by John Keegan
The comparative ineffectuality of Dönitz’s U-boat campaign of September 1939–July 1940 was to be sharply reversed after the fall of France. In the immediate aftermath, the German navy hurried supplies of torpedoes and other submarine warfare material to the French Bay of Biscay ports—Lorient, Brest, La Pallice, Saint-Nazaire, Bordeaux—which were henceforth to be the bases for its U-boats during the Battle of the Atlantic; the first arrived in the Bay of Biscay, at Lorient, on 7 July. The Biscay ports provided Dönitz’s submarine fleet with direct access to Britain’s Atlantic trade routes, shortening by hundreds of miles those from Germany’s bases and sparing it attack on passage in the constricted waters of the North Sea.
As soon as the Biscay bases were acquired, Dönitz embarked on the realisation of his plan to defeat Britain, and its surviving allies, by destruction of its Atlantic convoys. Advantage seemed on his side. The number of U-boats, which had to survive only one outward passage from German shipyards to the French ports in order to become effective, was increasing. The number of British escorts, and of replacements of British merchantmen lost to attack, was increasing much more slowly. Dönitz’s belief in his ability to win the naval—and thereby the European—war, by destruction of the Atlantic shipping trade, seemed ready to be realised.
By a strange reversal, the First World War fears of the Admiralty, that it lacked the escorts necessary to protect convoys, seemed about to be confirmed in a subsequent war twenty years later. In the second half of 1940, the Royal Navy, wholly committed to the concept of convoy as it was, was attempting to protect much larger convoys than it had organised in 1917–18 with far fewer warships. In 1918 a typical oceanic convoy of 16–22 merchantmen was protected by seven destroyers, first-class warships of a speed (over 30 knots) double that of a U-boat on the surface, where U-boats usually attacked. In the winter of 1940, convoys of as many as thirty ships or more might be protected by only one inadequate escort.
An example was Convoy SC7 (convoys were identified by acronyms, usually denoting point of departure, and numbered consecutively; those most used were HX, originating in Halifax, Nova Scotia; later New York, OB, outbound from Britain; CU, Caribbean–United Kingdom; MK, Mediterranean–United Kingdom; SL, Sierra Leone; PQ, Britain–North Russia). SC7 originated in Sydney, Novia Scotia, and consisted of thirty-five ships, all slow, four of them inland freighters from the American Great Lakes. The only escort was the sloop Scarborough, built in 1930 with a top speed of 14 knots and so slower than a surfaced U-boat. On the fourth day out, 8 October 1940, the convoy ran into a gale and that night into U-boats. Over the course of the next ten days, SC7, though reinforced by two more sloops and two corvettes, and attended by a Sunderland flying-boat, lost seventeen ships. The horror of the experience scarcely bears thought. For the torpedoed seamen, even if they were able to launch lifeboats or floats, there was no hope of rescue. The convoy could not stop; the escorts’ duty was to stay with the merchantmen. The survivors of sunken ships drowned or died of exposure.12
SC7 was a ghastly example of the pre-1917 fears of the anti-convoy Admiralty, which thought there were not enough escorts to protect merchantmen, encountering the plans of a commander, Dönitz, who had contrived means to maximise the offensive power of what should have been an inferior weapon, the U-boat. The U-boat was the realisation of an ancient conception, the idea of the invisible weapon. Most of its early forms had been devised to undermine, literally, the power of the British surface fleet, as had the first practical submarine, invented by the Irish-American, J. P. Holland, in 1900. The Holland boat, however, had been intended, like all its ineffectual predecessors, to attack submerged. The genius of Dönitz—he was a sort of evil genius—was to perceive that the submersibility of the U-boat should be used merely to protect it from counterattack, once its presence was detected, and that in offence it should be used on the surface, where it could achieve speeds superior to most of its targets, the merchantmen, and not greatly inferior to those of all but first-class escorts.
