Authors: Michael Gannon
The confidence and authority of the Tracking Room advanced dramatically. By Winn’s order, a large photograph of Admiral Dönitz looked down upon what Winn fancied was a shadow BdU.
5
Now, as he and Beesly observed the proceedings of individual U-boats, as revealed in their daily position reports back to BdU, and followed the organization of anti-convoy patrol line groups, as ordered by the BdU Operations transmitter, the two RN Reservists could do for British trade convoys what had not been possible in any reliable way before: provide evasive routing around known U-boat concentrations. Noting this watershed in anti-U-boat warfare, numerous historians have credited
the newly acquired signals intelligence with causing the precipitous decline in merchant ship sinkings that occurred in July and the remainder of 1941.
6
Before considering a third possible reason for the U-boat failures of July-December 1941, it would be well to remark that Germany’s own radio monitoring and cryptanalysis service,
B-Dienst (.Funkbeobach tungsdienst),
was not itself idle during this same period. Since before the war, in fact, its staff—500 in 1939,5,000 in 1942—at Tirpitzufer 72–76 in Berlin had been busily attacking various of the British codes and ciphers, gaining entry early on to the Royal Navy Administrative Code, Auxiliary Code, Merchant Navy Code, and Naval Code No. 1. It also had modest success against Naval Cipher No. 2, adopted on 20 August 1940. But it was Naval Cipher No. 3, introduced at the start of October 1941, that most engaged the energies of the B-Dienst cryptanalysts, since its content dealt specifically with Allied convoys. Originated in June 1941 for use by the British, United States, and Canadian navies in the Atlantic, and popularly called the Anglo-American Convoy Cipher, it quickly became the conduit for information about convoy departures, routes, diversions, and arrivals, as well as about stragglers.
By December 1941 B-Dienst was making its first breaks into the cipher, and two months later it was reading as much as 80 percent of the transmitted signals.
7
This was a perilous development for the convoy system, since Admiral Dönitz, often with movement and diversion information ten to twenty hours in advance, could place his U-boat patrol lines directly athwart convoy courses. Since the same intelligence source conveyed to him the Admiralty’s daily U-boat dispositions signal, Dönitz also knew what (in general) Winn’s Tracking Room knew about his U-boat positions, though he never tumbled to the conclusion that the accurate Admiralty signals on the point were based on cracked Enigma.
There were occasions when U-boats were lost under suspicious circumstances, or when German surface refueling tankers were swept from the sea in a flash (as five were in 3–15 June 1941), or when a convoy suddenly altered course around a patrol line, or when an ocean-f of U-boats failed to sight targets and Dönitz became sufficiently troubled
about cryptographic security that he consulted Konteradmiral Ludwig Stummel, whose Second Department (Operations) of the Naval Staff in Berlin superintended naval ciphers. Stummel always assured Dönitz that Admiralty information had to have come from shore-based radio direction finding, from air reconnaissance, or from surface ship sightings. The Enigma cipher, with its variable range and changes of settings, was invulnerable to enemy penetration by all known methods of mathematical analysis, Stummel insisted, and even if an entry should be made by accident, the information developed would be long out of date and operationally useless.
Dönitz’s own staff analyzed the suspicious cases and came to the same conclusion that ordinary noncryptographic means could account for them. Nonetheless, Dönitz took the precaution of introducing a series of complex position-disguising devices into the cipher transmissions, all of which were eventually solved by GC&CS. He also restricted the number of staff who had authorized access to U-boat positions information as well as the number who were permitted to tune in the U-boat wavelength.
8
As a result of each belligerent having the other side’s key, sea warfare in the Atlantic during the winter and spring of 1943 became in part a backroom chess game as the Tracking Room, playing Ultra, sought to outwit BdU, playing B-Dienst, and no doubt on numerous occasions one side was able to neutralize the other’s advantage. After the war, in a secret report (“three copies only”) dated 10 November 1945, Paymaster Commander WGS Tighe, R.N., of the Signals Division, Admiralty, presented a report titled “German Success Against British Codes and Ciphers.” It exists today in nineteen-page summary form.
