The Secret in Building 26

by Jim DeBrosse

  But their access wasn’t complete: the visitors were to be shown only the new technology, not the operational “results” of British codebreaking.

  On March 3, 1941, the Navy men visited Hut 3, where Alan Turing was refining his attacks on the naval Enigma. Turing described to the Navy men his method of Banburismus, the formidable challenges of the naval Enigma, and various German transmission systems. Although critics later complained the Americans were not given enough information, what Currier and Weeks were told and shown at Bletchley Park was significant—if disappointing.

  The machinery and methods that two of the Americans finally did see in Hut 3 were skimpy—the later hustle and bustle and technical sophistication of the 1944 and 1945 Bletchley Park that came to dominate its postwar public image was still a long way off. The Americans in 1941 saw four, maybe six Bombes at most, and what they saw was rather crude. The first models that printed their results did not arrive at Bletchley until the end of March, after the Americans had left.

  The first electromechanical Bombe, with its sets of noisily whirling commutators, had begun its work at Bletchley less than a year before. Its first break into Enigma had required a full week of runs.

  In that early Bombe, the commutators continued to spin for some time after a hit. When the machine at last stopped, its operators had to feel the relay switches in its circuits to discover which were warm and which were cold in order to identify the hit positions of the commutators. Then the machine had to be cranked back to where it had made the hit before it could restart its search. *11

  Turing no doubt told Currier and Weeks of his complex way of turning cribs into menus. But he no doubt told them as well that, to be effective, the Bombes required long, rare cribs containing loops and other telling patterns between the cipher and plain text.

  The Bombe attack relied, to some extent, on old-fashioned cryptological craftsmanship, not to mention the theft of enemy documents. But such captures, which allowed the naval system to be read for much of the remainder of 1941, did not come until after the Americans had left Bletchley.

  We do not know exactly how much of Bletchley’s technical history and frustrations were revealed to the visiting Americans. But if there had been a frank discussion, Weeks and Currier may have brought back a less-than-optimistic picture of the Bombe and Turing’s methods. The British had found that the Enigma was not going to yield to any simple or inexpensive solutions.

  The Navy’s team did not leave empty-handed, however: Weeks and Currier received a paper analog of a naval Enigma machine, which used strips of encoded paper rather than wheels for scrambling the plain text. They also were told the wiring of the machine’s eight code wheels and were assured that when enough copies of real Enigmas became available, the American Navy would receive one. In addition, Turing promised them much more would be exchanged in the future, including information about German errors in using the Enigma systems. *12

  Although all four Americans had learned enough from their visit to prevent them from thinking that the British had reneged on their agreement, Sinkov and Rosen had found a brighter scenario at Bletchley: they had learned of Britain’s longtime success against the Axis diplomatic and Air Force ciphers and of more recent breaks into the German Army Enigma networks. Currier concluded they had been “shown everything” and that the exchange had been open and equal.

  Nevertheless, Bletchley’s gifts to the Americans had come with some severe restrictions because the British felt the Ultra intelligence was crucial to their country’s survival. What the two teams saw and learned at Bletchley Park was to be held top secret, never written down, and conveyed by word only to their immediate superiors who themselves were expected to keep the secrets. The knowledge was never to be put to use before GCCS was contacted or, by implication, without its permission. The four Americans willingly signed ironbound oaths of secrecy before their departure in the first week of March 1941. By contrast, what the Americans gave to the British was offered without restrictions.

  The Americans also revealed to the British, in broad terms, their plans to build revolutionary codebreaking machines, including electronic ones. Rosen, the technological expert among the Americans, may have engaged in a bit of puffery, leading the British to expect far more advanced machines and more of them when they visited the United States later in the year. There was at least one omission as well: the Navy team seems to have kept the British unaware of Driscoll’s German naval project.

  Even so, the British, too, felt the exchange had been worthwhile. After all, they had gotten their hands on a Purple machine and had felt reassured that, based on the demeanor of Currier and Weeks, their secrets would be safe with OP20G. On the day the Americans were first shown the Bombe, Denniston wrote to “C” that “complete cooperation on every problem is now possible and we are drafting plans for its continuity when they return to the USA.” All that was needed now were secure telegraphic and courier links between the liaison officers. *13

  But trouble came almost the instant the four Americans left. British intelligence quickly informed GCCS that one of the men on the Army’s team had talked about his work so much on the British ship from London to the Royal Navy anchorage, in the Orkney Islands, that its officers knew what his visit had been about. *14

  In the next few months, there were further incidents that made the British regret opening Bletchley’s doors. GCCS received a letter in plain language from the American Army asking for a Bombe. The British view of the American State Department as an intelligence sieve was reaffirmed when a long newspaper article appeared in Britain describing the American plans to launch a new intelligence covert operation in England. The June 24, 1941, London Daily Express reported from New York that William “Wild Bill” Donovan “has a new hush-hush mission—to supervise the United States Secret Service and ally it with the British Secret Service. . . . The American ‘Mr. X,’ as he is known privately, will report directly to the President.”