The other ingredient of the Dönitz idea was that of the wolfpack (Rudel). His time as commander of a U-boat in the Great War had persuaded him that the deployment of single U-boats was wasteful. Better, he convinced himself in the war’s aftermath, to mass them in groups which could, first, detect convoys by forming a patrol line—similar to that organised by Nelson with his frigates—and then close for the kill. SC7 had had the misfortune to fall under attack by one of Dönitz’s earliest wolfpacks. It overwhelmed the escort. At one stage seven U-boats were operating against four escorts, since Scarborough had detached itself to hunt for one of the first predators, U-48, which it did not find; nor did it find its convoy again.
The other ingredient of the wolfpack method was central control from headquarters; after June 1940 from La Pallice. The medium of control was radio, just as it had been during von Spee’s cruiser campaign against British shipping in the Pacific and the Indian Ocean in 1914. Radio, as before, overcame the limitations of visual sighting, which so restricted Nelson’s ability to command the Mediterranean, because a single visual sighting, transmitted by radio, allowed La Pallice to concentrate a wolfpack against a convoy even if the U-boats constituting it had been scattered across several hundred miles of sea. Pack strategy, plus radio, was a deadly weapon against the convoy system.
All strategies, however, have weaknesses. Radio was the weakness of the pack system. Bletchley, supplied by intercepts from the listening stations, was provided with the material by which Dönitz controlled his U-boats. The difficulty was to break it. By late 1940, Bletchley had had no success against the German naval keys. Unlike those of the recently founded German air force (Luftwaffe), the Kriegsmarine’s operators came from a long-established signal service, which had strict procedures and severe schooling. Not only were German naval signallers trained not to make mistakes—for Bletchley the most fruitful source of breaks into the Luftwaffe traffic; the whole German naval signalling system operated on the belief that the enemy was listening. The Kriegsmarine therefore strove not only to keep enciphering secure but also to limit the amount of material transmitted, on the sound principle that the smaller the quantity of intercepts, the harder it is for an enemy to find a way into them.
Assurance of secure encipherment was attempted by two principal means: enlarging the number of rotors used in naval Enigma machines and designing certain keys to be used only by officers. Even before the war, naval Enigma operators were issued eight rotors from which to select three; from 1 February 1942 onwards Atlantic and Mediterranean U-boats used four rotors in an adapted machine.13 The “Officer” keys introduced were versions of the Heimisch key, the Süd key, and Triton, known as Shark at Bletchley, by far the most important since it was the key used in Atlantic U-boat operations from February 1942. Officer keys were regularly broken but usually with some delay.14
Limitation of material transmitted was achieved by the devising of “short” signals, a form of code which was enciphered within longer messages or used simply as answers to enquiries from U-boat headquarters at La Pallice (later Berlin). Most short signals, transmitted as “digraphs” (two-letter groups), referred to a chart of the Atlantic and adjoining waters, which was divided into an irregular grid. Bletchley, beginning with some captured material, managed to reconstruct some of the grid by April 1940. In May 1941, as a result of the celebrated capture of U-110, it reconstructed the grid of the whole North Atlantic and most of the Mediterranean. The Germans, who constantly reviewed the security of their signal system, became concerned in mid-1941 that the U-boat position transmissions might have been compromised and introduced a more complex short signal by relating positions at sea to fixed points of reference—Franz, Oscar, Herbert, etc.—arbitrarily chosen and changed at short intervals. When deciphered, a typical Enigma order to a U-boat now read: “If boat is in a fit condition for night attacks occupy as attacking area the northern waters of the 162-mile-squares [of the naval grid] whose central points lie 306 degrees 220 miles and 290 degrees 38
0 miles respectively from Point Franz. If boat not in a fit position, report by short-signal ‘No.’ “15
Bletchley managed to overcome the difficulty thus presented quite quickly, a vital matter since the position reports provided the data by which the Admiralty rerouted convoys, on passage, away from wolfpack patrol lines. Other short signals used by U-boats at sea were sighting and battle reports and announcements of expected dates of return to port. Most useful of all were the short weather reports, essential to Dönitz’s headquarters in positioning U-boats. Bad weather, paradoxically, was welcomed by convoy commodores and escort commanders, since it usually prevented U-boats from attacking. The short weather reports became a fruitful source of decrypts because, early in the Battle of the Atlantic, Bletchley found that they were rebroadcast by a meteorological shore station in a code that it could read; later, because the reports were made in three-letter groups, Bletchley discovered that U-boat operators were not using the fourth rotor on their Enigma machines, thus greatly simplifying the mathematics of decryption.16
BLETCHLEY AND THE BATTLE OF THE ATLANTIC
Winston Churchill confessed that he would much rather have faced the danger of a German invasion of Britain than had to sustain resistance to the U-boat war. The point is understandable. An invasion would have taken place within the dramatic unities of time, place and action. The U-boat war went on and on, always destructive and formless and apparently endless. As long as Dönitz could find boats and crews to send westward into the waters of the Atlantic, across which Britain’s necessities were convoyed, ships would be sunk, sailors drowned, cargoes lost, and the issue of the war suspended in precarious balance.
