by John Keegan
During January 1942 the B-dienst succeeded in reconstructing both the codebook for Naval Cypher No. 3 and the subtraction tables used with it. In consequence, it was able to read 80 per cent of the convoy traffic, often twenty to thirty hours in advance of the movements signalled. This warning allowed ample time for U-boats to be positioned in a convoy’s path, since the difference between the speed of an average convoy, seven to eight knots, and that of a surfaced U-boat, at least sixteen knots, meant that patrol lines and wolfpacks moved at twice the speed of their prey. In twenty-four hours, during which a convoy advanced 180 miles, a hunting U-boat could move 360 miles to cut it off, submerging only just before the moment of contact.
U-boat commanders were trained, according to Dönitz’s pre-war experiments with torpedo boats, to linger at the limit of visibility, until the fall of dark, on the convoy’s projected line of advance, then to surface, if possible within the convoy columns, fire salvoes from bow and stern tubes simultaneously, and to submerge and make their escape as the escorts appeared. Depending upon how much disruption had been caused in the convoy pattern, a second attack might be mounted.
The difficulty for the U-boats was to locate a convoy in the first place. The limit of visibility at best from a conning tower was ten miles; a patrol line of ten U-boats could therefore cover 220 miles of sea. Dönitz attempted to extend the area of sweep by securing the services of 1/KG40, a squadron of long-range Condor aircraft, from the Luftwaffe; but the growing efficiency of British air patrols in 1941–42, which had driven the U-boats into the central Atlantic, also put them beyond the Condors’ range; an effort to fly an autogiro on a cable from U-boat conning towers proved as impractical as it was dangerous. A convoy of fifty ships, meanwhile, occupied a front of only 2,400 yards. In the enormous spaces of the Atlantic—at least nine million square miles of operational waters—the area occupied by a convoy and the space covered by a questing U-boat patrol line were both relatively tiny. The one could be missed by the other with the greatest ease and usually was. Between 1 January and 31 May 1943, for example, the height of the Battle of the Atlantic, forty-eight out of eighty-six convoys sailed were not found by U-boats at all.20
Bad weather played a part, hiding the convoys from German eyes or forcing U-boats to seek shelter from the elements below periscope depth; so did routine alterations of course and emergency turns if contact were made. Deliberate rerouting of convoys, however, away from patrol lines and wolfpacks, located by Bletchley, was the most productive method. Indeed, it was Hut 8’s main task to provide such intelligence, which was Bletchley’s principal contribution to winning the Battle of the Atlantic and so, arguably, to assuring that the war would not be lost.
David Kahn, the great historian of cryptography, gives a dramatic account of one such rerouting contest in his book on the U-boat war, Seizing the Enigma. It describes the progress of Convoy SC127—so code-named for the point of departure from Sydney, Cape Breton Island at the mouth of the St. Lawrence River in Canada—towards Liverpool in England in April 1943. The direct distance, measured from Halifax, Nova Scotia, was about 4,000 miles. The planned course, including changes of direction, was longer. The course actually sailed, around identified U-boat traps, was longer still.
SC127 comprised more than fifty ships, arranged in thirteen columns, escorted by five Canadian warships. It first steered east, at about 7.5 knots, then turned slightly northeast towards a spot in the ocean designated Point F by the Admiralty’s Trade Movements Section. On 16 April, when it departed, both Bletchley and OP-20-G in Washington were decrypting Dönitz’s radioed instructions to his U-boats, and their encrypted reports, at a delay of three days. Both tracking organisations knew, however, that he had over sixty boats in the Atlantic (in fact sixty-three) and that twenty-five were located on SC127’s path due east of the Gulf of St. Lawrence. They formed a patrol line 650 miles long, aligned north-west–south-east, due south of Cape Farewell, the southernmost point of Greenland.
On 18 April Allied cryptanalysis—at either Bletchley or Washington or both—broke Dönitz’s transmission of the 17th, ordering the formation of the line code-named Titmouse. Though it referred to the U-boats involved only by their captains’ names, a new security measure, the key was one that had been solved; less clear were the boats’ locations, since neither Bletchley nor OP-20-G had comprehensively established the inner encryptions of the grid-square designations. Moreover, the B-dienst was reading correctly Naval Cypher No. 3 and so knew, from Admiralty transmissions to SC127, that it was aware of the existence of Titmouse.
