Codebreakers Victory Read online

  The Germans had good reason to believe their Enigmas were secure against cryptanalysis. Dr. Ray Miller, a computer scientist at the U.S. National Security Agency, has calculated the exact number of key settings faced by Enigma codebreakers. The possible permutations for the plugboards alone, he has determined, run to more than 500 trillion. And that was just one of the machine's five variable components. All the variables together multiply out to 3 x 10114. That number compares with only 1080 as the estimated number of atoms in the entire observable universe. "No wonder," he concluded, "the German cryptographers had confidence in their machine!"

  Did the Germans ever suspect that their enemies were reading their Enigma-encoded messages? At times when the Allies seemed to benefit from what seemed like amazing coincidences and incredible streaks of good fortune, questions were raised. Investigations were conducted. Always, at least until the very end of the war, the answers came back uniformly: the Enigma-based communications systems were inviolate. It was inconceivable that human minds could cope with such astronomical numbers of variations. If the Allies fared better than could be explained by brilliance or luck, the cause had to be secret agents, not penetration of the Enigma.

  Alternatives to the Enigma

  The first rotor code machine was developed by an American, Edward Ff. Hebern. He was also the inventor with the most grandiose ideas for his device.

  Hebern began with two electric typewriters connected by twenty-six wires in random fashion. Strike the key on the first and the other would type out its enciphered equivalent. But since the connections were fixed, decryption was too easy.

  He moved on to a rotor machine, filing his patent on it in 1921. This complex device used five rotors. He brought his machine to Washington and demonstrated it to the U.S. Navy's Code and Signals Section. Naval officials were "thrilled when he showed us what it could do." They seemed ready to commit to navywide use.

  This positive reaction was good enough for Hebern. Without having secured a signed order, he planned big. He sold a million dollars' worth of stock in his new company and built a substantial factory to produce his machines. The navy, however, was not accustomed to moving so swiftly. Not until 1923 did it convince a board to investigate the machine. In the end, it ordered only two machines for six hundred dollars each. Hebern's company filed for bankruptcy and he ended up in court, sued by his stockholders. Even though he later sold thirty-five machines to the navy, Hebern never succeeded in establishing a viable code machine business.

  In Stockholm another inventor resolved to try his luck with a rotor-type code machine. This was Arvid Damm, who already had patents for weaving looms. Although his code machines were clumsy constructions that invariably broke down during critical tests, he incorporated a company to market them. His smartest move was to add to his staff a young man named Boris Hagelin, son of a wealthy investor in the firm.

  One day in 1924, when Damm was in Paris, word came that the Swedish military was considering a mass purchase of Enigma machines. Hagelin resolved not to let that happen. He made quick changes to simplify the Damm model, making it more like an Enigma. The Swedish army placed a large order.

  At this point Damm died, and Hagelin took over. Under his direction, the company developed a very compact code machine, no larger than an attaché case. The French army gave him what seemed an impossible challenge: to produce a pocket-size machine. Not only that, they wanted a machine that could print out the ciphertext and could thus be operated by one man. Hagelin met the test. His clever little device used the same elements to handle both the deciphering and the printing. The French ordered five thousand of them.

  Hagelin came to the U.S. and got American cryptologic authorities interested. William Friedman made suggestions that he incorporated. In the end the U.S. Army adopted his machine as its field-use cryptographic system that offered at least midlevel security, and arranged with Hagelin to have the U.S. version mass-produced by a typewriter manufacturer. In World War II, GIs knew the not quite pocket-size but still very compact little wonder as the M-209.

  As David Kahn has said, Hagelin "became the first—and the only— man to become a millionaire from cryptology."

  Other Nations Make Their Choices

  For the U.S., the M-209 was considered suitable for tactical use, but not for the higher-level transmissions the enemy must never succeed in deciphering. To secure a machine promising full security, the military turned to Friedman. He developed an American equivalent of the Enigma that also applied the rotor principle. It was larger and more cumbersome than the Enigma but more secure. Friedman, for instance, had determined that the Enigma's "fast" rotor, the one that inched forward after 26 clicks of the keyboard, was most vulnerable to cryptanalysis when used as the right-hand entry rotor, so he avoided that placement in his machine. He received help from Frank Rowlett, who, seeing that Friedman's keying process was erratic and unreliable, came up with an electromechanical keying unit that worked much better. They received U.S. patents on their inventions. Their machine became known as Sigaba in the army and M-134 in the navy. It was never broken during the war.

  The British also studied the Enigma and saw ways to improve upon it. They added "Type X attachments," with the result that their machine became known as Typex. It, like the Sigaba, avoided reliance on glow lamps; it printed out its messages, enciphered one way, deciphered the other. Instead of the military Enigma's plugboard, it included two stationary entry rotors in addition to three stepping rotors. The Typex's rotor movements were much less regular than Enigma's. These modifications to the Enigma concepts gave Britain, too, a machine that handled top-level cryptography without ever being broken.

  By a fortunate happenstance, the British and American machines were similar enough that adapters could be used in exchanging enciphered communications between them.

