by DAVID KAHN
Quite possibly the finest feat of cryptanalysis performed by the Swedes, and the most far-reaching, was Arne Beurling’s solution of the German Siemens machine. Since German messages passed over Swedish wires just as German soldiers passed over Swedish rails, both the Wehrmacht in Norway and the German embassy in Stockholm took advantage of the machine’s online capabilities to wire messages directly to Berlin. The German Foreign Office called the machine the Geheimschreiber (“secret writer”). The teleprinters in the Swedish cryptanalytic bureau rapped out the German correspondence, and it was given to Beurling for an attempt at solution.
He observed at once that the ciphertext consisted of the 26 letters and six digits, a total of 32 characters, or 25. This suggested a cipher based on a teletypewriter to him, since he knew that teletypewriters used a five-hole punched tape. That was about all he knew, though, and he had to get a book on them to see how they worked. His studies—perhaps aided by an examination of patents—led him to the conclusion that a machine based on the Baudot code would encipher by shifting the positions of the five contacts, that each of these positions would very likely be controlled by a keywheel of its own, and that the number of control pins on the circumference of these wheels would vary from wheel to wheel to make the period as long as possible.
Since the key probably changed daily, Beurling selected the traffic for a single day, May 25, 1940, to work on. It covered the equivalent of two large sheets of paper. His analysis soon showed that his preliminary suppositions were correct, except that the substitution of the Baudot pulses was followed by a transposition. Very often the transposition had no effect. If, for example, pulses 1 and 2 were the same, the transposition of 1, 2, 3, 4, 5 into 2, 1, 3, 4, 5 would leave the character unchanged. Beurling took full advantage of these peculiarities to reconstruct the mechanism. He checked his work with new data from the traffic of May 27, found it was correct, and within two weeks of undertaking the job had solved the cipher. A Swedish mechanic constructed an apparatus to Beurling’s specifications, and though it looked monstrous and made a terrific racket, it printed out the German messages that the Swedes wanted to read.
To recover the daily keys, the cryptanalysts would work through the night, and in the morning, when the Swedish commander, Lieutenant General Olov Thornell, came in to ask, “What’s the news from the Germans today?” they were usually able to tell him. Twice when the Germans made threatening moves with their troops in Norway toward Sweden, Swedish troops, alerted by cryptanalyzed messages, moved swiftly into position and blocked the Germans. Their commander, General Niklaus von Falkenhorst, later extended congratulations to Thornell on the brilliance of his tactics. Thornell passed the felicitations on to the cryptanalysts.
In the spring of 1941, the Swedes cryptanalyzed other German military messages that, put together, spelled an invasion of Russia between June 20 and 25. Erik Bohemann, secretary general of the Swedish Foreign Office, passed the information to Sir Stafford Cripps, British ambassador to the Soviet Union, at a dinner in Stockholm while Cripps was passing through. This may not have come as news to Cripps, who may have known of the invasion from other sources, but it certainly reinforced any knowledge he had. Unfortunately, Stalin did not believe the British.
The dozens of diplomatic messages that clattered out of the Beurling mechanism told the Swedish Foreign Office what the Germans were really doing and thinking. They gave Foreign Minister Christian Günther advance warning of diplomatic notes that the German embassy was ordered to submit to him. The cryptanalysts tell a story that, after reading a particularly demanding note, they took the unusual step of notifying Günther of its contents by telephone, which they rarely used. (Later they sent it over by the regular messenger, who wore two shoulder holsters.) Günther promptly went on a “hunting trip,” and the German diplomat could not serve his demand until after the weekend. By then the Swedes had formulated a policy that enabled them to tell the Germans, with suitable regret, that they were unable to fulfill the requests.
And so Sweden’s cryptanalysts helped her navigate the perilous waters of neutrality while all about her raged the war.
