The Secret in Building 26

by Jim DeBrosse

  And so it went during one of the darkest months in the Battle of the Atlantic, when ninety-five merchant ships, hundreds of seamen, and 567,000 tons of Allied shipping went to the bottom of the Atlantic and Arctic oceans. In ten days alone that March, more than forty ships were sunk by U-boats. The battle seemed to be turning in favor of the Nazis, threatening the lifeline between America and En-gland and, ultimately, the outcome of the war. The loss of tonnage wasn’t some abstract number to be tallied and forgotten. It meant less fuel, less food, and more deprivation for the British fighting on the home front and fewer armaments and supplies for the Russians driv-ing the Nazis from Kharkov and the Allied troops trying to outfox Rommel in North Africa. Perhaps even more crucial, it meant delaying the all-important buildup to D Day and the liberation of Europe. As a U.S. Navy training manual pointed out at the time:

  If a submarine sinks two 6,000-ton ships and one 3,000-ton tanker, here is a typical account of what we have totally lost: 42 tanks, 8 six-inch howitzers, 88 twenty-five-pound guns, 40 two-pound guns, 24 armored cars, 50 Bren carriers, 5,210 tons of ammunition, 600 rifles, 428 tons of tank supplies, 2,000 tons of stores and 1,000 tanks of gasoline. . . . In order to knock out the same amount of equipment by air bombing (if all this material were on land and normally dispersed) the enemy would have to make three thousand successful bombing sorties.

  On March 10, 1943, the same day the Gorgas was torpedoed and sunk, the German Navy had switched to a new set of codes for its weather signals, breaking Britain’s tenuous hold on reading U-boat messages. Although the weather signals themselves did not reveal the location of the U-boats, the lists of codes and their plain-text translations had provided British codebreakers an important wedge in cracking U-boat radio signals and predicting where the wolf packs would strike next. The codebreaking blackout meant the U-boats could strike with a vengeance—without the Allies listening in. In the thirty-two-million-square-mile expanse of the Atlantic, finding a lone submarine with no clue to its location or course was akin to hunting for a roadside mailbox in all of Texas, without an address.

  Of course, the U-boats faced nearly the same challenge in finding the Allied convoys in their long voyages across the Atlantic. But unknown to the Allies, the Germans were reading the convoy ciphers with relative ease from 1941 through nearly all of 1943. With the help of Germany’s very competent radio-intelligence unit, known as B-Dienst, U-boat Control knew where to align its submarines along convoy routes, waiting for the kill.

  Grossadmiral Karl Dönitz, now supreme commander of the German Navy, kept in close radio contact with his skippers, requiring them to report their positions and fuel supplies with every communication. Once a convoy was located, Dönitz gathered the available U-boats into wolf packs to maximize their killing power and their defensive strength. The frequent radio contacts between headquarters and the U-boats gave the Allies ample opportunity to intercept and to home in on the signals with the latest direction-finding equipment. But direction finding was simply that: it could show the Allies what path a U-boat had been following but not its destination. To determine where the subs were going and why—whether to attack, regroup, refuel, or head home—the Allies needed to read the content of the radio messages, and that was a far more daunting challenge. In transmitting its signals, the Kriegsmarine made optimal use of the Nazis’ remarkable encryption machine, the Enigma. In simplest terms, the Enigma put German radio messages into an enciphered form of seemingly random letters. An Enigma machine on the receiver’s end, set up in exactly the same way, returned the cipher to plain text.

  The Germans were confident that their coding system was impenetrable—even if their machines were captured—and they had good reason to be. Theoretically, at least, the number of ciphering possibilities generated by the advanced naval Enigma of 1942 was far greater than the number of all the atoms in the observable universe—an incredible 2.0 × 10145 possible encryption settings compared to a mere 1.0 × 1080 known atoms. After purchasing the rights to the commercial version of the Enigma Cipher Machine in the 1920s, the German military began adding its own bedeviling improvements and innovations and continued to do so throughout most of the war. But what the Germans did not count on was that lapses in their own security and procedures would greatly reduce the number of unknown possibilities and bring their seemingly unconquerable cipher system within the grasp of adversaries who were far more clever than the Germans realized.