  After their terrible cryptographic disasters of the Great War, the Russians developed better code systems in World War II. They depended primarily on codebooks and the secure but burdensome onetime pad system, which required both sender and receiver to use the duplicate top sheets of keyed paper pads, and then discard the sheets after a single use. The Russian military's codebook systems relied on four series of numerical codes, with five-digit codes for the top-level strategic messages scaling down to two-digit codes for frontline exchanges. They also made some field use of Hagelin's M-209 machines, which they received in lend-lease from the U.S. and copied.

  Soviet espionage agents across Europe and in Japan had their own codes for communicating with Moscow. These were onetime numerical systems with the key numbers lifted from published statistics tables and given an additional scrambling. For the most part, these were systems that frustrated German cryptanalysts.

  As for the Italians, Mussolini's navy bought Enigmas but did not add plugboards to them. This simplified choice would have provided the Allies a cryptanalytic bonanza, except that the Italians made very limited use of their Enigmas. Much more fruitful for the Allies was the Italian use of a Hagelin machine for enciphering information concerning the Mediterranean convoys sent to supply Rommel's armies in North Africa. Otherwise, the Italian military services relied mainly on codebook systems, some of which proved very difficult to break.

  Japan's decisions about cryptography were influenced by a 1931 book, The American Black Chamber, written by U.S. cryptologist Herbert Yardley. In the twenties, Yardley's secret operation was breaking the codes of a number of nations, including Japan. The funding of the operation was supplied mostly by the U.S. State Department. When, in 1928, president-elect Herbert Hoover appointed Henry L. Stimson as his secretary of state, Yardley sought the approval of his new boss by sending him a batch of messages revealing the Japanese government's plans for gaining the best possible deal in a forthcoming international conference on the naval armaments levels the leading nations would be allowed to maintain. Yardley anticipated that the new secretary would welcome this inside information. Instead, Stimson took offense. "Gentlemen," he has been f
amously quoted as saying, "do not read each other's mail." Yardley's funding was withdrawn and the Black Chamber closed down. He got even—and temporarily rich—by writing his no-secrets-held-back best-seller.

  Forewarned by Yardley's irresponsible disclosures, the Japanese set about developing more-secure code systems. They were faced by a language problem, for their written language relied heavily on some two thousand ideographic characters that evolved from Chinese. These characters, however, convey meaning but not sounds. To represent their language's sounds, the Japanese added phonetic symbols called kana. Only the kana characters lent themselves to radio codes. For radio transmissions, the Japanese developed their own Morse code, using Morse dots and dashes to represent forty-eight kana characters. As an alternative, they transmitted radio messages in Romaji, a Romanized spelling of the kana characters.

  Japanese cryptologists inspected the available code machines and then developed their own design. The Imperial Navy began the work but then decided against machine encipherment, relying instead on codebooks whose number equivalents of essential words and phrases were further secured by a complex upper layer of scrambling. The Foreign Office took over the machine project for enciphering diplomatic messages.

  Their first try was known to the Japanese as the angoo-ki taipu A, the "Type A cipher machine," or as the 91-shiki Oo-bun In-ji-ki, "Alphabetical Typewriter 91." The 91 was derived from the year of the machine's development, since 1931 was 2591 in the Japanese calendar. Employing Romaji letters, the device had two electric typewriters, one for typing in the plaintext, the other for typing out the enciphered message for the radio operator to transmit. In addition to the two typewriters, it had an Enigma-like plugboard and an encipherment mechanism. The Japanese developed two versions of Type A, one for diplomats, the other for naval attachés.

  Soon after, however, the Foreign Office introduced a new machine, one the Japanese called angoo-ki taipu B, the B machine. It made its first appearance on March 20, 1939, and gradually replaced the A machine. Type B was, as will be discussed later, a machine based on principles that radically departed from those of any other code device. For the cryptanalyst, it presented a challenge of an entirely different order from that posed by the Enigma.

  So it was that in the aftermath of the Great War, and with another world conflict looming, each major nation gave hard thought to its cryptologic procedures and technologies. Most intelligence organizations changed from manual practices to the use of code machines. Others introduced codebook complexities that their intelligence leaders felt sure the enemy could not solve. Overall, the codemakers brought their art to the point where their systems seemed secure against the codebreakers—so secure, in fact, that Yardley, after viewing the demonstration of one such scrambler machine, lamented, "When all governments adopt such a system, crypt-analysis, as a profession, will die."

  The story to be told next, however, relates how one lesser nation found bright young men who, well before World War II commenced, pierced the tight security screen and proved that even code machines could be mastered. The cryptanalysts of Poland, struggling to meet the threat of a renascent German military, began the slow, faltering change that brought codebreaking into the ascendant and set the war on its course toward Allied victory.

  2

  Breaking the Enigma: Poles Show the Way

  On a Saturday in January 1929, a crate arrived at the customs office in Warsaw addressed to a German firm with offices in the city. The box bore a label specifying that its contents consisted of "radio equipment." The German firm's representative arrived and, claiming that the crate had been shipped from Berlin by mistake, demanded that it be returned to Germany before going through customs. Their suspicions aroused, the customs officials decided not to comply with the demand right away. They used the excuse that their people did not work on Saturday afternoons, but assured the representative that the box would be returned on Monday morning. By this stratagem the Polish Biuro Szyfrow, or "Cipher Bureau," alerted by the customs officials, had a whole weekend to investigate the suspicious "radio equipment."

