Months must pass, however, before the redesigned bombes, with all their thousands of soldered connections, would be available. How were the codebreakers to achieve at least partial success in the meantime?
British patience and meticulous attention to detail came to the rescue. GC&CS analysts had been studying the habits of German Enigma operators and had found two subtle mistakes that could be exploited.
The first became known as "Herivel's tip," after John Herivel, a young mathematician recruited by Welchman. Much like Rejewski, Herivel tried to put himself into the shoes of a German code clerk and imagine what the operator might do incorrectly because of laziness or work pressure. Herivel had an insight. At the beginning of each new encoding day, the German operator had a boring series of steps he had to go through. Following instructions, he must choose the correct set of three rotors out of the five available, slide the rotors in proper sequence onto the axle, turn their alphabet rings to the required positions and link up the proper arrangement of the plugboard cables. Then he was supposed to select three random letters for his message key. It was all a big bother. Suppose, Herivel asked himself, the lazy or hurried operator didn't take that final step? Suppose he sent his first message of the day using the same three letters as his rotor ring settings? Herivel suggested collecting the new day's first messages. If there was more than one shortcutting operator, there would be repeats—and the rotor settings could be surmised.
The second sloppy practice consisted of what BP labeled "cillis." The name may have been derived from the initials of one German clerk's girlfriend, which he used often instead of randomizing his three-letter selections. That was one type of cilli—the repeated use of familiar sequences, such as HIT and LER. Another form was supplied by German operators who, instead of plucking their three letters out of the air, simply lifted them from their keyboards. A sequence down from the Q key read QAY. One down from PFread WSX. Although these practices were expressly for bidden in the Enigma operators' manuals, lazy or rushed code clerks did resort to them, and from these cillis BP's clever analysts could determine the wheel order for the day as well as the setting for these particular messages. "Unbelievable?" Welchman wrote. "Yet it actually happened, and it went on happening until the bombes came, many months later."
Using these and similar ingenious methods, the BP crew early in 1940 began deciphering the Luftwaffe messages known as Red because that was the color of the pencil Welchman used to demark it from other systems.
By then it had been decided that Welchman and Turing would divide the main Enigma decrypting responsibilities between them. Welchman had moved into Hut 6, one of the wooden structures hastily erected on the park's grounds, and took over its operation when the young John Jeffreys became terminally ill. Welchman's team concerned itself with breaking German air force and army traffic, then passed the decrypts on to Hut 3. There, another team translated them, judged their importance and urgency and determined where they should be disseminated. Turing was responsible for Hut 8, heading up work on the naval Enigma signals, with Hut 4 as his analysis center.
When the bombes arrived in August 1940, allowing cribs to be put to use, the Hut 6 team simply accelerated the breaking of the Red cipher. It was of particular value because it was used in army/air force coordination and disclosed information about both services.
German Enigma operators continued their inadvertent cooperation. Each month, for example, the operators had to create new sheets covering the next month's ring settings, rotor orders and plugboard connections. One German operator decided he could save himself much work by simply rearranging blocks of settings from previous months' key sheets. Having broken those of the previous months, BP could quickly break the new settings.
In North Africa's Qattara Depression, a bored German officer reported every day the same message, "Nichts zu melden"—"Nothing new to report"—giving Welchman and his team a ready-made crib for solving the new Enigma setup.
The Britons' attention to detail steadily paid dividends. In her self-published book, England Needs You, about her life at Britain's Beaumanor intercept station, Joan Nicholls has told of two Germans, either friends or relatives, who served as code clerks at different stations in a panzer division network. They would end each of their messages to each other with the smallest of flourishes. One would sign off with "—••—", or X; the other would answer with "— —", M. It was, Nicholls wrote, "such a small transgression on their part, but we were able to log these two and, of course, the whereabouts of their unit, from North Africa through Sicily, Italy, France and finally into Germany."
The bombes quickly proved their worth. The Wren operators of a bombe would receive a "menu" of settings phoned to them from BP and set the rotors accordingly. The bombe would whir through its rounds, testing a Luftwaffe message. When it came to a Stop, this indicated that all the links on a menu were confirmed. From a successful Stop, the GC&CS team could determine the order in which the three rotors had been placed as well as the three letters which were the settings for the alphabet rings on the rotors. One of the Wrens would check out the settings on her replica Enigma. If German text appeared, and a member of the overseeing watch approved, the exultant cry would ring out: "Red's up!" The operators would then await their next menu.
Mechanically, the bombes were a bit tetchy. The stiff wire brushes on each rotor that connected the two sets of contacts could, in use, widen and cause the machine to short out. Before each run the Wrens applied tweezers to reposition each wire correctly. RAF mechanics also appeared, either regularly or on emergency call, to repair the bombes. With these measures of correction, the machines operated around the clock, week after week, month after month. With the Tabulating Machine Company mass-producing new units, several bombes could be set to work on the same menu, greatly reducing the time required for achieving a productive Stop.