9
In it Tighe described the Signals Division as “shocked” to have learned after the war from captured documents and interrogations of German cryptanalysts that Naval Cipher No. 3 had been thoroughly compromised, and that this security failure “not only cost us dearly in men and ships, but very nearly lost us the war.” That extravagance was matched by Tighe’s subsequent claim that “all successes obtained by U-boats against convoys HX229 and SC122 in March 1943, when 22 ships were sunk, can be directly traced to the information obtained by reading our signals.”
10
Certainly B-Dienst was at concert pitch prior to that battle of 16–20 March—called by the RN official history the “biggest convoy disaster of the war”—and Tracking Room records show that B-Dienst, hence Dönitz, had prior information about the ocean course of SC.122 and about both the original route and a diversion of HX.229. What the Germans did not know was how far along their courses the two convoys had proceeded by 15 March. That was the date, at 1918 GMT, when
U-91
(Kptlt. Heinz Walkerling), en route to a refueling rendezvous, sighted and followed a destroyer that was proceeding on a northeast course in qu BC 3486 (49°33'N, 40°35W). Thinking that it might be overhauling a convoy, Walkerling trailed it and was rewarded with the chance sighting of HX.229.
11
Various sources state that knowledge of the German reconstruction of Naval Cipher No. 3 was smelled out by American Naval Intelligence in March and May 1943.
12
But GC&CS had learned from Enigma solutions as early as December 1942 that B-Dienst was reading the cipher.
13
This is confirmed by Francis Harry (now Sir Harry) Hinsley, O.B.E., M.A., F.B.A., who was a team member of Hut 8 from its beginning, where his long tousled hair and worn corduroy trousers caused him to become known to Commander Winn, a frequent contact, as “Professor Corduroy.” Four decades later, in 1981–1984, Hinsley published the official history of British Intelligence in World War II. In June 1996, when the writer brought the Tighe Report to Hinsley’s attention, the quondam cryptanalyst, later President of St. John’s College and Professor of the History of International Relations at Cambridge University, recalled how dismayed Hut 8 was to learn that B-Dienst was looking over the Admiralty’s shoulder and that no one there was doing anything about it!
“We had a watch on to see if the Germans were reading Allied ciphers,” he told the writer. “It was awkward for OIC and Bletchley Park to know the Germans were reading the Anglo-American Convoy Cipher from December to August.” The responsible person for doing something about it was Paymaster Captain D. A. “Willie” Wilson, R.N., in charge of naval cipher security under the Director of the Signals Division, Admiralty. “Wilson was slow to conclude that the Germans
were reading the cipher, for which he was criticized at the time and after the war.” When he finally acknowledged the security breach, Wilson dawdled in replacing the compromised cipher. “His argument was that there was a large number of holders of the cipher spread all over the yard, that it couldn’t be replaced partially, and that he couldn’t replace it any faster than he was doing. Well, there was a certain amount of muttered criticism about the delay at Bletchley Park and at the OIC. The delay is an extraordinary story.”
As for Tighe, unlike Wilson he probably was not “indoctrinated,” that is, “in the know” about Ultra, “thus accounting for his naivete.” His report “was probably an attempt [by the Signals Division, Admiralty] to defend themselves, to cover-up and justify themselves.” A temporary modification of the cipher was introduced, finally, in June, and the cipher was completely changed to a secure key in August 1943, after which there were no further German entries. With that closure the Allies won the cryptographic war.
14
The paper victory did not mean, however, that Ultra was the one decisive factor either in the improvement of Allied fortunes at sea after June 1941 or in the pivotal defeat of the U-boats to come in spring 1943, despite the “Ultra Myth’s” persistence in claiming so.
15
Several features of Sigint (signals intelligence) work against that judgment. For one thing, there were frequent time delays between the interception of an Enigma transmission and its decryption. As we have seen, the average intervals in August, September, and October 1941 were fifty, forty-one, and twenty-six hours, respectively. Following the decryption of the Triton key in December 1942, the delays in decryption were frequently longer than those. When one reflects that a fast convoy might cover 240 nautical miles in a twenty-four-hour period and that U-boats might proceed between 320 and 370 nautical miles in the same period, it becomes evident that a much-delayed Ultra message might have little or no operational value.