Great Britain’s main cryptanalytic agency lay within her Foreign Office, which had taken over the personnel of the Admiralty’s Room 40 at the end of World War I. The Reverend William Montgomery, one of the solvers of the Zimmermann telegram, for example, joined the Foreign Office. Early in the 1920s, in a circular urging its diplomats to be more careful in the use of their codes, the Foreign Office told them that it was spending £12,000 a year, or almost $60,000, both in keeping British codes secret and in solving those of foreign governments, and that carelessness in handling codes was wasting much of this (or at least much of the part spent for British cryptography). The usual legends circulated among the diplomats about their code experts, some of whom had “made a life-long study of the work.” One story credited one of these wizards with solving a Turkish code during the war in less than five months, though he himself could not speak Turkish and had had to call in experts in the language to translate the messages. The Foreign Office reportedly considered no code as fully secret after it had been used for six months; consequently it changed all highly confidential codes every four months.
In 1939, the Foreign Office moved what it euphemistically called its Department of Communications to Bletchley Park, an estate and mansion in Bletchley, a town in Buckinghamshire about 50 miles northwest of London. It is far and away the most history-redolent black chamber of all. The British, of course, trace the land from a Roman encampment, through its award by William the Conquerer to Bishop Geoffrey of Constance for services rendered at the Battle of Hastings, down on through the ownership of various lords (most notably the two George Villierses, first and second dukes of Buckingham) and rich men of decreasing interest. A mansion was first built on the land in the 1870s and added to repeatedly; the Foreign Office, finding this too small, added many buildings, including a cafeteria and a large hall. Eventually 7,000 worked and trained there, including members of the armed services.
Britain urges cryptographic discipline
The War Office expanded its M.I. 1(b), the cryptanalytic agency started in World War I under Major Hay, to M.I. 8—the same name, coincidentally, as that held by Yardley’s organization. The Admiralty and the Air Ministry presumably had cryptologic agencies of their own. One of the first victories of the Admiralty’s unit was, surprisingly, in the domain of cryptography.
Since the beginning of the war, Admiralty secret communications had been read by the B-Dienst, with such disastrous results as the loss of Norway almost by default. The Germans continued to listen in to Admiralty messages during the critical summer of 1940 as Hitler prepared for Operation SEALION—his invasion of England. The cryptanalytic intelligence had long been entering into operational planning, and the Oberkommando der Kriegsmarine had come to depend on it. Suddenly, on August 20, as all England was bracing itself in its finest hour, and the sky above was streaked with contrails as the few earned their tribute from the many, the Admiralty, which had finally tumbled to the German cryptanalysis, changed its codes and ciphers. O.K.M. went deaf. The abrupt cutting off of quantities of information about British plans and dispositions caused, a German said, “a great setback for German naval strategy.” No longer could German vessels strike out at the greater British forces with foreknowledge or move deftly out of their way. British sea power rapidly gained its normal ascendancy. English ships shelled the invasion fleet in Channel ports. Air reconnaissance alone could not tell the Germans enough. The O.K.M., never very warm for SEALION, chilled still further. Eventually its coolness spread throughout O.K.W., and then to Hitler. It contributed to his ultimate decision to postpone SEALION indefinitely, and hence forever.
All of Britain’s cryptologic work seems to have been coordinated by the Foreign Office’s Department of Communications, which apparently handled strategic and primary cryptosystem solutions. All over the world, Britain had about 30,000 persons in communications i
ntelligence. Deputy director of the Department of Communications was a man who had already made a mark in the world by his cryptanalytic efforts. He was Nigel de Grey, who in 1917 had solved the Zimmermann telegram.
A British naval officer demonstrates the proper codebook security for when capture threatens
The department turned out solutions at a fairly rapid rate. On November 21, 1941, a Japanese diplomatic solution was given number 097975; on December 12, another Japanese diplomatic solution was numbered 098846;—indicating almost 300 solutions a week at that time (not Japanese alone, of course). A typical distribution of these solutions would send three copies each to the director of the department, the Foreign Office, and the War Office, two to the India Office, and one each to the Admiralty, the Air Ministry, the Colonial Office, the Dominion Office, M.I. 5 (counterintelligence), and Sir Edward Bridges, secretary to the Cabinet. The appearance of Bridges’ name on the list suggests that some of the British intercepts may have been read aloud at Cabinet meetings. In addition, Churchill, on August 5, 1940, ordered that a daily selection of original intelligence documents be submitted to him personally “in their original form,” which almost certainly included intercepts. Much of the cryptanalytic output must have gone to the Joint Intelligence Committee, which evaluated all intelligence. It was always chaired by a Foreign Office representative, who was Victor F. W. Cavendish-Bentinck throughout most of the war, and included the directors of military, naval, and air intelligence.
