee.
As irritating as Cherwell’s intervention was, Blackett conceded it was the right question to ask. Earlier in the year, Tizard had acknowledged that for all of his run-ins with Cherwell, the very fact that Churchill cared about science was not something to be taken for granted: “What previous prime minister of England ever had a scientific adviser continually at his elbow?” he observed to a parliamentary committee in February 1942. That Churchill listened so much to Cherwell infuriated Tizard, Blackett, and other scientists working on the war. But they were well aware, at least at some level, that it was the price to be paid for having a prime minister who would listen to them as well.1
Cherwell’s question prompted Blackett to reexamine a fundamental matter that had been largely overlooked, namely just how many escorts were needed for a convoy of a given size, and indeed what the optimal size for a convoy was in the first place. Blackett admitted he had dropped the ball:
Looking back, I think we operational research workers at the Admiralty made a bad mistake in not realising as soon as the group was formed in the spring of 1942 the vital importance of working out a theory of the best size for a convoy. However, it was not until late autumn that the problem became focused in our minds, largely through discussions that took place at the Prime Minister’s fortnightly U-boat meetings. The problem arose as to what was the best division of our limited shipbuilding resources between merchant ships and the anti-U-boat escort vessels. Every merchant vessel completed brought the United Kingdom additional much needed goods; every escort vessel completed added to the protection of the convoys and so reduced their losses by U-boat attacks and so saved more ships and cargoes. To make a quantitative comparison of the relative advantage of building escort vessels or merchant ships, one needed to know how many merchant vessels would be saved, that is not sunk, by each extra escort vessel protecting the convoys.2
Reviewing statistics for 1941 and 1942, Blackett’s team concluded that each additional escort vessel saved approximately two to three merchant vessels a year. That meant that, all things being equal, building one new destroyer a year was worth the same as building two or three new merchant ships a year. But, as Blackett also discovered, that was a finding of more theoretical than practical value, as shipyards could not easily change from producing one type of vessel to another: they were already pretty much committed to the building programs already under way.
The point about convoy size, though, leapt out from the analysis. It also plunged Blackett at once into yet another controversy with the military establishment. The Admiralty had long maintained that smaller convoys were preferable. Forty ships was considered the best size; convoys of more than 60 ships were flatly prohibited. There was no dispute that large convoys were harder to handle and manage and tended to overload port facilities. The navy, however, went further, insisting that small convoys also afforded the best protection.
The statistics disagreed. Blackett’s group, in an analysis completed January 27, 1943, found that each ship in a small convoy of 15 to 24 ships had a 2.3 percent chance of being sunk when the convoy was attacked by a U-boat, versus only 1.1 percent in a large convoy of 45 or more ships.3 The British navy had an obscure rule for establishing how many escort vessels were required for convoys of varying size. It probably went back to the First World War; Blackett was never able to track down exactly where it had come from. It gave the number of escorts as 3 + N/10, where N was the number of merchant ships in the convoy. Blackett pointed out that the rule even on its face contradicted the navy’s assertion that small convoys were safer. By this formula a 20-ship convoy required 5 escorts, while a 60-ship convoy required 9 to provide a commensurate level of protection. Yet if that were the case, why not combine three 20-ship convoys and allocate all 15 of their escorts to the resulting single 60-ship convoy, thereby almost doubling the number of escorts over that specified by the Admiralty’s formula?4
On February 5, Blackett presented his preliminary report on the value of escorts and aircraft to the Anti-U-boat Committee. (A cover note stated, “The paper has been submitted and discussed in detail with Lord Cherwell during the various stages of preparation. The responsibility for facts and figures rests however with Professor Blackett.”) The paper reaffirmed the value of air cover, especially VLR aircraft; even when aircraft were unable to carry out a successful attack, their mere presence forced the U-boats to submerge, and then often lose contact with the convoy. An average of even 8 hours of flying a day by VLR aircraft in support of a threatened convoy reduced losses by 64 percent. In its average service life of 40 sorties, each VLR aircraft saved 13 ships that would otherwise be torpedoed. But a key recommendation of Blackett’s was to increase at once the size of convoys. In fact, he pointed out, the number of ships lost in a 60-ship convoy was on average almost exactly the same as the number lost in a 20-ship convoy.5
