by Kahn, David
The Naval Section attacked the cryptosystems of foreign navies, while most of G.C.&C.S. concentrated on diplomatic solutions. In 1930, the Army Section was formed and, in 1936, the Air Section.
The most important cryptanalyst was Knox. In 1920, he had married a Room 40 assistant, Olive Roddam, and a year later, after selling most of his Great Western Railway stock, bought a damp, chilly, drafty house on 40 acres of woodland near High Wycombe, half an hour northwest of London. The cryptanalyst commuted to work by train, like his stockbroker neighbors, and in the coaches examined photographs of the papyrus rolls of the dramatist Herodas. In 1922, Cambridge University Press published the restored text of Herodas’s mimes and their English translation, with Knox as editor and with notes by his former classics tutor. Much of the book’s 465 pages consisted of densely printed commentary, mainly on Greek and Latin syntax and vocabulary. Because of the edition of Herodas and another contribution to classical scholarship, Knox was invited to become professor of Greek at Leeds University. But he turned it down, perhaps with regret.
He regarded his suburban neighbors with a certain cynicism: he was surprised at the prevalence of adultery when all the wives looked so much alike. But he played tennis with them, spinning the ball almost unreturnably. He bought a motorcycle for commuting as soon as he no longer had to study Herodas on the train; in 1931, he had the inevitable accident: he broke his leg, which left him with a slight limp. At about that time, frustrated by his work and yearning to return to Cambridge, he was kept from quitting by his wife’s reminding him of his duty to educate his sons and of the national importance of his work. This he carried out in a chaotic office at Broadway Buildings at hours that defied Foreign Office routine: he arrived early in the morning and departed at 4 P.M. He seemed to live on black coffee and chocolate.
Knox and his colleagues enjoyed fair success in those glory years of codebreaking. With Strachey and Clarke, Knox focused at first on the all-important American codes. Most posed few obstacles, and those that did, such as a superenciphered code introduced after the end of World War I, were solved after not too great a delay. Knox also successfully tackled Hungarian codes. A former czarist cryptanalyst, E. C. Fetterlein, tall, stolid, whose index finger was adorned with a large ruby given him for his successes in the years before the overthrow of Nicholas II, read Soviet Communist cryptograms with great panache—until his work was nullified in 1927 by deliberate exposure of the intercepts for political reasons. G.C.&C.S. solved French, Italian, Japanese, and Spanish messages, among others. Every week it distributed between sixty and eighty intercepts, each identified by a six-figure number.
Against only one country’s codes and ciphers did G.C.&C.S. fail: Germany. But it hardly seemed to matter. The former rival had been thoroughly defeated and all but stripped of her armed forces. The Admiralty, concerned more with Japan and America, had little interest in Britain’s former enemy. Technical circumstances further inhibited G.C.&C.S.’s feeble efforts. The German Foreign Office was using the one-time pad, the only theoretically and practically unbreakable cipher. German army radio traffic was hard to intercept in the British Isles. And with almost no German navy at first, there were almost no naval messages to intercept. Moreover, though in 1924 the Reichsmarine was still using the AFB code—the edition that Room 40 had broken—the difference between wartime and peacetime vocabulary made it difficult to solve messages.
Later, as the German navy gradually revived, studies showed that it was using a machine to encipher messages that first had been encoded in AFB. This so discouraged G.C.&C.S. that it stopped intercepting German naval messages altogether. For a decade it ignored Reichsmarine traffic. It thus lost the contact that is vital for keeping up with the small modifications that individually can be mastered—as the Poles did with the Enigma keying changes—but that in combination raise too great an obstacle. The five-man Naval Section directed its energies instead to the traffic of the French, Italian, Japanese, Soviet, and American navies.