Dönitz nevertheless, with some complacency, decided that SC127 would maintain its current course; he may have done so because he was also tracking the progress of the convoy following SC127, HX234, which had just made a sharply evasive change of course, and perhaps concluded that the British would not order two convoys simultaneously to divert from their planned line of advance, trusting to HX234’s alteration to distract Titmouse from SC127. In this he was wrong. On 20 April the Convoy and Routing Section in Washington, the U.S. Navy’s equivalent of the Admiralty Trade Division, ordered SC127 to make a radical change of direction, just before it reached the designated Point F due east of Cape Race in Newfoundland. Instead of continuing north-eastward it was to turn almost due north, leaving the coast of Newfoundland to port, and so evading by several hundred miles the grid squares in which the twenty-six U-boats of Titmouse were lying in wait for it.
Alerted by Titmouse’s failure to find SC127, Dönitz now formed another patrol line, code-named Woodpecker (bird names were currently in fashion at U-boat headquarters), and deployed it to the south, wrongly guessing that SC127 had gone in that direction. SC127 therefore proceeded untouched on its leisurely way. Two factors intervened to spare it. One, unfortunately, was that Titmouse had found HX234 and was beginning to sink ships; the sinkings would be sustained by yet another hastily formed wolfpack, Blackbird. The other was that because of SC127’s northward diversion, in one of the worst winters of the twentieth century which brought ice as far south as Newfoundland, it began to run into bergs and floes that forced it to change course yet again, into the Denmark Strait between Greenland and Iceland. The danger from ice also caused the convoy to slow down, thus throwing out German presumptions about its rate of progress towards other U-boat concentrations further east in the central Atlantic.
When on 22 April the convoy resumed the changed course ordered by Convoy and Routing in Washington, the allied cryptanalysts were again reading Dönitz’s Enigma transmissions, but now at only one or two days’ delay. As a result SC127 was redirected due east, thus sending it outside and to the north of the U-boat trap lying in the middle of the Denmark Strait. On 25 April the convoy passed from the control of the U.S. Commander-in-Chief Fleet (Cominch) to Western Approaches in Liverpool, the Admiralty command fighting the Battle of the Atlantic. Long-range aircraft, based in Iceland, had also appeared on 26 April to fly cover, an intervention that would have forced any U-boat surfacing (none did) to submerge and so lose contact with the convoy. On the last three days of its voyage, 29 April to 1 May, SC127 was continuously escorted by aircraft. On 2 May, having meanwhile detached twenty-two of its ships to Iceland and Scotland, it made port in Liverpool, having lost none of its number in seventeen fraught days at sea.21
SC127’s escape from attack by the wolfpacks was very largely due to Bletchley’s and OP-20-G’s ability to overhear Dönitz’s instructions to his U-boats. They did not yet enjoy complete dominance of the airwaves, however, nor had the Admiralty wholly overcome the B-dienst’s ability to decrypt Anglo-American convoy traffic. Not until June 1943 would changes to Naval Cypher No. 3 absolutely assure the security of the signals used to fight the Battle of the Atlantic and Bletchley—and OP-20-G—achieve mastery of naval Enigma. Nevertheless, the steering of SC127 through the traps laid for it by U-boat headquarters demonstrated that the corner had been turned.
In mid-March 1943, two convoys leaving New York, SC112 and HX229, had falle
n into the grip of two wolfpacks, “Robber Baron” and “Attacker-Driver,” and been massacred. Out of the two convoys’ ninety merchantmen, escorted by twenty warships, twenty-two were sunk, for the loss of only two U-boats. One was destroyed while returning to base far from the scene of action, by a British anti-submarine aircraft in the Bay of Biscay.