Too, at some critical times when the Germans changed Enigma settings, GC&CS went completely blind so far as decryption was concerned (though it and the Tracking Room had other, less accurate resources to fall back on, as discussed below). Such was the case during
the eleven-month drought following introduction of the fourth rotor. During the critical months January through May 1943 there were numerous failures of shorter duration, including ten days in January, seven days in February, the periods of 10–19 March, when the battle of SC.122/HX.229 was fought, and 26 April-5 May, when the protracted and definitive battle for Convoy ONS.5 was waged across 720 miles at sea (see chapters 4 through 7). And even when Triton decrypts were in hand, they were “not uncommonly” seventy-two hours old, making their operational value marginal. For all the above reasons, one has to look for another component reason for the Allies’ post-June 1941 success in protecting merchant shipping.
Here Sir Harry Hinsley helps us and in so doing sets aside any pro-Sigint bias he might be expected to have. Though he does say that “The difference between what the U-boats now [post-June 1941] achieved over-all and what they might have achieved—and, indeed, expected to achieve—was due to the great improvement in the evasive routing of the convoys that took place when GC and CS began to read the naval Enigma,” he stresses the fact that:
For the reduction in actual shipping losses the main cause was a change in the Admiralty’s policy. From 18 June it raised the minimum speed limit for independently routed ships from 13 to 15 knots, and there followed a dramatic decline in losses of “independents,” from 120 ships in the three months April to June to 25 ships between the beginning of July and the end of September.
16
Ciyptanalyst and historian David Kahn agrees that Sigint was not the main cause for the reduction in shipping losses:
… Intelligence did not always rule in the war against the U-boats. Other factors outweighed it. The July-August loss of tonnage fell to under a third of the May-June figure for reasons unrelated to B[letchley] P[ark]. More escorts were available and were accompanying convoys uninterruptedly across the Atlantic. The escorts’ experience made them more efficient. The minimum speed of ships sailing independently was raised from 13 to 15 knots. Air cover was increased.
17
It appears that all three factors examined above—loss of the U-boat aces, Allied signals intelligence, and improvements in the convoy system, especially the inclusion of 13-to-15-knot ships—had a cumulative effect in producing the turnaround of 1941; but that the last-named factor had the single most pronounced effect, as discussed below.
Whereas in World War I it took Great Britain nearly three years to institute a convoy system, thus finally stanching a trade hemorrhage that sorely threatened the island nation, in September 1939 she had in place detailed plans in the event of war for the immediate deployment of RN vessels to escort merchant ships in convoy. The first formation of ships in column sailed on 6 September. By the end of the first month of war, 900 ships had sailed in convoy without a loss, while thirty-nine Merchant Navy vessels that had been independently routed were sunk by U-boats. In the beginning there were few available escort ships—destroyers, frigates, sloops, and corvettes—that were equipped for antisubmarine warfare. The majority of destroyers were assigned to the Home Fleet for use against the Kriegsmarine’s high seas fleet. Still, the Trade Division of the Admiralty was able in September 1939 to develop three major convoy routes that could be protected with the naval assets at hand:
1. OA and OB: Outward bound from Britain to North America and Africa
2. HX: Homeward bound to Britain from Halifax, Nova Scotia
3. SL: Homeward bound from Sierra Leone
During the first four months of the war, 5,756 individual ship sailings were made in convoy, suffering only four sinkings by U-boats. The German opportunities to attack the convoys were greatly improved in the following summer, however, by their not having to sortie any longer from Baltic and North Sea bases, which required a lengthy roundabout the north of Scotland before reaching the North Atlantic sea lanes. With the surrender of France to German ground forces in June 1940, Admiral Donitz was able to base his boats on the Brittany
peninsula and Western French coast directly abutting the Bay of Biscay and the Atlantic. There, at Brest, Lorient, St.-Nazaire, La Pallice, La Rochelle, and Bordeaux, the Organisation Todt erected bombproof U-boat shelters, some of them with steel-reinforced walls three meters thick and even stouter carapaces seven meters in depth.