Then, in July 1936, Spanish General Francisco Franco, fed up with the republican government in Madrid, rose in insurrection. Within days, Fascist Italy and Nazi Germany began sending him weapons. Their troops followed. As the civil war in Spain widened, the Mediterranean swarmed with troop transports, merchant vessels, warships. The ether hummed with messages, many of them Italian. G.C.&C.S. had long ago reconstructed Italy’s main naval code, in large part, Denniston remarked, “because of the delightful Italian habit of encyphering long political leaders [editorials] from the daily press.” These solutions had enabled Britain to track the movements of Italian warships during dictator Benito Mussolini’s invasion of Abyssinia in 1935, which threatened the Mideast bulwark of the very foundation of British power: India. Now the danger was more acute. Mussolini, hostile to Britain after her opposition to his aggression, was calling the Mediterranean mare nostrum and trying to make it an Italian lake. Just as his Ethiopian invasion had threatened British-controlled Egypt, the Spanish civil war imperiled Gibraltar, and Britain began to see that for the first time since 1798, she was threatened with being squeezed out of the Mediterranean.
It was in these circumstances that G.C.&C.S. realized that the Italian navy had introduced a cipher machine, soon identified as the commercial Enigma. Knox attacked the problem. The machine, including its rotor wiring, was known to G.C.&C.S., so it did not have to be reconstructed; only the daily keys had to be recovered. And since the commercial model lacked the plugboard of the military version, the problem was simplified. Knox used a technique that everyone who tried to solve rotor messages seems to have hit upon, one that has become known as la méthode des bâtons, the rod method, so called from the wooden rods onto which were glued strips of paper with the cipher alphabet of each rotor written on it.
Using the rod method, Knox succeeded in determining the keys for some messages and thus in reading other messages enciphered in that day’s key. He then attacked the messages of the Franco forces and of the German forces in Spain, which used the commercial Enigma. By 1937, he had succeeded.
These successes, together with the rise in German naval traffic volume with the Kriegsmarine’s arrival in some force in the Mediterranean, perhaps gave him hope that he might solve the German naval Enigma. G.C.&C.S. wanted to intercept lower-level messages in the hope of matching them to Enigma messages and so obtaining lengthy cribs, but lack of men and gear prevented this. And though Knox determined that the naval Enigma had a plugboard, he made no further progress.
The increase in international tension led to an important intelligence development within the Royal Navy. After Mussolini’s aggression in 1935, the deputy chief of the naval staff, Admiral William James, became concerned about the ability of the Naval Intelligence Division to cope with an emergency. His concerns were exacerbated during the Spanish civil war by the lack of intelligence about foreign naval forces, particularly submarines, blockading the Republican-held areas. During the last two years of World War I, James had headed Room 40, technically Section 25 of the Naval Intelligence Division, and he knew that it was the lack of coordination between the intelligence and operations divisions that had enabled the German High Seas Fleet to escape at the Battle of Jutland. To avoid a recurrence of that flaw, James proposed a center for intelligence that would directly serve operations. The director of naval intelligence began to plan the expansion of the insignificant movements section of the intelligence division into such a unit. It would collect and evaluate all operational intelligence and disseminate what was needed to the fleet. He assigned the expansion to Paymaster Lieutenant Commander Norman Denning, a round-faced man in his thirties.
By June 1937, the nucleus of the unit had been formed. Its first task was to track Italian submarines, and one of its early experiences led to a fundamental rule in handling communications intelligence. G.C.&C.S. had sent the new body the gist of the solution of a message to two Italian submarines that Mussolini had placed at Franco’s disposition. The message said that any merchant vessel
attempting to break the blockade was to be sunk outside of Spanish waters. This appeared unbelievable in the political situation, and the Foreign Office questioned it. An investigation showed that the G.C.&C.S. evaluator had placed his own interpretation on a not too clear text. From that time forward, G.C.&C.S. provided only the originals of solutions to the Admiralty.