It was well that Hut 6 was breaking Red because over in Hut 8 Turing was having a far harder time. He was facing a much more formidable opponent. While Welchman was dealing with Hermann Göring, who was as lax with his Enigma systems as he was with his personal fitness, Turing had to match wits with Admiral Karl Dönitz, the much more rigorous leader of the U-boat command. The German navy drilled its Enigma operators in following strict security procedures, kept introducing improvements into its systems and made regular changes in key methods. As an example of Turing's difficulties, the navy Enigma gave operators a choice of three out of eight rotors, while other German clerks selected three out of only five—a difference that put much greater barriers in the way of the cryptanalyst.
Before Turing eventually triumphed over the naval Enigma, a story told in a later chapter, he and his team experienced a harrowing time. Every day of continuing failure conjured up new visions of men trapped in their sinking vessels or escaping only to freeze to death in the North Atlantic's frigid waters; of desperately needed food, gasoline and other supplies plunging uselessly to the ocean floor; of Britain's ability to survive hanging in the balance.
Welchman made another significant contribution to the early development of Bletchley Park. He looked ahead and realized how inadequately Dilly Knox was preparing for the intelligence war that was to come. The small staff Knox had gathered would be inundated, Welchman foresaw, once the Enigma was conquered and masses of German intercepts began pouring in. Welchman developed an organization plan calling for a major expansion of the forces at BP. With his plan quickly approved, he became needed as much for his leadership and administrative skills as for his codebreaking abilities.
BP'S Triumph over "Fish"
In the autumn of 1940, BP's analysts were faced with a cryptologic challenge decidedly different from that presented by the Enigma. This new traffic was based not on Morse code but on the international code—the Baudot-Murray code—developed for teleprinter machines. Instead of dots and dashes, Baudot-Murray employed electrical impulses triggered by holes and spaces in paper tapes. Unlike the Morse system, whose characters varied in length f
rom one dot (letter E) to five dashes (numeral 0), each symbol in the Baudot code was represented by a group of five equal-length hole (x) or space (o) elements. Thus, in the nonsecret international system, A was xxooo, B was xooxx, and so forth. It was, in short, another forerunner of digital computers' binary system.
The Germans, it became clear, were adding a second machine to the standard Teletype machine to encipher its output. Messages transmitted by radio were most often sent automatically at high speed and were direction beamed so that interception was more difficult. In addition, each of the three German armed services was using a different cipher machine. Because the Germans called one of their systems Sägefisch, or "Sawfish," BP chose "Fish" as shorthand for this whole separate type of transmission.
How was BP to cope with the immense new problems presented by Fish? The question took on greater urgency as it became evident that after intermittent use at the beginning, the Germans were relying on Fish ever more heavily. BP also determined that the system was being used for higher-level communications, such as those between army commands, or from headquarters to commanders in the field. Further, the messages tended to be much longer, running to thousands of characters, compared to the few hundred typical of Enigma traffic. And there was the threat that the Teletype code machine might in time completely replace the Enigma. Harry Hinsley has written that Fish represented "intellectual, technological and organisational problems of a still higher order than those presented by the Enigma."
But BP had triumphed over Enigma. They were confident they could do the same with Fish.
Britain's top intelligence authorities came to a decision. To tackle three different types of Fish encoding machines would require too great a commitment of resources. Intelligence directed that the effort be concentrated on the cipher machine used by the army, subsequently known to be produced by the Lorenz firm. A greater flow of information about the Wehrmacht was, at that point, the prime need of Allied war planners. Carrying on the piscatorial analogy, BP called the Lorenz machine Tunny.
Hinsley was right: Tunny made a formidable opponent. In addition to two drive wheels, it used ten rotors, and none of them sat idle; they all worked together in linked sequences. The rotors were different from those of the Enigma. Instead of being wired internally, each enciphering rotor had around its rim a number of spring-powered pins that could be either retracted or extended to form either a hole or a space. To encode a message, the machine applied an additive system invented in 1918 by an American, Gilbert Vernam. The sender's machine automatically added to each plaintext letter a random letter, resulting in still a third letter which was sent over the air. The receiver's machine automatically canceled the additive, leaving the original character to be printed out.
BP's John Tiltman, a man who had entered Oxford at the age of thirteen, set his mind to unraveling Tunny messages by hand methods. He was making only slow progress until, once again, a German operator error opened the door. The operator had a long message of nearly four thousand characters to be sent from his high command post to another. The operator set up his Lorenz machine correctly and sent an indicator so the receiver could set up his machine. Yet trouble developed. After the sender had patiently typed out the long text and transmitted it, the receiver radioed back that he hadn't got it, so please send it again. The two of them took the absolutely forbidden tack of turning their machines back to the same initial settings of the rotors. The sender then committed a second mistake. Probably bored by having to repeat the message, he began to take shortcuts. The first word was Spruchnummer—"message number." He abbreviated it to Spruchnr. With similar cuts, the second message came out about five hundred characters shorter than the first. If the two messages had been identical, they would have been no help to Tiltman. But by crossruffing the streams of dissimilar ciphertexts against each other, he was able to recover both messages completely. What was more, he found the elements that had been added by the Lorenz.