Yet, despite that perception, the Battle of the Atlantic, like any great battle, can be seen in retrospect to have had chronology and shape. It divides into five broad periods. From September 1939 until July 1940, the combat between Dönitz’s U-boat arm and the Royal Navy was not strictly a Battle of the Atlantic, since Germany’s lack of forward bases made passage into great waters difficult and largely confined the U-boats to the seas around Britain. There were rarely more than ten U-boats on station, often as few as four, and, though few were lost, only nineteen in the first ten months of the war, little shipping was sunk either. Dönitz’s dream of deploying 300 U-boats, to sink 100,000 tons of shipping a month (about twenty ships, given the current average size of ocean-going merchantmen), seemed a fantasy.
Then, following the fall of France and the Franco-German armistice of July 1940, Germany acquired occupation rights over French territory which included the ports of the Atlantic coast. Dönitz at once set up his U-boat command headquarters there, in the chateau of Kerneval near Lorient, and began to bring his flotillas out of the narrow waters of the Baltic and North seas to the Bay of Biscay. Sinkings at first rose but then fell again as the Royal Navy deployed more escorts and Dönitz was obliged to commit a huge proportion of newly built U-boats to training. Throughout the war, the Kriegsmarine never skimped on training, subjecting new boats and crews to as much as a year’s practice in the Baltic before allowing them to the “front.”
The third period of the Battle of the Atlantic began in April 1941, the month after Winston Churchill coined the phrase “Battle of the Atlantic.” Dönitz had by then assembled enough trained U-boat crews to begin organising patrol lines and wolfpacks in the central North Atlantic, though his boats had been driven into those deeper waters, away from the sea approaches to the British Isles, by the increasing numbers of Royal Navy escorts and Coastal Command aircraft. Sinkings rose but the Admiralty also had considerable success during the year in routing convoys away from U-boat patrol lines, thanks to Bletchley decrypts. In September 1941, for example, when thirty-two U-boats were on patrol in the North Atlantic, twelve sank no ships at all, and only four sank more than 10,000 tons, or two ships each.
Dönitz’s prospects changed suddenly with the inception of the fourth period in January 1942, when he was able to withdraw his U-boats from the central North Atlantic, break up his packs and patrol lines and deploy individual boats, under now often highly experienced skippers, against the coastwise shipping of the United States on its east coast and in the Caribbean. U-boat captains described the next six months as their “Happy Time.” Targets were numerous, so were sinkings. In January 26 U-boats operating in American waters sank 400,966 tons of shipping, 71 cargo ships or tankers, for no losses at all. February was worse proportionately: 18 U-boats sank 344,494 tons, 57 ships. In April, after a very bad March when 406,046 tons were sunk, 31 U-boats sank 133 ships of 641,053 gross tons; and so the awful summer went on. By the end of August, when the Americans at last instituted proper anti-submarine measures, 609 ships, of 3,122,456 gross tons, had been sunk, for the loss of 22 U-boats, out of 184 engaged.17
The extent of Dönitz’s success was due to the refusal of the United States Navy to institute convoy at the outset, in a bizarre repetition of the British Admiralty’s policy of 1914–16. Admiral Ernest King, Chief of Naval Operations, formed the view that weakly escorted convoys would merely provide more plentiful targets than individually sailed ships, and so left America’s coastwise traffic to its fate. In his defence, it is argued that he used such warships as he could muster in Atlantic waters to escort the convoys taking American troops to Britain, and that not one suffered loss; it was also the case that he was meanwhile waging a life-and-death struggle with the Japanese navy in the Pacific, which consumed almost every serviceable warship his navy possessed. Nevertheless, there was undoubtedly an American anti-convoy prejudice, as evidenced by the U.S. Navy’s organisation, as by the Admiralty in the First World War and again at the outbreak of the Second, of U-boat “hunting” groups which, as reason should have taught and experience did teach, found few if any U-boats to attack. By 1941 the Royal Navy was wholly committed to the correct view that, if U-boats were to be found and sunk, they had to be presented with targets to attack that could defend themselves, in short, convoys with strong close escorts.
