The Bletchley Park Codebreakers (30 page)

At a higher level of diplomatic calculation, Britain also was seeking to budge America away from her neutrality, and Churchill in particular saw technical exchanges among military experts in many areas as one way to do so. In a frank speech to the House of Commons, Churchill said his aim was to see the United States and Britain ‘somewhat mixed up together’. In growing numbers, American military experts began arriving in Britain in the summer of 1940 to study how the British military was performing in the fight against Germany, and they were handed reams of data on the effectiveness of fighter and bomber aircraft, on radar, and other extremely secret information. Co-operation on intelligence was just one dimension of a much broader British charm offensive, and it apparently so overwhelmed American Brigadier General George V. Strong (who in August had cabled back that Britain was ‘a gold mine’ of technical information) that in September Strong, on his own initiative, asked the British if they would be interested in a full exchange of codebreaking results, including US work on Japanese diplomatic ciphers. The British staff officials were astonished, but immediately accepted.

Strong then had to convince Washington. He cabled, asking for an urgent reply. William Friedman at the SIS immediately saw the advantages. Though he did not know of British progress against the Enigma, he was well aware how behind the United States was in dealing with German military traffic in general. German signals were impossible to pick up from the continental United States, and the Army was essentially devoid of knowledge about the German military cryptographic systems. The US Army had not even begun
paying much attention to German traffic, military or diplomatic, until October 1939, concentrating instead on Japan, Russia and Mexico. The benefits of an exchange of raw traffic, at a minimum, were manifest. Friedman was more cautious about the idea of exchanging cryptanalytic research or results, but thought it might work on a strict item-by-item quid pro quo.

The US Navy, for its part, had been able to intercept some Enigma signals from the east coast of the United States, but had made no significant progress in decrypting the material – and it had a growing need for operational intelligence about German U-boats. Yet the Navy was so suspicious of British motives that Safford would agree to nothing beyond an exchange of traffic.

By the late autumn of 1940, the US Navy had at last agreed to go along with Friedman’s proposal to exchange technical data with GC&CS, and the British had agreed to accept a visit by American technical experts. Roosevelt gave his blessing on 24 October and the next day the US Army’s chief signal officer. General Mauborgne, forwarded to the US Army staff a list of what the SIS proposed giving to the British – including its solution of the Purple machine. But this was far from constituting a formal agreement. The planned visit by the American technical experts was viewed by both sides as really just an initial exploratory conversation. And internal British memoranda make abundantly clear that the British plan was to stonewall if questions about the Enigma came up: the British directors of military, air and naval intelligence conferred and concluded that ‘a full interchange on Germany and Italy cannot be entertained at this stage’. Instead, when the American ‘expert’ arrived, ‘steps will be taken to steer him away from our most secret subjects’:

‘Y’ referred to the interception and direction-finding of enemy signals and to the decryption of low-level tactical and field codes, and most definitely did not include the ‘innermost secret’ of the Enigma. It was not until after the four American cryptanalysts – Abraham Sinkov and Leo Rosen from the Army, and Prescott Currier and Robert Weeks of the Navy – had been in the country three weeks that a decision was finally made to reveal to them that the Enigma was in the process of being broken and the methods that were being used. A memorandum from ‘C’ to Churchill dated 26 February 1941, reported that the British Chiefs of Staff ‘on balance’ now favoured ‘revealing to our American colleagues the progress which we have made in probing the German Armed Force cryptography’. But ‘C’ assured the Prime Minister that the discussions would be on technical aspects of cryptanalysis only and would not extend to ‘the results’, i.e., actual intelligence derived from Enigma traffic. A handwritten note from Churchill at the bottom of the memo reads ‘As proposed. WSC. 27.2’.

On 3 March 1941, the Americans were informed of the British success against the Enigma. But there were strict limitations. A handwritten agreement dated that same day and signed by Weeks stated that they should inform no one of the information they had acquired, except the head of their section, Commander Safford, and him ‘by word of mouth only’. (The Army representatives agreed to a similar restriction.) The Americans would be given the wiring of the naval Enigma rotors, but agreed to ‘disclose that only when it is decided to work on the problem’.

There has been considerable controversy in subsequent years over the exact nature of the exchange of information that occurred during this visit. An account by Safford, written many years later, claimed that the Americans had been double-crossed; the United States gave the British its most closely held secrets about Japanese codes, even supplying a copy of the Japanese Purple machine that the US Army cryptanalysts had reconstructed, while the British gave the Americans nothing in return.

But the true situation was more complex. Co-operation on Japanese codes began almost immediately and was definitely a two-way street. Even while the Sinkov mission was at GC&CS, the British cipher unit in Singapore received radioed orders to begin sharing information on JN-25 with the US Navy’s intercept and decrypting unit at Cavite in the Philippines, known as Station Cast. In February 1941, the two bureaux
in the Far East exchanged liaison officers and thereafter regularly sent each other recovered code groups. (By June 1941 only about 1,100 code groups of the 55,000 used in the new JN-25B codebook had been identified between the two units, and these were mostly groups that stood for numbers and other mundane and common words or symbols that yielded little of intelligence value. By the time of Pearl Harbor the total of recovered groups was still under 4,000, and a change on 1 August in the additive book used to conceal the code groups’ identities in actual messages further blocked attempts to read traffic. At the time of Pearl Harbor not a single JN-25 message had ever been read currently, and since the December 1940 codebook change no JN-25 messages had yielded more than fragments, of no intelligence value.)