In February 1939, two months after Germany announced that she would build as much tonnage in submarines as the British had, the nucleus evolved into the Operational Intelligence Centre (O.I.C.)—Section 8 of the Naval Intelligence Division. To head it, the director of naval intelligence named a retired contemporary, a one-time watch-keeper in Room 40, a man widely regarded as impossible to rattle, Rear Admiral John W. (Jock) Clayton. The O.I.C. had four sections: Italy and Japan, submarines, direction-finding, and surface ships and outside liaison, headed by Denning. Liaison with G.C.&C.S. was ensured by the presence in O.I.C. of a small party from Naval Section under Commander M. G. Saunders. The intercept and direction-finding stations were administered and controlled not by O.I.C. but by a combined signals division–intelligence division group, DSD/NID 9, under Commander Humphrey Sandwith. He and Denning pressed successfully to increase the number of intercept stations. Denning, seeing which way the wind was blowing, had a map drawn with Berlin at the center to give him the German point of view. Similarly, G.C.&C.S.’s Naval Section, which in 1937 had begun attacking German naval traffic for the first time since 1928, started a German subsection in May 1938 on Quex Sinclair’s recommendation. It consisted of an officer and a clerk, but no cryptanalysts for itself alone.
As the O.I.C. was coming into being, Sinclair, driven by a realization of the value of cryptanalytic information, the need for quiet in the work, the requirement for secrecy, and the likelihood of expansion in case of war, had a landed estate called Bletchley Park in the railroad junction town of Bletchley, some fifty miles northwest of London, purchased for G.C.&C.S. The estate consisted of a manor house greatly expanded in Victorian style in the 1870s by stockbroker Herbert S. Leon and the surrounding acres of landscaped grounds.
Since the house could not accommodate the hundred-odd employees, G.C.&C.S. built “huts”—long, narrow, peak-roofed, one-story wooden temporaries that huddled near the big house, which was headquarters. In July 1939, a few months after Britain had guaranteed to aid Poland in case of German aggression, a few weeks after Hitler and Mussolini had formalized the Axis, a few days after the Soviet Union had rejected a British-French proposal to block further Nazi aggression, Sinclair ordered G.C.&C.S. to move to Bletchley Park, later usually referred to as B.P.
Dillwyn Knox went to an office in former servants’ quarters called the Cottage, a small, square, brick building. There he pursued his work on Enigma with the cipher machine given him by the Poles. But though they had shown that it was possible to climb this cryptologic Everest, many difficulties remained. In particular, the naval Enigma resisted solution. Its keying system and the navy’s more careful use of it offered no handholds for the cryptanalysts.
A few weeks later, as German-Polish tensions rose, the German battlewagon Schleswig-Holstein visited Danzig. She was there purportedly for a twenty-fifth anniversary memorial service for the dead of the Magdeburg, who were buried in a Danzig cemetery. Early in the morning of September 1, 1939, she trained her 11-inch guns upon a Polish military depot 400 yards away. At 4:48 a.m., without warning, they fired the first shots of World War II. The depot’s wooden buildings burst into flames. Moments later, hundreds of miles away, Hitler’s armies crossed Poland’s borders. Britain’s guarantee went into effect. Two days later, the island kingdom, whose survival rested upon its ability to rule the sea, was at war—unable to read the main enemy’s main naval cipher.
7
PHANTOMS
FIVE DAYS AFTER HITLER’S ARMIES INVADED POLAND, THEY surged to within forty miles of Warsaw. The Poles did not need their Daniels—their cryptanalysts—to read the handwriting on this wall. The army high command prepared to evacuate. The codebreakers burned papers and packed Enigma replicas, bombes, perforated sheets, and Polish cipher machines in heavy crates. On the night of September 5 they quit their offices for the Vilna railroad station. The evacuation train, leaving the next evening, took three days to cover the 125 miles to Brzesc (Brest), to the east of Warsaw. There Rejewski, Różycki, and Zygalski continued their journey by car. The codebreaking equipment was destroyed. The trio drove south, first in a mobile direction-finding vehicle with Major Ciȩżki and his family, later in a tiny car, with Rejewski bumping uncomfortably atop some batteries. After six days of detours, rain, and frantic searches for gasoline, the group, which included Colonel Langer, arrived at the tiny village of Kuty at the tip of the tongue of Poland that then protruded into neutral Romania. The village, whose normal population was 400, was packed with a confused mass of soldiers, diplomats, bureaucrats, automobiles, and horse-drawn carts, all striving to squeeze onto the two-lane bridge over the Cheremosh River, which marked the border. The codebreakers crossed during the night of September 17.