  But America’s aid was too valuable and its Navy too involved in the Atlantic to be cast aside. Bletchley Park had no choice but to fulfill its promises. It sent a list of Enigma settings requested by OP20G to Washington in late May and early June 1941, as well as a copy of the bigram tables, which were vital to determining the indicators for setting up the machines.

  Sharing Enigma research with the Americans was worrisome enough for the British. Sharing Ultra secrets seemed unthinkable. Even after the U.S. Navy had begun escorting merchant ships across the Atlantic in 1941 and its vessels were being hunted by German subs in an undeclared war, British leaders resisted the idea of sharing U-boat locations with the Americans. Britain had its own very stringent rules for using Ultra in its naval operations: if the information could not be covered up with a phony source, such as an airplane sighting, it could not be used at all. “C” wrote to Churchill in June 1941 with his fears about supplying Ultra information from U-boat intercepts.

  I find myself unable to devise any safe means of wrapping up the information in a manner which would not imperil this source. . . . [I]t [is] well-nigh impossible that the information could have been secured by an agent, and however much we insist that it came from a highly-placed source, I greatly doubt the enemy being for a moment deceived, should there be any indiscretion in the U.S.A. That this might occur cannot be ruled out, as the Americans are not in any sense as security minded as one would wish.

  IN MID-1941, Agnes Driscoll and Laurance Safford were still convinced that they had found a way to break into the Enigma without the aid of a machine—by generating a huge paper catalog of the plain-cipher letter combinations it produced. Catalog attacks were a mainstay of codebreakers in the West because they were logical and obvious, even if labor intensive. *15

  What Driscoll and Safford either didn’t realize or resisted, despite the information brought back from England by Currier and Weeks five months earlier, was that they were dealing with a whole new breed of the Enigma that demanded a much more powerful attack and the a
id of advanced machines. Driscoll’s approach could get only a small part of the solution for the steckered three-wheel Enigma—that is, the starting positions of the code wheels—and only after the much tougher keys to the wheel order, steckers, and ring settings had been found.

  It was Driscoll’s dogged attachment to her catalog attack that proved her undoing. Indeed, Driscoll and her colleagues at OP20G were ignorant of their own ignorance about the challenges of the Enigma. Driscoll launched her effort assuming that the best German systems and enciphering machines would be little more complex than earlier commercial Enigmas. Driscoll’s mistaken image of a simple electric code-wheel machine and her realization that the Navy could not steal or capture a machine or its supporting documents led her to a historic commitment: the American attack, she decided, would be based on approaches that needed little knowledge of German operational procedures and that would be as free as possible of a dependency on German procedural errors.

  Typical of her work habits, “Miss Aggie” left no record of her anti-Enigma methods. And there are few references to her projects in late 1940 and 1941 in the histories and technical documents written by her associates. But it is clear that her goal, like Safford’s, was to make the American attack independent of the British and of everything but intercepted messages. Just as important, the attack had to be “economical,” since she and Safford did not anticipate the Navy providing them with expensive equipment or a large staff. Indeed, in the last months of 1940 and the first quarter of 1941, OP20G did not have the machinery or even a fraction of the personnel needed for a successful catalog attack against the Enigma. The only path left for her was a tried-and-true approach—and a dead end.

  IN AUGUST 1941, five months after the American visit to Bletchley, Denniston traveled to America to cement relations with the codebreakers there, leading to his historic encounter with Driscoll. Denniston’s aims in Washington were not only to smooth over the conflicts so far but to coordinate, appease, and control the Americans. His first step was to persuade Friedman at the Army’s Signal Intelligence Service (SIS) that England had no Bombe to spare for them. But he also had to prevent Friedman from turning to IBM to build one of his own. His goal was to persuade Friedman that SIS had no need to venture beyond its research role in grappling with German Army and Air Force codes. England would tell SIS everything they needed to know about methods and would supply them with timely information as soon as the American Army had any real involvement in Europe. In a last-gasp tactic, Denniston was prepared to invite SIS to send some of its mathematicians to Bletchley so that they could feel they were part of the system and save face. But Denniston hoped that offer would not be necessary; the British continued to have deep fears about American security.

  On the positive side, Denniston urged Friedman’s group to concentrate its energies and, he thought, its vast technological resources on Japanese systems. But he was surprised and disappointed after he toured SIS headquarters to find it had few tabulators and none of the more advanced machines that Rosen had alluded to.

  Still, Denniston got full compliance from the SIS. He came to consider SIS “our friends” who wanted to learn from—and be subordinate to—Bletchley’s codebreakers. Denniston concluded that Friedman was the real force in American cryptanalysis, and the two became close friends. Friedman may have sealed the friendship by sending Denniston, an avid golfer, a package of golf balls, which were in short supply in England during the war. Denniston sent a heartfelt thank-you note, adding, “Believe me, no one loses a ball here these days. It may add an hour or so to the round, but balls must be found!”

