The Secret in Building 26

by Jim DeBrosse

  In a January 20 memo, Engstrom asked Meader to “advise Mr. Desch of the urgent necessity” of using a one-size-fits-all wheel. But on this point, Desch sided with his engineering instincts rather than the needs of codebreakers. His first two prototypes—Adam and Eve—both used different-size wheels.

  Turing, however, was able to predict at least one engineering pitfall that Desch and his team had not: laboratory testing alone could not anticipate all the problems with the high-speed brushes used to sense the commutators’ electrical connections. The British knew from their more advanced trials that the brushes had a tendency to bounce over the contacts—a glitch that caused numerous false hits and misfirings. Just as Turing had done at Bletchley Park, the Americans had first tested the contacts on an oscillograph and not detected any problems. In his memo, Turing cited the British experience with “brush bounce”: “Such a demonstration was made by Commander Travis and Flowers and myself [using an oscillograph] at Malvern, and yet when it came to the point of lining [our Bombe] up for a trial menu, it failed on account of bounce.”

  Engstrom passed the warning on to Meader in a January 5 memo, as well as some more general advice from Turing, who knew all too well how suddenly and dramatically the Germans could change their Enigma systems. Engstrom put the NCR team on guard to keep the Bombe design as fluid as possible “in order to permit the addition of new features, or the change of those already incorporated, from time to time as rapidly as possible.”

  Turing’s memo was not all negative. He gave understated praise to a Desch innovation that eliminated the need for a circuit breaker on the Diagonal Board and possibly reduced the number of false stops at the same time. “I think they have got something here, but it remains to be seen how great are the transients [stops] that remain,” he wrote. Desch’s idea of using electrical switches on the Diagonal Board rather than plugs and a circuit breaker was even more remarkable in Turing’s view, given that the Americans had proceeded much on their own. “Starting from scratch on the design of a Bombe, this method is about as good as our own,” he wrote.

  What the Turing memo makes clear is that the Desch machine already was different enough from the British Bombes—in its central drive and gearing system for spinning the commutator wheels, in its menu setup, rewind feature, and particularly in its electronics—to label it a truly original American machine, not just a colonial knockoff.

  The British had used a system of electrical relays in their first Bombes to identify the positions of the wheels when a hit was made. The system had proved reliable, but Desch had felt that for a more powerful machine a relay system would be too slow. He turned to his experience with the NCR electronic calculator and created a digital electronic tracking and control system that amazed the Navy’s engineers. In his final version of the Dayton Bombe, his system of several hundred tubes did more than record the position of a hit, it acted as a control center for the machine. It was able to track the wheel positions, signal the motor and the clutches, then return the wheels to their hit positions. Electronics again played the central role in testing for stecker contradictions. At that point, the wheel locations and steckers were printed and the operator could signal the machine to restart its search.

  While Wenger, Engstrom, and other top officials at OP20G no doubt read Turing’s report and passed on what they might have learned, Desch himself was never shown the memo, to protect his morale—a delicacy of consideration that later amused Desch. “After the war, they showed me the reports and they weren’t very complimentary,” he told Tropp in his 1973 interview, then added, chuckling, “[The British] must have had to swallow hard because we had to build 100 machines for them.” In fairness to the British, Bob Mumma spoke up in that same interview: “They never dreamed they might do like we did given their limited resources.”

  But the truth is, the British tried with their limited resources to take the same machine-intensive approach. Desch and his team didn’t realize the British had already built fifty three-wheel Bombes by early 1943 and were manufacturing almost two per week, albeit with limited success against the four-wheel naval Enigma.

  THE TEMPO AT NCR picked up as spring of 1943 approached, but the men in Dayton were not keeping pace with the war. Desch still was bogged down in the Navy’s highest-priority project, nowhere near achieving his goals, while brave young men were dying in the icy waters of the Atlantic.