Co-operation between the British and American codebreakers, in the true sense of the word, also began on German diplomatic codes. On Enigma, however, the British drew a careful line between ‘research’ and ‘exploitation’. It was one thing to have a technical discussion about cryptanalytic theory; it was quite another to let anyone else in on the actual intelligence operation – the system that transformed intercepted German military communications into military intelligence. The Americans were told about the Enigma and were shown the bombes, the electro-mechanical devices Bletchley Park had developed to recover the daily Enigma settings. But the visitors were not permitted to take notes, nor was GC&CS terribly quick in replying to subsequent requests for further details. On 15 July 1941, Washington requested information about daily Enigma settings; a month later the material ‘was still being copied’. Denniston was meanwhile ‘aghast’ to receive a letter from the United States asking for a copy of the bombe, explaining that ‘we avoid as far as possible putting anything on paper on this subject’.

Denniston was scheduled to travel to Washington on 16 August. ‘Perhaps the most important purpose of the visit is to clear up the position concerning E traffic’, he wrote in a memorandum to ‘C’. Denniston proposed that he should explain to the Americans during his visit that GC&CS had received so far only 6 of the 36 bombes it had ordered and could not spare one. Allowing an American firm to build a bombe was out of the question on security grounds, he wrote; for the American cryptanalysts, the Enigma was at best a ‘new and very interesting problem’, but for Britain it was ‘almost life-blood to our effort’. But perhaps, Denniston suggested, it might be possible for
the US codebreaking bureaux to send some ‘young mathematicians’ to Britain to work with GC&CS directly on the Enigma. ‘C’ shot down every one of these suggestions. He replied that he was ‘a little uneasy about the proposal for young mathematicians’ and that ‘I should feel inclined not to mention’ the bombe situation at all lest that might be used as an argument by the Americans to have them built in the United States, which the British were adamantly opposed to.

On 1 October 1941, Denniston did apparently write Safford a letter supplying the information OP-20-G had been after, but the letter went astray. On 27 November, Admiral Leigh Noyes, the US Director of Naval Communication (under whose jurisdiction OP-20-G fell), sent a long complaint to the British alleging that they had failed to carry out their end of the bargain. The British reply, from ‘C, did little to mollify his concerns, for it argued rather jesuitically that on naval Enigma, ‘everything asked for has been supplied’, which did not exactly seem to the Americans to reflect a spirit of true co-operation. Finally Denniston realized that his letter to Safford had not been received and cleared the matter up; by 12 December the letter had arrived and Noyes stated that everyone was now satisfied.

Nevertheless the British position was untenable in the long run, for it still insisted that neither the US Army nor the US Navy need participate directly in ‘exploiting’ German Enigma traffic – that is, in decrypting it and distributing the results on an operational basis. When it came to control of the product, that was a British prerogative, period. The Admiralty would ‘pass German naval intelligence to Navy Department when U.S.A. was affected’, Washington was informed. As for the US Army, they clearly were
not
affected by the ground war in Europe:

‘Interested’ in this case was clearly meant in the strict sense of the word, that is, when the US Army had an interest at stake in the actual
fighting. With the entry of the United States into the war, both OP-20-G and Arlington Hall (as the US Army SIS establishment came to be known after its move in 1942 to the former girls’ school of that name in Arlington, Virginia) renewed their demands for direct involvement in decrypting German Enigma traffic. And now British opposition began to assume a different complexion. Knowledge is power, and being able to control the distribution of such a valuable form of knowledge as Enigma intelligence gave the British a not inconsiderable point of leverage in the alliance. It was not so much that the British were out to directly manipulate the information or withhold it when military necessity dictated it be provided to field commanders, British or American. But many decisions in negotiations over military strategy between the new allies were tilted by arguments over the relative contribution each side was making to the war effort, and the Enigma was one bargaining chip: the British monopoly over this intelligence was one thing that made the Americans undeniably beholden to them. By the same token, the growth of a significant independent American capability would make the British – and British views – that much more dispensable.

Denniston was alarmed by the new American demands for fuller participation in the Enigma work, and at once sent a message to the British liaison in Washington:

For the US Navy, what finally broke the British claim to monopoly was GC&CS’s sudden failure to continue reading the Atlantic U-boat traffic. On 1 February 1942, the Atlantic U-boat networks changed from the three-wheel Enigma to a four-wheel machine. Running the bombe through all possible positions for one wheel order of a three-wheel machine typically took about fifteen minutes of machine time
(the set-up time for each run added another ten or fifteen minutes, and additional time was required for testing the results of the run). The addition of a fourth wheel meant that the time required for each run on the three-wheel bombes was multiplied by a factor of twenty-six – the machine would have to be run through all possible wheel orders
and
at every possible position of the fourth wheel; testing all 336 wheel orders at all twenty-six positions of the fourth wheel could thus in theory take several thousand hours on a three-wheel bombe. Even with all sixteen bombes that were available at this time running simultaneously, and even exploiting methods that could reduce the number of wheel orders that needed to be tried, recovering each daily setting might take two days or more of effort. But there were other problems as well, notably a lack of suitable cribs. What was needed was a new, four-wheel bombe that could run at much higher speeds.