Day 7: "A hit, a miss, then a parade of successes"
WAVES tend to bombe operations at Navy codebreaking headquarters in Washington, DC.
By Jim DeBrosse
Dayton Daily News (OH) - Saturday, March 3, 2001
MAY 24, 1943
IN A SECURED INNER ROOM in NCR's Building 26, not long after lunch, Phil Bochicchio was conducting a test run on "Adam" - one of two prototype Bombes - when its high-pitched whine suddenly died and the machine shut off. At first, he thought it was just one more electrical short in the temperamental device.
But then, as it was designed to do, the Bombe came back to life and began slowly to rewind. That could mean only one thing: It had scored a "hit" on part of an Enigma message fed into the machine.
When a hit occurred, a complete circuit surged through all 64 of the Bombe's fast-spinning "commutator" wheels - possible only if it had arrived at the positions of all four rotors on the Enigma machine that may have created the message. The surge set off a braking mechanism that stopped the Bombe's wheels within four to five revolutions, then rewound them back to their hit positions and printed out the rotor sequence.
That's exactly what happened, but Bochicchio, the floor manager on the project, still was skeptical. He had a fellow engineer run the same encrypted message through "Eve," the other prototype. It shut off in precisely the same rotor positions.
Bochicchio gave the print-out to Lt. Cmdr. Ralph Meader, the Navy official in charge of NCR's massive Nazi-codebreaking effort. Meader wired the results to Naval intelligence in Washington, D.C. Codebreakers there could hardly believe their good fortune. The rotor positions enabled them to unscramble a series of intercepted German radio messages during the next two weeks, revealing that the Germans were moving 17 submarines to attack westbound Allied Convoy GUS-7A. The convoy's escort carrier, the USS Bogue, sent its pilots to hunt down and scatter the wolfpack - sinking one sub and damaging three others.
Bochicchio wouldn't learn until June 5 how important that inaugural hit was, when Howard Engstrom, head of the Navy's machine decoding section, sent an encrypted teletype message to NCR informing the engineers that the May 24 hit alone "had paid for the entire cost of the NCR project."
But there was no time to celebrate. Even at this late stage of development, the NCR engineers were struggling with the machine's most basic part. "Those large rotors seemed destined to overheat, lose their shape and create faulty electrical signals," historian Colin Burke wrote in Information and Secrecy. "The metal sensing brushes also seemed to have been meant only for slow and dependable tabulators. As quick-fixes were made to those parts, more oil leaks developed on the prototypes. Those problems raised such fears about the production model's design that assembly was halted. The situation was so grave that all message processing at Dayton was suspended."
And the news would grow worse for Joe Desch and his team before it got better.
JUNE 18, 1943
The order from Naval intelligence seemed unreal. The Navy wanted a whole new Bombe design, with an automatic method of switching rotors and greater machine speeds - and with no delays in production.
Desch was stunned. In effect, the Navy wanted a machine approaching the complexity of a modern computer.
Leading theorists knew even then that electronic digital processing would become the basis for modern computers, but more practical types like Desch, who knew the limits of industrial production, also realized that it would take years to put such a machine into operation.
Desch had been right. "There was no practical, high-speed substitute for the many hard-wired rotor wheels that could be rearranged on the Bombe's drive spindles," Burke wrote. "Thousands of tubes in very dense circuits would be needed to imitate all possible rotor wiring on the 64 wheels on a Bombe."
Desch also realized that the miniature electronic tubes at the time, unlike today's microchips, simply weren't reliable enough to do the job.
He protested to Meader, who usually went along with whatever his naval superiors wanted. This time, Meader supported Desch. He dashed off a rejection letter, telling the Navy brass "that too much money had already been invested, too much precious material had been used and that the first model was too close to production to be thrown aside," Burke wrote.
The Navy had little choice but to renew its faith in Desch and hope that his 100 machines, when they finally reached Washington, would do the job.
By the last week of July, 15 Bombes had been assembled at NCR - but none would work properly for long.
