THE CODEBREAKERS

by DAVID KAHN

  During the drive to isolate Tobruk, for instance, the Fernmeldeaufklärung Company overheard a radiotelephone conversation in clear at 10:30 a.m. June 16, 1942, between the 29th Indian Brigade and the 7th Armored Division. From this it appeared that the garrison of the El Adem box, or strong point, intended to attack the Germans that night. The information was passed to Rommel and his intrepid 90th Light Division, who attacked at once, catching the British so off balance that instead of their pummeling the Germans, Rommel captured El Adem. This enabled him to surround and isolate Tobruk, which unexpectedly capitulated on the 20th, allowing enormous quantities of stores to fall into German hands and giving the daring Panzer leader his opportunity to strike immediately for Suez. It was aid of this sort that prompted Rommel’s intelligence officer to call Seeböhm’s Fernmeldeaufklärung Company “a very important factor in Rommel’s victories.” The company could also have independently read the Fellers messages with a furnished copy of the BLACK code to save time in getting the information to Rommel.

  On July 10, the swirling desert warfare brought the Afrika Korps staff headquarters directly into the path of a British armored thrust. In a brief, fierce spurt of action, the brilliant Seeböhm was killed and most of his unit wiped out or captured. Many of their records fell into British hands. This loss deprived the company’s replacements of a great deal of necessary information, and at the same time enabled the British to correct many radio-security mistakes. Rommel thus lost the microscope that scrutinized the enemy lines and presented to him so many bits of information.

  At about the same time he lost his telescope. The United States appears to have had some suspicion of the leak earlier in the spring, when two officers came out from Washington to check on Fellers’ security measures. They cleared him, and perhaps this lulled their fears until new information reached the Allies. Apparently a prisoner of war told the British of the intercepts, and the British, who had themselves broken the BLACK code and its superencipherment, using it to read other traffic, now began to pick up Fellers’ messages within an hour after he filed them. After ten days of studying his “long, detailed, and extremely pessimistic” reports, they notified American authorities late in June of the leak and perhaps of Fellers’ attitude. Fellers himself was never told of the German solutions, but was recalled to Washington, returning in July.* Later messages from Cairo still contained some noteworthy observations but no broad view of the situation. And when the new military attaché there began using the M-138 strip cipher, which defied all Axis attempts at solution, it cut Rommel off from the strategic intelligence on which he had so long depended.

  The loss occurred just as he was crossing the frontier into Egypt and seemed to have the Pyramids and victory almost within his grasp. The British 8th Army fell back to its fortified positions at El Alamein, and on July 2 Auchinleck jabbed out with the first of a series of counterattacks. Rommel, deprived of his most valuable source of information, could no longer take the expeditious measures for defense and offense that he was previously enabled to. On July 4, he reported that he was going over to the defensive. Meanwhile, Britain succeeded in reinforcing Malta, and attacks from there pinched the Axis pipeline. Rommel clamored in vain for fuel.

  At the same time, the 8th Army built up a powerful force in secrecy, and concealed not only the date but the direction of its main thrust. Two divisions arrived with 240 guns and 150 tanks. In the old days, the Afrika Korps would have learned of it from Fellers’ messages; this time they never knew the two were there. The British had profited from their capture of the Fernmeldeaufklärung files to institute an improved call-sign procedure, tauten cryptographic discipline forward of divisional headquarters, introduce radiotelephone codes, impose rigid wireless silence on reserve formations, pad out real messages with dummy traffic, and create an entire fake signals network in the southern sector. The new Fernmeldeaufklärung staff had neither the talent nor the experience to penetrate these disguises and sift the true from the false. The Germans, who had been used to the constant flow of information from Seeböhm’s men, had to depend almost exclusively upon air reconnaissance, without any radio-intelligence corrective. And camouflage fooled it. Hundreds of tanks and guns were hidden beneath dummy trucks; large supply depots were created so slowly in the south that it looked as if they could not be ready for several months.