Several more weeks of analysis and discussion among Blackett’s group produced some further insights into what was going on. A convoy’s size made almost no difference in the chances that the convoy would be sighted and attacked by a U-boat. It also made almost no difference in the absolute number of ships that would be sunk once a U-boat had penetrated the destroyer screen around the convoy, since there were always more than enough targets and each U-boat had a limited number of torpedoes. So there was safety in numbers simply because one ship was less likely to be picked out from the crowd. Looking at the actual number of ships torpedoed per convoy sailed (including stragglers who failed to keep up with the escorts), the researchers found that it worked out to almost exactly 0.9 ship per convoy regardless of convoy size.6
The mathematician J. H. C. Whitehead shortly afterward produced a paper explaining further why a large convoy did not need as many escorts per ship to afford the same level of protection. Even the 3 + N/10 formula underestimated the benefits of combining small convoys. The chances of a U-boat penetrating the escort screen, Whitehead noted, was basically a function of the linear spacing of the escorts around the convoy’s perimeter. But the length of the perimeter increased very slowly with increasing size of the convoy. It was a familiar geometric principle that occurred frequently in biology: the area within a perimeter increases with the square of the perimeter’s length. To maintain the same linear spacing around the perimeter, the number of destroyers required increased as the square root of the number of ships they were assigned to guard. A 78-ship convoy had a perimeter only one sixth longer than that of a 40-ship convoy; 7 escorts on the larger convoy provided the same level of protection as 6 escorts on the smaller one. Thus by far the most efficient use of the available surface escorts was to employ them in the largest possible convoys practical. Increasing the average size of convoy even by 50 percent, Blackett calculated, would reduce the total number of merchant ship sinkings by half.7
Blackett was well aware what a serious matter he was taking on his shoulders. He and his co-workers “had proved intellectually to themselves that big convoys were safer than small ones,” as Blackett later said, but “before we advised the Navy to make this major change, we had to decide whether we really believed in our own analysis.” Blackett did so for himself by deciding “that if I were to send my children across the Atlantic at the height of the U-boat attacks I would have sent them in a big rather than a small convoy.”8
But at the Anti-U-boat Committee the Admiralty was unimpressed, and dug in its heels. At the February 24 meeting, Lord Leathers, the minister of war transport, urged that the artificial limit of 60 ships be waived as the only way to relieve the increasing backlog of supplies awaiting shipment across the Atlantic. The first sea lord, Admiral of the Fleet Sir Dudley Pound, replied that the “sailing of such large convoys was asking for trouble” and reiterated that 40 ships was the optimal size.
Blackett tried again a week later, pointing out that even if losses increased with larger convoys, total imports would increase as well given that significant number
s of ships were now being delayed by the 60-ship rule:
More ships are now presenting themselves for convoy than can be accepted within the 60 ship maximum rule. The reason for the imposition of this limit is the expectation of heavy losses with larger convoys. The object of this note is to point out that it is inconceivable that the increase of losses could be large enough to offset the immediate gains in imports by adopting larger convoys.9
But, he pointed out, the assumption that losses would increase at all was “unduly pessimistic”; he was just presenting a worst-case possibility, and the evidence was that losses would likely not increase even in absolute numbers. Blackett advanced all of these arguments yet again in an unusually forceful paper produced later that spring (forthrightly titled “The Case for Large Convoys”). In it he also noted that a related advantage of fewer sailings with larger convoys was that the available long-range and VLR aircraft would be able to cover a greater percentage of ships during their perilous passage through the still inadequately protected “air gap.” The calculation was rather complicated, but conferring with the Coastal Command ORS he predicted that air cover would increase 40 percent.
In the end it was the press of import requirements that forced the Admiralty to back down. Blackett’s arguments helped offer some assurance that the change would at least not be a disaster, even if he had failed to convince Pound that it would be a virtue in itself. By late spring and early summer 1943, 80-ship convoys were common. As Blackett had predicted, the large convoys were far more economical in their use of surface escorts. The change allowed a large number of destroyers to be freed up the following year to support the D-Day landings. Blackett later calculated that had he thought to look into the problem in spring 1942 instead of a year later, a million tons of shipping, 20 percent of the total losses, could have been saved in the interim through the adoption of large convoys. But in the rush of other work no one in the Admiralty operational research section had thought of it.
Blackett's War Page 30