  DENNISTON FOUND QUITE a different reception at the Navy’s OP20G headquarters, where he met with the leaders of the naval communications section. Denniston agreed to furnish more information on French and Italian systems, and he established procedures for secure communications with England. But a few days later, during an intense meeting with Safford, he probably admitted that the American Navy deserved more help with the German Enigma because its ships and sailors were virtually at war in the Atlantic. Even so, Denniston continued to press for affirmation of what he believed the Americans had promised: to do nothing more than research on European naval systems.

  Always the gentleman, Denniston held his tongue when he learned that Currier and Weeks had given information about Bletchley’s work to others at OP20G besides Safford, violating their handwritten agreement. For when Denniston met with Agnes Meyer Driscoll, it was clear that she had been privy to those secrets. She was probably using Britain’s knowledge of the Enigma’s wheel wiring by then, and she seemed informed of the Bombes and even Banburismus, he concluded.

  But his biggest shock no doubt came when Driscoll told him that the American Navy didn’t want a Bombe, didn’t want to use it, and thought little of the other methods against Enigma that Turing had revealed.

  Driscoll declared she had found a far better approach than the British—a special catalog method. With a bit more work, she said, it could become operational. It was, she claimed, much simpler, demanded less material than the Bletchley attacks, and would be better able to withstand changes in the Enigma systems. It would need only a few daily messages and very short, commonly worded cribs—not the hard-to-find ones the Bombe required. No revolutionary machinery would be needed. And her attack would be more effective in dealing with the problem of Enigma wheel turnovers after only a few letters were encrypted.

  She showed Denniston, who by this point must have been appalled, a sample solution based on an eight-letter crib. She believed that, even with fewer than two dozen people using her soon-to-be completed catalog, messages could be solved within a few days. Those claims probably seemed outlandish to Denniston, given Britain’s need for hundreds of people and very expensive machines to penetrate any Enigma system.

  Denniston remained composed—a British gentleman to the last—even though he must have felt insulted. And although his goal to keep the naval Enigma under British control was not endangered by Driscoll’s method, he seemed threatened by her attitude. He was unable to dissuade Driscoll from pursuing her catalog attack. *16

  Regardless of how much detail he may have supplied, Denniston stressed that most catalog methods, when faced with a complex machine like the naval Enigma, could not solve enough of the machine’s settings to work on their own. And he no doubt emphasized that the Bombe—despite its reliance on hard-to-find cribs, expensive methods such as Banburismus, and the exploitation of German procedural errors—was so far the only machine to have penetrated the Enigma on a timely basis.

  But when he offered to provide more information about the Bombes and Turing’s methods, Driscoll wasn’t interested. He might have been willing to supply OP20G with a Bombe when one became available, but if so, Driscoll certainly wasn’t requesting one.

  “Miss Aggie” did bend a little. She admitted that she was somewhat “stumped” in her quest to fully understand the Enigma. (Denniston might have suppressed a smile at this point.) She had been unable to build a fully working Enigma from the paper analog and the other documents Currier and Weeks had brought from England. She voiced frustration over the double turnover of one of the Enigma code wheels, and she demanded clarification on how it and a number of other Enigma features worked.

  Denniston also had to bend. He promised to send responses to all her questions and asked her to compose a list for him. He also promised to forward relevant codebooks and a working naval Enigma machine when one was available.

  But when he invited Driscoll to come to England and to learn firsthand about British methods, as well as to inform Bletchley about hers, she turned him down cold. A visit was out of the question, she said, because of her health. Yet she did not suggest that any of her crew take her place, nor did she invite a British expert to Washington to work directly with her.

  The meeting seems to have concluded with Driscoll reaffirming her faith in her approach. She promised to inform Bletchley Park of the details of her superior method a
nd, indeed, did send off her list of questions to Denniston. The list was very specific and did not reflect what Leigh Noyes later insisted was true: that OP20G had expected the British to supply everything about the Enigma, without specific requests.

  Driscoll’s list of questions only further reflected her hubris. It did not include anything about the Bombe and or any of the other British attacks. She was just after enough details about the Enigma and the German naval systems to allow her to advance her own attack. So, with a tiny staff of four, Driscoll pressed on with her catalog approach to Enigma. *17

  Driscoll’s work continued to have Safford’s full support. In late 1941, he approved her requests for more resources, although she had admitted that one of her attacks had failed. By the end of the year, her team had more than tripled, to fourteen people. Letter pairs were generated and punched on tabulator cards, using precious machine time and manpower, while Driscoll, thinking she needed an actual Enigma machine, awaited her copy from the British.

  In mid-December 1941, Driscoll finally sent the British some information on her special method, with only cursory answers to the few questions Denniston had posed four months before. Driscoll again declared her faith in her approach, but GCCS concluded that it had “apparently failed.” For one thing, it could not, as she had claimed, overcome the problem of turnover—that is, the tumbling of the Enigma’s wheels before enough letters had been enciphered to identify the wheel being used. And as Turing pointed out to her in a letter in October 1941, her method would take seventy-two thousand hours—more than eight years—to find a solution. Given Driscoll’s obstinacy, Bletchley Park began to have second thoughts about providing more technical information.