  7

  Troubles with Adam and Eve

  March 19, 1943—Dayton, Ohio

  AS ONE OF the first maintenance technicians assigned by the Navy to the U.S. Bombe project, Phil Bochicchio, who was to become the project’s floor supervisor, arrived in Dayton in March 1943 with orders to report to the U.S. Naval Computing Machine Laboratory, no street address given. But when Bochicchio’s train pulled into Union Terminal, no one at the USO lounge there had ever heard of the laboratory, nor was there a listing for it in the phone book. Bochicchio, a twenty-year-old machinist’s mate from New Jersey, wasn’t shy about asking. He walked the two blocks north from the train station to downtown, then the five or six blocks around to the police and fire stations and the Navy recruiting office. None had heard of the lab, either. Finally, at the Dayton Municipal Building, a Chamber of Commerce official offered to drive Bochicchio out to NCR, where the man knew the Navy had several other projects in progress. At NCR, Bochicchio promptly reported to the Navy liaison officer, but the officer seemed clueless as well, Bochicchio recalled: “He looked at my orders and said, ‘I don’t know what to do with you.’ I said, ‘Well, great. Just give me something to do.’ ”

  For the next two or three weeks, Bochicchio designed toolsat NCR before he learned anything about the project that hadsummoned him to Dayton. Unknown to him, the Navy and FBI were completing a thorough background check before giving him his orders. “Finally, I got a letter from my dad saying [to] call home as soon as I can. When I called, my dad said, ‘What kind of trouble are you in, boy? The FBI has been here, naval intelligence has been here. All the neighbors are wondering.’ ”

  Bochicchio wasn’t in trouble, but he was in for a lot of hard work over the next two years. As floor manager, he was in charge of setting up and debugging the sensitive decrypting machines. Unlike the later production models, which stood more than seven feet tall and weighed two and a half tons each, the first two prototypes, Adam and Eve, were small enough to stand on sawhorses, but like all complex, unproven machinery, they were prone to seemingly endless glitches.

  Bochicchio said a big part of Desch’s engineering headache was trying to please both the Navy theoreticians, who designed the logic of the machine and insisted on speed, and the Navy codebreakers, who wanted something reliable and easy to operate and maintain. “The mathematician thinks one way, the cryptologist thinks another. And you’re sitting in the middle, and you have to try to figure out how in the hell to give them both what they need from the machine.”

  Following Turing’s visit, debate between Navy engineers and codebreakers reached its height over the design of the fast wheel on the machine. In March 1943, Engstrom repeated his “request,” in no uncertain terms, for a one-size wheel design. This time around, Desch complied with his boss’s wishes, although the stresses on the single fast wheel proved to be a source of continuing frustration for him and his design team, even after the first machines went into production.

  OP20G’s September 1942 memo requesting money for the Bombe project had promised to deliver the first NCR machines to the Navy in February 1943—a promise that was unrealistic but nonetheless backed by Desch himself. A month after the approval of his design, Desch had written to Meader on October 23, 1942, that he would need close to sixty sailors for assembly and repair work on the Bombes, beginning at the end of the year, “so that they will be thoroughly trained in the servicing of the equipment when it is ready for use.”

  If Americans suffered from any universal flaw during World War II, it was the belief that Yankee industrial know-how could work miracles overnight.
Hence, if the first British Bombes were designed and produced in eight months, why then American industry ought to be able to do the same, even with a far more complex machine, in little more than half the time.

  But by February 1943, after five months of intensive development, all Desch had been able to fashion were two wheezy prototypes that were prone to oil leaks and breakdowns after only a few minutes of operating at the speeds demanded. A month later, when the German U-boats unleashed their fury against Allied shipping in an all-out effort to turn the war, Desch still hadn’t produced a working machine.

  AS THE SPRING U-boat offensive opened, the Germans changed some of their codes and tightened up their procedures so that the Allies were again shut out of the submarine code systems. They remained blind for more than a week during what became the worst month for the Allies in the Battle of the Atlantic. More than twice as many Allied merchant ships (ninety-five) went to the bottom in March as in February.