  First conceived of by Arthur Scherbius in 1922, the Enigma was a marvel of the machine age. Looking much like a portable typewriter with its own wooden carrying case, it scrambled each keystroke of a message through a series of tumbling rotors that turned the result into seemingly random gibberish. When a clerk pressed a letter on its keyboard, the corresponding cipher-text letter lit up on a display panel at the top of the machine and was written down by a second clerk. Once the entire message had been scrambled into cipher text, it was radioed by Morse code from the submarine to headquarters or vice versa.

  Pressing down a key advanced the first rotor forward one letter and sent an electric current through the wiring of the naval Enigma’s four rotors. All but the last rotor were joined to a ring with a tumbler pin on it that, in turn, advanced the next rotor. So, after the first rotor moved through all twenty-six letters of the alphabet, it would tumble the second rotor and, after the second rotor advanced twenty-six letters, it would tumble the third, much like the mileage numbers on the odometer of a car. As a result, the naval Enigma would never repeat the same position of its scrambling wheels for hundreds of thousands of keystrokes (26 × 26 × 26 × 26 = 456,976; however, a complex feature of the Enigma, a double turnover of one of the rotors, reduced the number of positions by some 18,000).

  The fourth and final disk, called the Umkehrwalze, or reflector, served to reroute the current back through the first three wheels, ensuring that the plain text and the scrambled text were always reciprocal at the same settings. That way, the clerk receiving a ciphered message could set up the machine the same way, then type in the scrambled text and automatically reproduce the original plain text. The German military liked the economy and simplicity of the Enigma’s operation: one machine could do both tasks of enciphering and deciphering messages.

  The ever-changing wheel positions of the Enigma were only the beginning of a codebreaker’s nightmare. Each rotor was wired to scramble the letters in a different way, and all the rotors but the fourth could be ordered differently within the machine, from left to right. The different wirings for the rotors and the different orders for placing each rotor inside the machine created another staggering set of enciphering possibilities—26.0 × 10105—again, far greater than all the atoms in the known universe. Fortunately for the Allies, the capture of several Enigma wheels and a machine during the early part of the war revealed the wiring inside the rotors and vastly reduced the number of possibilities.

  But the hurdles for codebreakers didn’t stop there. Each of the four rotors could be set manually to a different starting position, at letter A and on through Z, with the starting letters visible through small windows on the Enigma casing. The total number of possible initial settings for the four rotors was, again, 456,976. Likewise, the tumbler rings on each of the first three rotors, which caused the adjacent rotor to advance, could also be set to different starting positions, multiplying the possibilities by yet another 456,976.

  If all that weren’t enough to drive codebreakers mad, the Germans added what looked like a tiny telephone switchboard to the front of the machine, on which double-ended wires, called Steckers in German, could be plugged into jacks for swapping individual letters, so that A became E, for instance, and vice versa. Anywhere from one to thirteen different letters could be steckered in this way. The stecker board proved to be one of the biggest headaches in breaking the Enigma. With twenty letters steckered to one another and six letters unplugged, the standard practice during the war, the number of plugboard possibilities alone was nearly 533 trillion.
<
br />   In the blind man’s game of finding and sinking enemy vessels at sea, the German Navy realized the importance of making their Enigma operating procedures no less secure than the machine itself. While the German Air Force and Army allowed their cipher clerks to choose at random the starting positions of the rotors—which were then encrypted and sent along with the message to the receiving clerk—the German Navy didn’t trust such potentially sloppy methods, and wisely so. The laziness of some Army and Air Force clerks led to repeated and predictable starting positions, such as HIT (the first three letters of Hitler) or, in the case of one love-struck cipher clerk, the frequent use of CIL, which happened to be the first three letters of his girlfriend’s name, Cilla. British codebreakers, never short on humor, dubbed that whole class of starting position giveaways “Cillies” or “psillies.”