The battle around SC112 and HX229—the two became intermixed, as SC127 and HX234 almost did—was the costliest of the war and later reckoned to mark the crisis of the Atlantic battle itself. Any more battles like SC112/HX229 and Dönitz would have won. As it was, the others that followed tilted against him. Later in March the two convoys SC123/HX230, escorted by one of the new small “jeep” carriers, outfaced two wolfpacks, “Seawolf” and “Seadevil,” and got home to Britain with the loss of only one merchantman; HX231 and ONS176 did not fare as well, but there was no massacre. In April, 313,000 tons of shipping was lost, but so were fourteen U-boats. In May, when Dönitz sent three wolfpacks, “Woodpecker,” “Blackbird” and “Ram,” comprising sixty U-boats, against ONS5, sustaining contact for ten days, twelve merchantmen were sunk, but so were eleven U-boats, seven in a single night. At the end of May, in which thirty-four U-boats were sunk, he decided that the exchange rate could no longer be borne. “In May in the Atlantic the sinking of about 10,000 tons [approximately two ships] had to be paid for by the loss of a boat, while not long ago [he meant during the most profitable period in 1942] a loss came only with the sinking of 100,000 tons. Thus losses in May have reached an intolerable level.” He accordingly ordered “a temporary shift to areas less endangered by aircraft,” by which he meant away from those flown off the escort carriers, from Britain over the Bay of Biscay and from Iceland, Ireland and North America into the former “air gap” in the mid-Atlantic. It was an admission of defeat, effectively total defeat, for, though new weapons and new U-boat technology would allow sinkings to continue, they would never again approach the level of 1942 to early 1943, let alone his projected “war-winning” total of nine million tons of merchant shipping a year. Dönitz had lost.
What part had the cipher battle played in the defeat of the U-boats? It was greatly to the Germans’ disadvantage that from the middle of 1943 onwards they lost their ability to read British transmissions. During its best days, in early 1943, the B-dienst was able to give Dönitz advance warning by as much as thirty hours of convoy movements and was breaking up to 10 per cent of the convoy traffic in real time. After an unjustifiably laborious and time-consuming examination of the security of Naval Cypher No. 3, the design of an improved super-encipherment, Naval Cypher No. 5, and its distribution to the British and American fleets, the B-dienst found itself shut out. Nevertheless, until then, its contribution had been formidable. Dönitz, lacking the access to long-range air patrol enjoyed by the British and Americans, never could have achieved the number of convoy interceptions he did by the formation of U-boat patrol lines alone. The ocean was too large, the number of U-boats too few. They needed direction onto the convoy tracks and that, from 1941 to 1943, was provided by the B-dienst.
Even so, most convoys crossed the ocean, east–west or west–east, without interference, even when the Battle of the Atlantic was at its worst. In the five wartime months of 1939, 700 ships were sailed in convoy across the North Atlantic to the British Isles, and only five were sunk; sinkings in inbound convoys were far more damaging to Britain’s war effort than in outward convoys for, though the ship was lost in both cases, so in the first case also was the cargo; outward-bound ships were often sailing empty in ballast, at worst carrying export goods to pay for purchases. In 1940, 5,434 ships arrived and 133 were sunk. In 1941, 12,057 ships arrived and 153 were lost. The total of sinkings in convoy, even the weakly escorted convoys of those first three war years, totalled 291, or .02 per cent of those sailed.
In 1943, the year of the biggest convoy battles, when Dönitz had more than 300 U-boats in commission and was organising wolfpacks of as many as 40 boats, 9,097 ships arrived across the North Atlantic and 139 were lost; in 1944, 12,007 arrived and 11 were lost; and in the five war months of 1945, 5,857 arrived and six were lost. Out of a total of 838 convoys and 35,449 ships sailed in 1943–45, 325 ships were lost altogether, a percentage loss of .009.22
These figures need adjustment. They do not include sinkings in outward-bound convoys, in convoys sailed elsewhere than the North Atlantic or of ships not in convoy, which were numerous; nor do they include sinkings by aircraft, surface warships or mines. They make no allowance for the fact that many ships crossed the Atlantic several times, exposed to risk on each occasion, as were their crews. They are a record only of transits by ships loaded with war material, including essential foodstuffs, fuel and raw materials, on the crucial route, the North Atlantic lifeline between North America and Britain—in short, on the battleground of the Battle of the Atlantic, the course of which so preoccupied and tormented Churchill during the dark years of the war.