  The destruction might have been even worse had another surge of insight among the British codebreakers not allowed the three-wheel Bombes to be useful again. U-boats were required to report the sighting of any Allied convoy to Admiral Dönitz’s headquarters, using a special short-signal code so that Allied direction-finding equipment wouldn’t have time to home in on their transmissions. However, the British had a copy of the latest codebook for those short signals, captured from U-559 in November 1942. The codebooks, plus an order to U-boat skippers to report weather conditions and convoy sightings in the Enigma’s simpler three-wheel mode, allowed Bletchley’s Hut 8 to solve Shark for 90 of the 112 days between March 10 and June 30, 1943.

  Although the decryptions were by no means immediate, with many keys taking weeks to solve, old-fashioned cryptanalysis had again saved the day, and it was a good thing. The British were having as little luck as the Americans with their two separate designs for a four-wheel Bombe. During tests in April, the projected fast-wheel speed of one prototype had to be drastically reduced from three thousand rpm to twelve hundred. The first production models, introduced in June, were to prove less than reliable as well.

  By mid-1943, with its ongoing problems as well as OP20G’s promise to let the British oversee its Enigma efforts, England wished all the best for the NCR program. After all, the British reentry into Shark had been based on such an unpredictable foundation—captured materials and enemy error—that Bletchley Park didn’t dare accuse the Americans of wasting resources on developing a four-wheel Bombe. The NCR Bombe offered perhaps the best hope of a consistent break into Shark.

  Despite all the setbacks in Dayton, OP20G had no choice but to continue its investment in the project, and it bet millions of dollars more that Desch and his engineers could overcome the next set of technical hurdles. They instructed him to rush to the production stage and stand ready to build as many as three hundred machines to tackle what the British feared most—changes in the Enigma systems that might call for all 336 wheel orders to be tested, and with only weak cribs to help.

  While Desch and his men fretted over the final design of the Bombes, other sections of NCR were setting production records, something that made the Navy optimistic about the Building 26 project, if and when it moved into the manufacturing phase. Despite its early problems meeting Navy demands in 1941, NCR went on to establish one of the best corporate records of the war. In January 1942, NCR earned the first of its five “E” awards from the Army and Navy—an honor given to companies that were deemed to have gone beyond the call of duty. Only one other company in America, Dayton’s Frigidaire, earned an equal number of E awards.

  Part of NCR’s success was due to the company’s improved ability to direct outside subcontractors. By mid-1943, the engineers and technicians in Building 26 were amazed at how quickly they could get anything they wanted from NCR and the hundreds of subcontractors on the project. Thousands of different precision parts were produced for each Bombe. “From an engineering standpoint, it was wonderful,” said Lou Sandor, one of the NCR engineers assigned to Desch’s team.

  The abilities to produce reliable precision parts and to deliver them on time were to be essential to the complex assembly and testing operation. Gears would have to be made to specific tolerances and be hard enough to withstand high speeds, sudden braking, and intense stresses. The commutators would have to be exactly balanced so that they would not distort while spinning. Their embedded metal contacts would have to be cut and milled to exact lengths so that electrical pulses to the rest of the machine were perfectly timed. Even the miles of electrical cable for each machine would have to be specially designed and ordered.

  Secrecy requirements made the supervising of subcontractors tricky. They could not be told the why of their work, and every effort was made to prevent them from guessing. For instance, the number 26 was never used on parts specifications. The manufacturers of the commutators were told to number their contacts from 00 to 25. The cable manufacturers were told to make the cords in twenty-eight, not twenty-six, different colors.

  Even when the secrecy problems were overcome, creating a functioning Bombe was a tall order, especially when the first wheel of the machine was designed to rotate at close to two thousand rpm—down from a hoped-for 3,440 rpm in September 1942 but still as fast as a car engine cruising at highway speeds. At such velocities, keeping the wheels in balance and in their original shape was a formidable task, especially since Engstrom had nixed Desch’s idea of different-size wheels.