The intelligence from these solutions went also to the United States, but so closely did Britain guard her cryptanalytic capabilities that for more than a year she would give the United States information based on the cryptanalyses but would not name the source. In January, 1941, however, a four-man American cryptanalytic mission accompanied a PURPLE machine to England to establish technical cooperation with British cryptanalysts. Britain had not cracked the PURPLE machine, but they had more in the way of cryptanalyzed intercepts than the United States, and this was the quid pro quo. This cooperation between the two English-speaking nations in the most sensitive of areas tells the depths of their friendship. The American Signal Intelligence Service and OP-20-G radioed the PURPLE keys to London daily. Cooperation extended to the small Australian communications-intelligence unit and to the unit at Singapore, and Canada assisted in making sure that all got all Japanese intercepts.
Some of the most important British communications intelligence resulted, however, not from the scribblings and quiet cogitations of reticent cryptanalysts, but from the explosive sexual charms of a British secret agent in America. Her unlocking of several hearts gave Britain access to vast treasuries of intelligence. She was an American, the daughter of a Marine Corps Major. Her real name was Amy Elizabeth Thorpe but she was known in espionage by her codename, CYNTHIA. She had had her first sexual experience at 14, and was pregnant when, at 19, she married a junior British diplomat, whom she later divorced; at the start of her espionage work she had just turned 30. A moderately attractive blond, tall and with prominent features, soft-voiced, a good listener, and with a sensuality that was indefinable but very much present, she served British Security Coordination, Britain’s intelligence organization in the United States, not for money, but for thrills.
In the winter of 1940-1941, B.S.C. assigned CYNTHIA the task of obtaining the Italian naval cryptosystem. She managed an introduction to the Italian naval attaché in the embassy at Washington, Admiral Alberto Lais. Within a few weeks he was infatuated. When she was certain of her power over him, she told him directly that she wanted the naval code. Lais, despite his age and experience, agreed without any protest to betray his country for a woman. He arranged for her to meet his cipher clerk, who produced the codebook (and, presumably, any superencipherment tables for it) for a fee. It was promptly photostated and returned to the safe; the photostats went to London.
A few months after the English cryptanalysts received them, CYNTHIA’s feat paid off. As Churchill obliquely put it, “Towards the end of March [1941] it was evident that a major movement of the Italian Fleet, probably towards the Aegean, was impending.” Admiral Cunningham, commanding in the Mediterranean, whose cryptanalytic unit probably had been given copies of the Italian code and superencipherment, sensed by March 25 the Italian sortie against British convoys carrying troops to aid Greece. Two days later he slipped out of Alexandria after dark and set his course so foresightedly that at dawn a scouting plane contacted the enemy squadron. Though the Italians were also reading Cunningham’s messages and so took action that increased the difficulty of his attack, he destroyed the cruisers Pola, Fiume, and Zara, and damaged the battleship Vittorio Veneto in the Battle of Cape Matapan. The victory, Churchill said, “disposed of all challenge to British naval mastery in the Eastern Mediterranean at this critical time.”
A few days later, the State Department declared Lais persona non grata as a result of sabotage plans that he had disclosed to CYNTHIA, who had maintained profitable contact with him. At dockside, he spent his last few minutes with CYNTHIA, ignoring his weeping family. His departure enabled her to turn to her next assignment, at the embassy of Vichy France.
She gained entrance by posing as a newspaperwoman. During the wait to interview the ambassador, she chatted for an hour with the press attaché, Captain Charles Brousse—and captivated him. By July of 1941, after allowing him to seduce her, she “confided” that she was an American agent and urged him to work for the real France against the Laval government. Soon she was getting a plaintext copy of every incoming and outgoing telegram of the embassy, plus a daily report that Brousse wrote to fill in the missing details.