  So when General Bernard Montgomery opened fire with a thousand cannon on the German positions at Alamein on October 23, it came as a complete surprise to the Afrika Korps. Rommel had been so certain that nothing would happen for a while that he had gone to Austria to convalesce. He flew back at once to take personal charge of the battle, but by the time he arrived it had already been lost. Hampered by shortages of oil, men, and armor, he could only shift his divisions about in desperate but futile attempts to recover. The defeat became a rout, and the Afrika Korps fled west across the desert, leaving a battlefield littered with hundreds of destroyed or useless tanks and troop-carriers. A few months later the Germans were driven out of Africa, then out of Crete, then up the boot of Italy—always retreating, never again advancing. The Battle of Alamein marked the turning of the Allied hinge of fate. “Before Alamein we never had a victory,” Churchill said. “After Alamein we never had a defeat.”

  That change in fortune had revolved, to no small degree, upon cryptology.

  * This division carries into the practical sphere the distinction that codes operate upon texts linguistically whereas ciphers operate nonlinguistically.

  * No Pers Z representative appears to have attended—probably a reflection of the high-level personal dislikes and power struggles between Göring and Himmler on the one hand and Ribbentrop and the military on the other. At one point Göring tried to bring Pers Z within the ambit of the Forschungsamt.

  * The term “Nachrichten” reflects this, since it means not only “communications” or “signals” but also “intelligence.” In nonmilitary contexts, it means “news” or “information.”

  * Gherardi stayed on until Italy’s declaration of war upon the United States closed the embassy. After the war, he coolly asked for his old job back—and got it! He held it until the secret finally leaked out; then, after several interrogations, he resigned, in August, 1949.

  * Later in 1942 he was awarded the Distinguished Service Medal for his work as military attaché, which “contributed materially to the tactical and technical development of our Armed Forces.” The citation also stated that “His reports to the War Department were models of clarity and accuracy.”

  15

  DUEL IN THE ETHER: NEUTRALS AND ALLIES

  THE BELLIGERENTS were not the only ones who availed themselves of the valuable intelligence of cryptanalysis. Vichy France, for example, installed about 50 cryptanalysts and clerks in a villa outside Lyons. Their success seems to have been limited. A newspaper story reported their 1941 failure to solve the systems of former French Minister of the Interior Georges Mandel, Churchill’s friend, then in Vichy custody for his attempt to set up a resistance government in North Africa with himself as premier after the fall of France. At least part of their work was directed at the Free French and the underground, but they never communicated any of their results to the Germans. In fact, one member of the bureau, Charles Eyraud, later known as the author of a fine modern work on cryptanalysis, himself burned all the bureau’s papers when the Germans occupied all of France.

  Almost certainly the best of the nonbelligerent cryptanalysts, and perhaps one of the best in the war, was that of the precarious neutral, Sweden. At first she used codebreaking primarily to see whether Hitler planned to grant her the same sort of military protection that he so generously accorded Norway and Denmark. His preparation for occupying those two countries was one of the best-kept secrets of the war, and Sweden did not want to be caught napping. Later she used the intelligence to keep abreast of a variety of political events.

  Except for a brief interlude back about the turn of the century, when R. Torpadie so impressed th
e Swedish authorities by solving a nomenclator of 1632 for a historical study that they commissioned him to set up a cryptologic bureau called Room 100, Swedish cryptology got its real start with Yves Gyldén. His father, Olof, the head of the Royal Naval School, had been financially interested in Arvid Damm’s cipher machines. Yves, who got his un-Swedish first name from his French mother, became cryptologically interested and subjected them to every possible cryptanalytic test. The interest thus kindled in cryptology remained with him throughout a business career with the pharmaceutical firm of Astra, founded by his grandfather. In 1931, a tall, grave man of 36, Gyldén published his Chifferbyråernas insatser i världskriget till lands, a keen, perceptive study of World War I cryptology and its effects. Its 139 pages were later translated into English by the U.S. Army Signal Corps as The Contributions of the Cryptographic Bureaus in the World War, and portions were published in the Revue Militaire Française. This book demolished the lingering myth of chamber analysis, demonstrated the crucial role of errors and of torrents of ciphertext, and generally crystallized the lessons of World War I and catalyzed the evolution of the cryptology of today.