  The Adam and Eve prototypes were prone to sparks and short circuits that ruined decoding runs and oil leaks that created maintenance nightmares. The technical hurdles continued as Desch moved during the spring and summer of 1943 to his second set of prototypes, Cain and Abel, and then to his production machines. To make proper contact with the spinning wheels, the sensing brushes on the face of the machine had to be kept oil-free. Power from the five-horsepower electrical motor had to be distributed evenly through a labyrinth of motors, shafts, gears, and clutches, or the resulting stresses would cause the machine to overheat and grind to a halt. The line shafts had to be beveled to precise tolerances so that the rewind mechanism would return the machine to the exact location of the hit. In short, the NCR Bombe was like a high-performance race-car engine being pushed to its limits—before anyone even knew for certain it would work.

  BESIDES GRAPPLING WITH the Bombe’s technical flaws, Desch also had to contend with his boss and unwanted houseguest, Ralph Meader. The project’s naval supervisor was constantly on Desch’s back. Mumma said the pressure on Desch “just got worse and worse. . . . We worked seven days a week, you know, and pretty long hours. One Easter Sunday, I know, I laid off and didn’t come to work and Joe was awful mad about it.” Vince Gulden, another member of Desch’s engineering team, remembered working ninety-hour weeks that spring. “Meader was on the quarterdeck with a whip and he was giving us stories about [the Battle of the] Coral Sea—how it wouldn’t have happened if you guys had been there and finished earlier. We heard that more than once.” The Bombe, of course, would not have helped the Americans against the Japanese code systems in the Pacific, so Meader was either bluffing or ignorant about the purpose of the NCR machines. The former seems more likely.

  Guilt was a common method for motivating workers in war plants all over the country. Posters on bulletin boards, next to time clocks, on factory walls, and in break rooms often showed factory workers shoulder to shoulder with soldiers and sailors. The idea was to create an attitude in workers that they were “production soldiers” on a par with the men fighting on the front—and to remind them that, unlike those in combat, they were not risking their lives and should therefore pull their weight by working harder. The factory-worker-as-warrior message was so prevalent by 1944 that an official of the National Association of Manufacturers was moved to quip, “The boys in the foxholes would, on their return, be forced to employ a press agent to convince the public that soldiers, too, had something to do with our victory.”

  When he could
get away for an hour or two, Desch liked to retreat to his garden plot off Wilmington Pike, near a local dairy at the time, where he grew everything from corn to kohlrabi, his daughter said. And he loved to whistle, mostly Sousa marches, snatches of classical music, and romantic movie scores. But the supersecrecy of his work kept him from confiding in the one person who might have been able to help him handle the stress: his wife, Dorothy.

  Desch had a surprisingly romantic and sentimental side. His letters to Dorothy were always addressed to “Sweetie-pie” and often ended with strings of Xs for kisses. On a business trip to New York in 1938, three years after their marriage, he wrote her two letters on the same day, morning and evening, both telling her how much he missed her.

  When he could find the time, he loved to go out dancing with Dorothy. He was an excellent ballroom dancer, especially at waltzes, Debbie Anderson said. “Mom said more than once they cleared the floor” at the Biltmore Hotel, a popular Dayton nightspot at the time. But by spring of 1943, a night out with his wife had become a rarity: Desch was captive to the top secret work inside Building 26.

  Despite the endless hours of work both in Dayton and Washington, the Americans still had not made their hoped-for decisive contribution to cracking the U-boat codes. That was evident in the round of intelligence negotiations between Britain and the United States in May 1943. The U-boat onslaught in the spring and the delays in both Bombe programs strained the GCCS-OP20G relationship. So had demands by America’s Admiral Ernest J. King that his ships and planes be allowed to attack the special refueler U-boats, called “milk cows” (Milchkühe), even if the pinpoint attacks might raise German suspicions that the Enigma was being read.