Clay Blair, the most meticulous historian of the U-boat war, has concluded, by analysis of the statistics, that Churchill’s fears were exaggerated and that Dönitz, even when he acquired his 300 boats, never threatened Britain with starvation, as the country undoubtedly was so threatened in 1917, nor even came close to inflicting defeat on the Allied anti-submarine forces. Indeed, Blair actively argues that reliance on evasive rerouting of convoys away from U-boat patrol lines, by intelligence derived from Bletchley and later OP-20-G decipherment, positively reduced the rate of U-boat destruction by lessening the frequency of contact between the wolfpacks and the convoy escorts and so effectively prolonged the Atlantic battle. By the end of 1941, he writes, “it was clear . . . that the British could not rely so completely on convoy ‘evasion’ much longer. In addition to strengthening defensive convoy escort, they needed to hurl offensive air and submarine forces at U-boat construction yards, training areas, bases and pens, the Bay of Biscay and elsewhere to kill U-boats at a much higher rate.”23
By the middle of 1943, a variety of measures adopted by the British, Americans and Canadians had greatly increased the rate of U-boat destruction both in the vicinity of convoys and on the oceanic approaches. In May 1943, of the forty-nine U-boats which sailed on patrol to the North Atlantic convoy routes, eighteen were lost, a destruction rate of over one in three. Many were commanded by men sailing as captains for the first time; eleven, however, had experienced skippers. Between them, the forty-nine boats sank only two merchantmen.24
What factors, as opposed to intelligence derived from decryption of Enigma, so set back Dönitz’s dream of victory through convoy destruction? They were many, ranging from a successful effort to increase the size of the merchant fleet supplying Britain, by charter and requisition, to such technological advances as the installation in escorts of high-frequency direction-finding (HF/DF or Huff-Duff) and the adoption of centrimetric radar, but also including improvements in underwater weapons, additions to the number and improvements in the quality and organisation of escorts, the introduction of escort (“jeep”) carriers, the commitment of more anti-submarine aircraft, with improved equipment, to the routes leading from U-boat bases to the Atlantic, as well as of Very Long Range (VLR) aircraft to the middle Atlantic “air gap,” improvements in underwater detection methods and many other measures.
ENLARGEMENT OF THE MERCHANT FLEET
Britain began the war with a merchant fleet—soon to be christened by Winston Churchill, in one of his flashes of linguistic inspiration, the Merchant Navy—of 3,000 ships, totalling 17.5 million gross registered tons. By the end of 1941, when the United States (with 1,400 ships of 8.5 million gross tons) entered the war, the Merchant Navy had lost 1,124 ships, including neutrals sailing with British war cargoes, totalling 5.3 million gross tons. However, 483 ships had been acquired from German-occupied Norway, Greece and Holland, and another 137 requisitioned as prizes from the enemy, totalling four million gross tons. British yards meanwhile built ships of about two million gross tons, with the result that between
September 1939 and December 1941, the month of Pearl Harbor, the Merchant Navy effectively increased in size to 3,600 ships totalling 20.7 million gross tons. In retrospect it seems astonishing that owners, to say nothing of their employed crews, were prepared to acquiesce in the chartering of their ships into the Battle of the Atlantic. Their readiness to take the risk, commercially and personally, can be explained only by what Clay Blair has demonstrated to be the relatively low rate of loss.25
In what may be characterised crudely as a personal struggle between Churchill and Dönitz, Churchill can thus be seen to have been outbuilding and outchartering Dönitz in the period September 1939 to December 1941. Measured in terms of tonnage sunk, Dönitz’s preferred index of success, Britain was keeping ahead. After 7 December 1941, when the might of United States industry was thrown unequivocally into the balance, the Anglo-American alliance drew unchallengeably away. At the forefront of the tonnage struggle was the dynamic American industrialist Henry Kaiser, a civil engineer who had built the Hoover Dam and other major works during Roosevelt’s programme of reconstruction after 1932. Requested to devote his time-and-motion techniques to ship construction, Kaiser designed a standardised merchant ship, the Liberty ship, modelled on a British general-purpose freighter, and built shipyards on the west and east coasts that could turn it out, partly by prefabrication, in as little as four days. The average was forty-two days and, by the Kaiser method, 2,710 were built during the war. The number included the superior Victory ship, which could do eighteen as against eleven knots. Kaiser also built the T2 and superior T3 tankers and numbers of escort carriers. When the output of the Kaiser shipyards is set against U-boat sinkings in the North Atlantic, 1,006 ships in 1942 but only 31 in 1944, it becomes clear that Dönitz’s hope of winning a tonnage war was quite misplaced. As had not been the case in the First World War, Germany’s enemies could outbuild the U-boat’s ability to sink. However hard Dönitz drove his captains, statistics, by mid-1943, had turned decisively against them.26