  In the German Navy, the Enigma operators were required to select the starting positions from a book containing long lists of randomly generated letter pairs, then to convert those letter pairs to a second set of letter pairs for each starting position of the wheels. These bigrams, or doubly disguised letters, were then sent at the beginning and end of each Enigma message as an indicator for the receiving clerk. The clerk deciphered the indicator from his own codebook and set his Enigma to the same starting position.

  In fact, the entire sequence of steps for setting up the advanced U-boat Enigma was dictated by the Kriegsmarine high command. Before sending a U-boat message on a given day, the Enigma operator checked his current list of “keys” from headquarters. Going by the list, he selected three wheels from a box of eight, plus the special fourth rotor from a different store of components offering four choices. He placed all four rotors in the proper order within the machine along with the three tumbler rings, then set the starting positions of the wheels and rings, and, finally, plugged in the steckers to the correct letter sockets.

  Breaking into the Enigma was like trying to open a series of locked doors. Each door could have thousands or even trillions upon trillions of possible keys, but only one would open it. And not until all the doors were opened could the message be read. Luckily, some of those doors had been pried open through the capture of Enigma components and operating instructions. From experience, the British knew that the toughest keys to find were the selection of wheels, the wheel order inside the machine, and the steckers. Once those were found, it was relatively simple to find the ring settings and the Grundstellung—that is, the starting positions of the wheels. And once the grund for the day was identified, it was easy to read the rest of that day’s messages.

  From the 1920s, when the German military adopted the Enigma, to the end of World War II, there was a never-ending search for methods and special machines to attack the cipher. By 1943, the Allies were building some of the most complex machines ever devised to try to conquer the Enigma and other Axis code systems.

  Since the mid-1970s, much attention has been given to the British successes in cracking the Nazi Enigma and, more recently, to the early Polish breakthrough that laid the groundwork for the British. The Poles called their first decoding machine a “Bombe,” perhaps after the brand of ice-cream cones favored by the codebreakers or perhaps because a loose chunk of metal dropped to the floor whenever the machine arrived at a solution. Just before Poland fell to the Nazi blitzkrieg in 1939, the Poles shared their secret success with both the British and the French, but it was the British who capitalized on this remarkable feat. Operating from the top secret Government Code and Cypher School (GCCS) at Bletchley Park, outside of London, British codebreakers refined and further mechanized the device, based on the advanced work of mathematicians Alan Turing and Gordon Welchman. Their machine, which first appeared in 1940, was also dubbed a Bombe (pronounced bomb), perhaps as a nod to the contribution of the Poles or, others say, because of its menacing ticking sound.

  In the broadest terms, the British Bombe worked like a series of Enigma machines in reverse. Spinning commutator wheels were wired to simulate the rotors on the Enigma, with twenty-six electrical contacts running past twenty-six metal sensing brushes for tracking the possible letter pairings of cipher and plain text. To do their work, the Bombes needed special clues the British called “cribs.” A crib was a word or phrase that codebreakers suspected as the underlying plain text matching a stretch of nonsense letters in a message. The codebreakers had to not only surmise the underlying plain text but be able to pinpoint its exact location, letter by letter, within the ciphered message. For example, if the suspected crib was HEIL HITLER and the corresponding Enigma cipher was ZQAT KFBCMJ, the plain-cipher pairings would H-Z, E-Q, I-A, and so on.

  Having been fed a crib and told which sets of Enigma wheels to test, the machine crunched through hundreds of thousands of possible wheel and plugboard settings—and, more important, eliminated those settings that were inconsistent with the design of the Enigma—until it arrived at a setting that may have generated the matching cipher text. Once the rotor positions and the plugboard settings were known, it took just a few more simple steps before the scrambled text could be read by running it through a captured Enigma machine or its analog.

  The British, with the help of information from Poland’s codebreakers and the brilliance of its own cryptanalysts at Bletchley Park, made inroads on a variety of Axis codes and machines, including the Enigma, by 1940. Winston Churchill’s pet code name for the top secret operations at Bletchley Park was “Boniface,” now better known publicly as “Ultra.” Although Britain never attained a fully secure grip on any of the German code systems, it had learned to handle with reasonable success the three-wheel Enigma machine by 1941.