The plaintext undoubtedly enabled the British to reconstruct French diplomatic codes, if they had not already done so.* But in March of 1942, London asked British Security Coordination to obtain the French naval code, which was used both by naval attachés and fleet commanders. This might have stemmed ultimately from an order by Churchill himself. He was then mounting a force to seize French-owned Madagascar to keep it from becoming a Japanese submarine base and he feared that Vichy might reinforce the island from Dakar just on the possibility that England might attempt such a seizure. “I therefore asked for extreme vigilance about any convoys or shipping which might pass from Dakar to the island, toward which our forces were already about to start,” he wrote. Supervision of Vichy naval signals constituted, of course, one aspect of this vigilance.
CYNTHIA first asked her lover to get the code for her, but he replied that it was an impossible task, since only the chief of the code room, one Benoit, and his assistant had access to that tightly guarded sanctuary. CYNTHIA approached them, but failed with both.
Undaunted, she switched tactics. She and Brousse—who was now totally infatuated and willing to assist her in any of her plans—arrived late one night at the embassy. They explained as tactfully as possible to the watchman the difficulty of obtaining hotel rooms in wartime Washington and, smoothing his qualms with a tip, went in to spend the night on a divan on the first floor. They repeated this several times, until the watchman became used to it. One night in June, 1942, they clambered out of their taxicab in festive mood with a bottle of champagne, which they invited the watchman to share. He was happy to do so. A few minutes later, he had sunk into a drugged slumber. The “cab driver,” an expert locksmith working for B.S.C., worked three hours and discovered the combination of the safe in the code room. But he did not have enough time to take the codebooks for photostating, and CYNTHIA and Brousse had to return two nights later.
It would be almost impossible to drug the watchman again; and furthermore it was inadvisable, for CYNTHIA felt that he was growing mistrustful of their repeated visits. She sensed that he would look in on them that evening, and so she prepared a counterstratagem that would allay his suspicions. When, as expected, the watchman walked into the room twenty minutes later, she was totally nude. It was utterly convincing. The watchman retired in confusion and did not bother them again.
They l
et the locksmith in through a window; he removed the codebooks and their accompanying tables of encipherment and handed them to another agent outside, who had them photostated in a nearby house by other B.S.C. operatives. By 4 a.m. they were back in their safe with no sign that they had ever been abstracted; 24 hours later the photostats reached England.
It was by then too late to help with the capture of Madagascar, which had gone off without a hitch the previous month. But plans were now afoot for the Allied landing in North Africa, and the photostated code helped keep the Anglo-American forces informed of the movement—or, rather, nonmovement—of the units of the Vichy French fleet at Toulon, Casablanca, and Alexandria during the invasion. Thus was England once again helped by a Lady Godiva.
No such dramatic feats were required by the British or by anyone else to read American diplomatic codes. The cryptanalysts who worked on them did not even have to furrow their brows excessively. For these codes of a great power were, from before World War I to the middle of World War II, as puny as those of many smaller nations. The United States must have been the laughingstock of every cryptanalyst in the world. And during World War I, the twenties, and the thirties, American diplomacy must have been conducted largely in an international goldfish bowl.
During that period, the Department of State entrusted its code compilation to the chief of its Bureau of Indexes and Archives, later the Division of Communications and Records, which handled both the code-making and the coding. For nearly all those years, the post was held by David A. Salmon, a career employee whose knowledge of cryptology was limited to what he had learned on the job.
He inherited from his predecessor, John R. Buck, the practice of designating American diplomatic codes by the color of their binding. Thus, since before World War I, the United States had had a RED and a BLUE code, both using five-figure groups. The RED was the older of the two, and State had given the Navy some copies of it for communication at outposts between diplomats and naval officers. In 1912, soon after Woodrow Wilson took office, the President’s Commission on Economy and Efficiency asked the State, War, and Navy departments to consider a standard interdepartmental code and “better and less expensive methods of enciphering cablegrams.” The RED code had by then become, in Salmon’s unconscious pun, an “open book.” Nearly a year later, the three departments finally agreed to use, as an emergency cryptosystem—the Vigenere! They knew it in a slightly different arrangement, called the Larrabee, in which the plaintext alphabet was repeated above each ciphertext alphabet and the keyletter was printed in large type at the left. Cryptographically, however, it was identical with the system that Kasiski had demolished half a century before. Not only did they merely contemplate using the system, which was bad enough; they actually did use it. And making things as easy as possible for foreign cryptanalysts were the short keywords: State used PEKIN and POKES in 1917.