The conclusion of the American Happy Time confronted Dönitz with the need to risk his U-boats against such targets again. The inception of the period that ensued, the fifth and climactic stage of the Battle of the Atlantic, from September 1942 until May 1943, ushered in a dreadful episode in sea warfare, marked by heavy losses of merchant ships and tragic loss of life, all suffered in North Atlantic weather at its worst. Nevertheless, the crisis can be seen in another light. It was the moment in the maritime conflict between the Kriegsmarine and its opponents—the Royal Navy, the Royal Canadian Navy, the United States Navy and their associated air forces—when Dönitz was forced, in the classical terms of sea warfare, to give battle. He had argued throughout his life as a professional naval officer that there was a victory waiting to be won between a surface fleet and its submarine enemy. At the end of 1942 he was challenged to win such a victory—and lost.
The part played by Bletchley Park in that victory, though crucial, was complex. Its struggle was two-sided, not unilateral, for the German navy had its own interception and decryption service—the Beobachtungs (Observation) dienst, known as the B-dienst—and it had a record of considerable success against the Royal Navy’s disguised transmissions. Because of Britain’s efficiency in breaking into Germany’s codes during the First World War, an undeniable complacency prevailed in the Naval Intelligence Department during the post-war years and well into the Second World War. The British believed that decryption was a one-way traffic even if, at first, they were unable to break Enigma. Because also they had unwisely trumpeted their decoding achievements of 1914–18, they had put the Germans on their mettle so that, long before the outbreak of war in 1939, the B-dienst was breaking the then current Admiralty code, a system of five-digit groups called the Naval Code, which was super-enciphered mathematically. Through carelessness, the more secure Naval Cypher was also betrayed to the enemy. It was a familiar story; a cipher officer used the super-encipherment book of the Naval Cypher to super-encipher messages sent in the Naval Code
. As the latter could be read, the former was quickly broken and was read currently and continuously until 20 August 1940.18
The English-language department of the B-dienst employed 900 people before the war; the number would rise to 5,000 by 1942. The B-dienst was located at German naval headquarters in Berlin and was led by Wilhelm Tranow, a radio technician who had first been employed to test the security of Enigma. The security of their ciphers was a matter of deep concern to the Germans throughout the war. It was constantly reviewed, as their own was by the British. Both navies remained convinced, nonetheless, that they could not be overheard, the Germans with far better reason. They correctly reasoned that, even were the British able to acquire three of the four elements of the Enigma system—the machine, the setting list, the indicators and the tables of bigrams which designated the grid-squares on the oceanic chart—they would still not be able to read messages. They discounted the possibility of the British acquiring all four and, as the result of a prolonged investigation during 1942, instituted new precautions against the operators’ resorting to short-cuts. More to the British disadvantage, they also altered the U-boats’ Enigma machines to accept a new, fourth rotor which, in combination with an adapted reflector, multiplied by twenty-six the number of possible keys.19
The result was that between 1 February 1942 and the following December, Bletchley lost its way into Enigma altogether, with a calamitous impact on sinkings. The effect was heightened by a sudden German breakthrough into British naval codes. Suspecting, correctly, that both Naval Cypher No. 2, the successor to the Naval Code, and the first Naval Cypher had been penetrated, as the latter had been since September 1941, the Admiralty had introduced in December 1941 the new Naval Cypher No. 3, still a super-enciphered code, not a cipher proper. It worked by the traditional method, the adding of numbers, from a book of number groups, to the groups indexed in the main code (“Cypher”) book and was decoded by subtraction. The books were issued to the Royal, Royal Canadian and United States Navies for use in the passage of convoys across the North Atlantic.