Studying this information, another young Cambridge graduate, William T. Tutte, realized that certain patterns tended to repeat after forty-one bits. From this he deduced that the first rotor had forty-one pins. In four months of intense concentration, Tutte worked out the machine's complete internal structure. His discoveries resulted in the building of a simulated Tunny. Later, when the capture of a German unit permitted comparison, the BP team exulted, "We got it right."
Throughout 1942, work on the Lorenz intercepts had to be done by hand methods. The decrypts yielded some useful information, even though many of them were weeks old before the cryptanalysts could deliver them.
Max Newman, who had stirred the mathematical imagination of young Alan Turing at Cambridge, also entered the picture. Seeing that decryption of Fish traffic would remain of limited value until the process was mechanized, he proposed a device that would go beyond the strictly electromechanical functioning of Turing's bombes; it would make use of the emerging technology of electronics by including vacuum tubes, which worked much more swiftly than electrical switches.
Newman's machine resembled an eight-foot-high cupboard. The women working on it dubbed it "Heath Robinson," the name of the cartoonist who, in Britain, created the same sort of absurdly overcomplicated devices that Rube Goldberg did in America.
Comical or not, Heath Robinson could scan one thousand telegraphic symbols a second. It was designed to keep two paper teleprinter tapes in synchronization at thirty miles per hour. One tape carried the enciphered text of a message; the other contained the wheel patterns worked out by the codebreakers. The machine's comparison of the two tapes determined the settings of the five most active rotors and greatly speeded up the process of decryption.
It was a promising start, but Heath Robinson soon showed severe weaknesses. It caused the sprocket holes in the tapes to stretch, ruining the synchronization, or it completely broke the tapes. It sometimes became so hot that it began to smoke.
Turing, at that point in early 1943 a member of the Fish team, suggested that Newman call on the services of Tom Flowers, who had worked at Britain's Post Office Research Station, and who had developed postal equipment that used vacuum tubes. After the years of secrecy had ended, Flowers recalled about the Heath Robinson machine, "I was brought in to make it work, but I very soon came to the conclusion that it would never work. It was dependent on paper tape being driven at high speeds by means of spiked wheels, and the paper couldn't stand up to it."
He saw that combating the Lorenz required a fully electronic machine, one that would use as many as fifteen hundred vacuum tubes. The machine he envisioned would do a job much faster because it would not require the synchronization of two tapes. Its one tape would carry the enciphered message and would be read photoelectrically. To forestall the tape's tearing itself apart, it would run on smooth-surfaced wheels. Instead of running on a second tape, the wheel patterns would be generated electronically. He was confident his machine could process five thousand characters a second and could thus spin through all the possible combinations of keys for the Lorenz in about an hour.
The authorities at BP, having to husband their funds, thought Flowers's plans were too big and impractical. The vacuum tubes were unreliable; they burned out too quickly. Besides, since this was already the spring of 1943, the war would be over before he could possibly produce a machine.
Flowers and his Post Office colleagues were not to be dissuaded. They would build the machine without BP support. Flowers found that if vacuum tubes weren't turned on and off, if they were just allowed to run on and on, they were quite reliable. As for the completion date, he and his team, working on their own, produced Colossus I in ten months.
The results were dramatic. Colossus reduced the time required to decrypt Fish messages from weeks to hours. The timing was also dramatic. The machine came into use soon enough to provide invaluable information for Eisenhower's D-Day planning. Its decrypts helped prove that the Germans were swallowing the Allies' deception programs for the invasion. During the remainder of
the war, it kept Allied leaders informed of the decisions of the German high command.
In volume, Fish decrypts numbered far less than the monthly averages of ninety thousand decrypts from Enigma traffic in the war's final phases. But the Fish messages were longer, some of them running to ten thousand characters, and they provided an ear to the most intimate planning of Hitler and his top generals, their discussions, orders and reports on the disposition and strengths of their commands.
As suppliers of intelligence, Enigma and Fish complemented each other. While Enigma decrypts most often revealed information of operational and tactical value, Fish supplied knowledge of strategic importance.
As the war progressed, the Germans relied more and more heavily on their Fish communications. These networks proliferated, rising from six links in July 1943 to twenty-six in early 1944. The Germans also kept changing and improving their systems. But Flowers and his men stayed with them, building ever more powerful Colossi. With BP approval, they had twelve machines working by the end of the war. Their Colossus II used twenty-five hundred vacuum tubes. George Vergine, an American who worked with the Colossi at BP, later recalled having one of the suppliers at the vacuum tube manufacturing plant ask him, "What the hell are you doing with these things—shooting them at the Jerries?"
Hinsley called the availability of Fish decrypts "the outstanding signals intelligence achievement in this last phase of the war."
Unknowingly, the engineers who produced Colossi were reaching another milestone. Their machines were forebears of the modern digital computer, a success that should have been attributed to British science but that had to be suppressed by the long postwar secrecy imposed on all things cryptographic. On orders from Churchill after the war, the Colossi were destroyed and Tom Flowers burned the blueprints.
Codebreakers Victory Page 6