  But when the German Navy switched to a four-wheel Enigma in February 1942, the challenge facing the British was overwhelming. The new fourth wheel still acted as a reflector and was stationary, but its daunting new feature was that it could now be set to start at any letter of the alphabet, just like the first three rotors of the machine. While the German Army and Air Force continued to use three-wheel machines, the new naval system, called Triton by the Nazis and M4 Shark by the British, added one more door of protection to the U-boat messages. Shark increased the number of possible wheel orders to be tested on the Bombes from sixty to more than three hundred. The hurdle was too much for Britain’s twenty-some hard-pressed Bombes, which had been designed to attack a three-rotor Enigma, and was far beyond the reach of its codebreakers’ hand methods.

  From early 1942 until December of that year, when the capture of new Enigma materials came to their aid, British codebreakers were almost totally shut out from the U-boat radio traffic. No wonder that North Atlantic sinkings more than tripled in the last half of 1942 compared to the last half of 1941—from six hundred thousand tons to two million tons. Even when codebreakers finally managed to read U-boat messages again in late 1942, the decoding took days and sometimes weeks, and the information often came too late to be of use in tracking down U-boats or shifting convoys away from wolf packs.

  To grapple with Shark, the British had begun work on a new high-speed, four-wheel Bombe not long after they learned from messages in late 1941 that the Germans were going to introduce the new Enigma. Progress was slow and, even as late as the spring of 1942, the British felt they had little hope for success against the trillions of new possible wheel orders.

  But by this time the British weren’t alone in their struggle. After a misguided start on the Enigma problem in late 1940 and a visit by American codebreakers to Bletchley Park in early 1941, the U.S. Navy began work on a top secret program in early 1942 to develop a Bombe of its own. It gathered its own Bletchley Park of bright young engineers and theoreticians, recruited from some of the nation’s best colleges.

  Under some of the most stringent security measures of the war, the codebreaking machine would be designed and produced in Dayton, Ohio, at the National Cash Register Company, then a world leader in electronics. Only two companies in the United States had the technical capability at that time to produce such a ma
rvel: IBM and Dayton’s NCR. The obvious choice for the Navy at that time was NCR, where chairman of the board Colonel Edward A. Deeds had a long working relationship with Vannevar Bush, the famed MIT scientist who had forged an alliance between military and academic researchers through his National Defense Research Committee. Even more propitious, NCR had at its disposal eleven city blocks of mostly idle factories and office buildings and a regional network of skilled labor and parts suppliers, all waiting to be put to work. NCR, unlike IBM, had been ordered by the War Production Board to stop making its chief product, cash registers, to conserve much-needed materials. The project would be entirely self-contained inside NCR’s Building 26, so named because it was the twenty-sixth structure to rise on the company’s ninety-acre business campus. It was only a coincidence, noted by researchers at the time, that the work in Building 26 would revolve around the seemingly endless possibilities in scrambling the twenty-six letters of the alphabet.

  Like the British, the Americans at first put their faith in the rapidly emerging field of computational electronics—precursors to modern computers—as the best and perhaps only hope against the nearly impenetrable possibilities of the Shark Enigma. The initial aim of the four-million-dollar Navy project was lofty: to construct a state-of-the-art electronic decrypting machine fast and smart enough to crack Shark in a matter of hours, not days or weeks. The program at NCR would be equal in priority to the Manhattan Project, aimed at developing the atom bomb, and perhaps second only to it in hastening an end to the war. Because of the wartime shortage of men, the project would recruit some six hundred young women through the Navy’s WAVES auxiliary to perform the delicate handwork in assembling the machines. None of the women was to learn the vital nature of their work until fifty years after the war.

  Although the official U.S. naval histories have exaggerated the contribution of Ultra and the Bombes in subduing the German U-boat threat, the latest data do show that, by drastically reducing the time needed to decrypt Shark messages, the Bombes helped find and predict U-boat locations in one of every four U-boat sinkings from 1943 until the end of the war. Harder to gauge is the extent to which those pinpoint strikes disrupted the German wolf packs and kept them at bay, but they were certainly an Allied advantage in the Battle of the Atlantic.