It was perhaps Desch 's darkest hour. The Navy was ready to scrap the project: a year of intensive work and millions of dollars were about to be declared a waste. But Desch relied again on his practical bent to save the day. "At the very last minute, he made a discovery that revived hope," Burke wrote. "Running the Bombe's Bakelite code wheels at extreme speeds was again causing invisible distortions, leading to false electrical contacts. Desch predicted that careful storage, handling and refurbishing would solve the problem."
Again, the Navy trusted in his judgment, and the project was spared.
Although no one knew for certain that the NCR Bombes would work in the long run, Desch 's optimism - and a mounting fear of the Nazi submarine wolfpacks - led Meader to ship the first of 120 machines to Washington before testing was complete.
A covered annex was added from the rear entrance of Building 26 to the railroad tracks, where the machines were loaded on flatcars in the dead of night. In Washington, they were unloaded at a top-secret complex on Nebraska Avenue near Tenley Square, once the campus of a prestigious all-girls' school. Inside deceptively quaint old dormitories of ivy-covered brick, the Navy ran the massive codebreaking effort it had taken over from the British.
Many of the WAVES based in Dayton, including Catherine Racz, went along as operators. Racz was one of the four-person teams assigned to each of the 120 machines, lined up in long, noisy rows in Building 4. The teams got "menus" from Naval intelligence, with beginning rotor positions and plugboard settings and a string of encrypted message whose plain text either was known or suspected. Each Bombe was the equivalent of running 16 Enigma machines in reverse at extremely high speeds. The more urgent a message, the more Bombes were assigned to crack it.
Whenever a machine scored a hit, "a bell went off, and all the big shots showed up" to retrieve the print-out of the machine's wheel positions, Racz recalled. "Evidently it was something important. But we didn't even know then what we were doing. We just did what we were told."
Despite the earlier problems with the machines, they were eventually tweaked to the point of reliability, said Bochicchio, who set up the Bombes once they reached Washington. All the machines continued to leak small amounts of oil, he said, but by using thicker lubricant and reducing the oil pressure inside the machines, the problem was minimized.
"We knew we could do it," Bochicchio said. "You try one way, then another. It just takes time."
By December, when all 120 machines were operating, the Desch Bombe was proving itself worthy of the Navy's investment. Consider that in June 1943, when the first machines were under development, it took an average of 600 hours to decrypt an Enigma naval message. From December 1943 until the end of the war, when all 120 machines came on line, it took an average of just 18 hours.
"As important," Burke wrote, " Desch 's Bombes proved to be very, very reliable. After their first shakedown runs, they could be used 24 hours a day."
Even though the final product wasn't as fast as had been hoped - the speed of the Bombe's main wheel had to be slowed from 3,600 rpms to 1,725 rpms to keep it from warping and breaking apart - the Desch design sported several striking innovations, including "a digital electronic tracking and control system that astounded the Navy's engineers," according to Burke.
Desch 's automatic braking and reverse feature, though pooh-poohed by British codebreaking pioneer Alan Turing, worked reliably and enabled the operator to restart the search for consistencies within the message without delay once a hit had been made.
"Despite all the false starts, delays and problems," Burke wrote, " Desch had built one of the most complex machines in the world."
The NCR machine was 200 times faster than the original Polish Bomba and at least 20 times faster than the Turing Bombe. The Desch design ran 30 percent faster than even Britain's 1943 version of a four-wheel Bombe, with a number of other advantages, including the ability to run more than one test per run.
The Brits at last recognized their effectiveness, as well as the abilities of U.S. Navy codebreakers, and handed over most of the deciphering operations in the Atlantic to the Americans.
Historians differ over how much the Bombe shortened the war. But by even the most conservative estimates, the Ultra codebreaking effort, including the NCR Bombe, lopped at least a year off the fighting, sparing hundreds of thousands of lives on both sides of the conflict, Burke said.
Sunday: Bringing the truth to light