  Five years after the influential little book was published, Sweden set up a cryptologic bureau. It was headed by Colonel C. G. Warburg, a gentleman who had fallen off a horse, broken both arms and legs, and needed a sinecure. He proved as incompetent in cryptology as in equitation, and was replaced by a naval officer who won the respect of the experts who later served under him. During the late 1930s Gyldén gave many talks on crypt-analysis to Swedes. He also sowed the seeds of a valuable cooperation with the other Scandinavian countries when he lectured in Oslo and stimulated a Captain Rocher-Lund to set up Norway’s first cryptologic office. In 1939, during a 12-hour war game, Gyldén headed the cryptanalytical office that solved 38 of the 56 rather simple cryptograms transmitted by the “invaders.” Sweden’s preparations extended to recruiting talks at Uppsala University, where coeds were entertained with the intrigues of cryptology and sold on the idea that they could become good codebreakers. Other personnel were drawn from the winners of cipher-solving contests which the cryptanalysts got the newspapers to run.

  When war broke out, the Swedish cryptanalysts numbered 22. All were paid the magnificent sum of half a crown a day (raised later by stages to two crowns), as a result of which most of them engaged in a kind of part-time cryptanalysis—working for the government in the morning and at regular jobs to get money to live on in the afternoon. They were installed first in the Gray House, Sweden’s Defense Ministry building, and afterwards in an old house at Carlaplan 4, since demolished and replaced by Sveriges Radio; they finally settled down in an old, drafty, noncentrally-heated apartment house at Styrmans-gatan 2. (A branch was also established in a modern apartment house in Strandvagen in 1943.)

  In 1940 the cryptanalysts were divided by language, though some of the mathematicians shifted from group to group. The four units were: No. 1, for Romance languages, primarily French and Italian, headed by Gyldén, who had spent ten years in France and was fluent in that language; No. 2, for German, in which one of the brightest workers was Carl-Otto Segerdahl, a young mathematician; No. 3, for English, which attacked American and British systems and was headed by Dr. Olof von Feilitzen, 32, a librarian whose English is better than that of many Americans; No. 4, for Russian, headed by Dr. Arne Beurling, 35, a big, slow-talking, quietly handsome professor of mathematics at Uppsala University, who in 1952 became a member of the Institute for Advanced Study at Princeton. Beurling, one of the war’s finest cryptanalysts, also determined the unknown ciphers of other countries and made the initial breaks. Gyldén, as the founder, was a kind of first among equals; he also taught new recruits. These came in at such a pace that by the time he left in 1941 the group had grown to 500, and by the end of the war to 1,000.

  Messages, too, poured in. Teletypewriters, cut directly into Swedish post-office circuits, duplicated messages sent over those wires. Norway, Denmark, and Finland forwarded their intercepts to Sweden, which had the only effective cryptanalytic center among them, and these messages enabled Sweden to make very fruitful comparisons between the same text enciphered in different keys. She paid her Nordic associates back with the information gained in the resultant cryptanalysis—sometimes with valuable results.

  Early in 1940, just before the German occupation of Norway, Nazi agents there, who were concentrated in the German-Norwegian shipping lines and in the large fishing and fish-processing firms, were ordered to pass back information on ship movements and weather. They disguised the data as sales prices, offers, and tonnage reports on fishing, and transmitted by telephone and radio. But the Norwegian authorities had intercepted the telephone calls, which dealt with prices in a highly suspicious manner. They sent recordings to Sweden, where Segerdahl discovered that the five-digit “prices” actually represented the transposed and monalphabetically enciphered numbers of ships in Lloyd’s Register. The solutions enabled Norway to break up at least one of the rings in February, though others continued to operate.