  Carefully disassembling the crate, Cipher Bureau experts found that it contained not radio equipment but a code machine. They took photos, made a diagram of its construction and measured its dimensions before repackaging it. Their information was tucked away in a file in the Intelligence Cipher Department.

  This was the Poles' introduction to the Enigma.

  The people of Poland had good reason to be sensitive to their neighbors' communications equipment. After being wiped off the map of Europe in the 1790s by a series of partitions that awarded its lands to Austria, Prussia and Russia, Poland had been re-created with the agreement of the Allies even before the cessation of hostilities in 1918. With their nation locked in a vise between Germany and the USSR, Polish leaders felt a desperate need to read the other countries' intentions by breaking their codes. The importance of codebreaking had been borne home to them in 1920 when decrypts of Russian military messages helped stave off an attempt by the new Soviet Union's armies to march through Poland and link up with pro-Communist German revolutionaries. In those early days, the Germans' pencil-and-paper codes were also yielding to Polish cryptanalysts.

  During the 1920s the Poles did two things that proved critical to their later cryptanalytic triumphs. In 1921 they signed a political and military pact of mutual assistance with France, since the French wanted to hold their old German nemesis in check with the threat of a two-front war. In addition, the leaders of Poland's Cipher Bureau in 1929 enrolled some twenty young mathematics students in a course in cryptology at the University of Poznan.

  The latter move was in response to a disturbing change: the Poles could no longer decipher the messages the Germans sent by radio. The head of the Cipher Bureau surmised that their potential enemy had switched from codebook to machine cryptography. Coping with code machines, they reasoned—long before the French or British came to the same conclusion—required advanced mathematical skills rather than the linguistic bent that had heretofore left cryptanalysis largely to philologists and classical scholars.

  Faced with successively tougher trial cryptograms, most of the Poznan students dropped out. Three of the brightest—Marian Rejewski, Jerzy Różycki and Henryk Zygalski—continued their interest in cryptography and were eventually employed by the Cipher Bureau. The stage was set for the contents of that sequestered file of information about the German code machine to be reexamined and the attack on Enigma to begin.

  The first step was to acquire one of the commercial Enigmas then on the market. For cover, the Cipher Bureau made the purchase through a cooperative Warsaw electrical firm whose business interests could logically justify the machine's use. Comparison of the machine with the file photos and diagrams proved that it was almost identical to that earlier model and suggested that Enigma was the German military's choice for mechanical encipherment. Experiments with the purchased machine, however, verified that the Germans had made changes that increased the Enigma's security. The Poles realized that they could make no headway against the Germans' military Enigma until they learned to build a working model of it.

  At this point, in December 1931, the Cipher Bureau received some unexpected help from Poland's French allies.

  The Captain and the Turncoat

  France, as noted earlier, had been a leader in cryptology before and during World War I. In the postwar period, however, the military command relaxed this phase of its operation, reducing the number of army cryptanalysts to just eight. These analysts bothered themselves only with some of Germany's simpler pencil-and-paper codes. Army leaders had grown complacent because Germany's military preparations were severely restricted by the Treaty of Versailles. No matter how adept the Germans became in cryptography, the forces they were allowed could in no way compare with the superb armies of France.

  Fortunately for the eventual course of the war, this thinking was reversed by Captain Gustave Bertrand, who had worked in cryptology since the cl
ose of the Great War. Bertrand believed strongly that the coming era of machine cryptography required new approaches, including a readiness to buy or steal opponents' cipher secrets. He formed a new army intelligence department specifically for this purpose.

  In the summer of 1931 he received a letter from a young German employed in the cipher section of the German Foreign Office. Hans-Thilo Schmidt was nursing grievances because of his low pay and conspicuous lack of success in comparison with that of his older brother, a high-ranking officer who had arranged for Schmidt's job as a code worker. Schmidt also longed for the richer life that French intelligence payouts could provide. He was willing, for a price, to assume the code name of H.E.—in French the acronym sounds like Asché—and turn over to Bertrand the secrets of the very machine that his brother Rudolf had approved for German army use.

  Bertrand's first meeting with Schmidt was arranged by a wily go-between named Rodolphe Lemoine, code-named Rex. At the meeting, Schmidt supplied documents that included the instruction manual for the use of the Enigma and the directions for setting its keys. Bertrand's enthusiasm for his find, however, was dampened by France's leading cryptanalyst of the time, who found the material of little value without an indication of the wiring of Enigma's rotors and the actual keys in use for a given period. Bertrand received the same lukewarm response from British intelligence. Not one to give up, he sent photocopies of the two booklets to Warsaw by diplomatic courier while he himself flew there to meet with the Poles.

  There, his reception was much warmer. The Poles immediately recognized the value of the booklets Bertrand had received from Asché. As recalled by Rejewski, "Asché's documents were like manna from heaven, and all doors were immediately opened."