  The Swedes not only used cryptology against foreign espionage, they sometimes used espionage against foreign cryptology. In one case, they tapped a telephone call between the Italian military attaché in Stockholm and his colleague in Oslo. The recording sounded absolutely unintelligible, and the Swedes at first thought that the Italians had used a telephone scrambler. When they determined that they had not, the recording was sent to the language department at Uppsala, where it was found to be a Sicilian dialect rendered incomprehensible by the attaché’s over-liberal use of cursewords. Eventually the sense was sorted out, and the conversation proved to comprise the Stockholm attaché’s explanations of how to use the military attaché code, which the Oslo man—who was railing at the idiots in Rome who would send him such a code—could not fathom. Between the explosions of the colorful Sicilian equivalents for “dunce” and “jackass” and still other expletives were references to operating procedure, meanings of specific codewords, and so on. Needless to say, it proved a great help to Gyldén in his Italian-code solutions.

  The Swedes also obtained much help from their own Foreign Office in the form of diplomatic notes sent and received, reports of notes verbales, aides-mémoires of conversations with various ambassadors, and other memoranda. This is common practice in all countries, but the Swedish cryptanalysts carried it to a peak of perfection by using as their liaison man a former foreign minister. Rickard Sandler, 56, had served in that post from 1932 to 1939; he had also filled in as premier for 18 months in 1925 and 1926, and in 1934 had been elected president of the League of Nations Assembly. Spare and round-faced, Sandler had been bitten by the cryptology bug, and in 1943 he wrote a book on famous ciphers. But he proved inept as a cryptanalyst, unable to solve what the Swedes regarded as the simplest of practical problems—Norwegian one-part codes. However, he was a great success in making sure that the Foreign Office reported every scrap of information promptly to the crypt-analysts. So well did he have his contacts trained that the Foreign Office even reported the time of departure of an ambassador’s car from the Foreign Office building. With this little datum, the cryptanalysts—knowing the message he had been given and estimating how long it would take the ambassador to drive to the embassy and have a message of that length encoded and sent to the telegraph office—could more easily pick out the cryptogram corresponding to that message from the embassy’s daily file.

  As usual, the Swedish cryptanalysts were greatly helped by lazy or stupid encoders. Clerks repeatedly violated the most elementary rules by failing to superencipher and forgetting to bisect messages. The worst bungler the Swedes came across was the German consul at Stavanger, whose numerous blunders became the vulnerable heel of many a German message. His name—almost too fittingly—was F. W. Achilles. The Swedes appreciated his help so much that they hung a large photograph of him in their office. “He was very fat and he looked like a gorilla,” Segerdahl said. “I never met the man personally, but I considered
him my best friend in the German diplomatic service!”

  The Swedes also read messages in other German systems—a double transposition for the military attaché and two substitution systems for the troops. The latter gave them an unexpected peek into the sex habits of German soldiers. The Wehrmacht provided women from the Baltic states and concentration camps as prostitutes for the occupation forces in Norway, and the vessels were naturally awaited with great eagerness. Their arrivals and departures formed the subject of excited communication between units, and not infrequently a radioman in a port from which a ship had just sailed would recommend one of the girls to a fellow signalman in the port to which the ship was headed. The reasons were sometimes quite specific, and the Swedes came to think that they knew the girls almost as well by cryptologic means as the soldiers did by carnal.

  But errors, circular messages, and all the other aids would not have helped the Swedes much if they were not as clever as they were. They became so attuned to French procedure in regard to a multiplicity of codes—at one time the French had eleven in simultaneous use—that they could tell when the French regarded them as compromised (after about four years) and began sending material in them that they wanted others to read. Usually this tried to implant the idea that the French were acting only out of the most moral considerations in a given situation, probably to distract attention from their real motives. Many phrases from the messages in these compromised codes later showed up in the French Yellow Books, the official governmental statements of their positions. The Swedes also solved an American-British code in which U-boat warnings were transmitted—probably the same that Germany’s B-Dienst read—and thus got a free ride in safeguarding their own merchantmen.