The Romanian authorities confined the military personnel but left the civilians on their own, so the three young cryptanalysts caught a train to Bucharest. At the French embassy, they received visas promptly when they mentioned Bertrand, who had begun to search for them the day they entered Romania. Soon they were taking a succession of trains through Yugoslavia and Italy to France. They and a dozen other members of the Biuro Szyfrów arrived in Paris on October 1, some by special airplane, some by the Orient Express.
Five days later, the last of Poland’s forces surrendered to Germany. The defeat demonstrated an elemental point about intelligence: unlike guns or morale, it is a secondary factor in war. All the Polish codebreaking, all the heartrending efforts and the heroic successes, had helped the Polish military not at all. Intelligence can work only through strength.
But France was strong, or was regarded as such, and on October 20 the fifteen Polish cryptanalysts resumed their war against Germany, working under Bertrand. He now headed the so-called Section d’Examen (Examination Section), which performed radio intercept, traffic analysis, and cryptanalysis, of the General Staff’s Fifth Bureau, the mobilized form of the peacetime Service de Renseignements. As offices and billet for his section, Bertrand had requisitioned the Château de Vignolles, a large, three-story villa and associated outbuildings near the town of Gretz-Armainvilliers, some 25 miles northeast of Paris. The villa and its unit was called P.C. (“Poste de Commandement”) Bruno. Attached to it as what Bertrand called his Z Team were the Poles, who enrolled in the Polish army in France, and, as his D Team, seven émigré Spanish Republican cryptanalysts, whom Bertrand had enlisted in the Foreign Legion. Several dozen French cryptanalysts and support personnel brought the section’s total to seventy. Also present was Captain Kenneth Macfarlane, a British liaison officer, promptly nicknamed Pinky because of his rosy complexion; he had a direct teletypewriter line to Britain for exchanging results with Bletchley Park. The whole section ate meals together, and their differing tastes and temperaments led to sulks and arguments. Often, however, the problems were liquidated, so to speak, at the bar.
Rejewski and the others took up where they had left off. For weeks, the Enigma modifications of December 15, 1938, continued to defeat them. But toward the end of December 1939, assisted by a set of 1,560 Zygalski sheets that G.C.&C.S. had punched, P.C. Bruno recovered a German army key for October 28. A few weeks later, its solution of a Luftwaffe key for January 6, 1940, showed that the Wehrmacht had introduced no new procedures that would baffle the Allied cryptanalysts.
By that time Britain had achieved two remarkable breakthroughs that gave it the lead in Enigma cryptanalysis. These resulted from a shift in G.C.&C.S.’s thinking.
Toward the end of the 1930s, Alastair Denniston, the head of G.C.&C.S., had come to realize what the Poles had understood a decade earlier: that the shift to cipher machines required using mathematicians as cryptanalysts. So in the late summer
of 1938, as British Prime Minister Neville Chamberlain flew repeatedly to Germany to appease Hitler, the British codebreaking agency held a series of courses in cryptology primarily for mathematicians. The move brought into the closed world of British communications intelligence the fresh thinking of a mathematician of world-historical importance.
Alan Turing was a prodigy, a genius. A tallish, dark-haired, powerfully built man of twenty-seven with sunken cheeks and deep-set blue eyes, he wore unpressed clothes, picked at the flesh around his fingernails until it bled, stammered, fell into long silences, rarely made eye contact, sidled through doors, ran long-distance races, and had, by the time he arrived at B.P., made two fundamental contributions to knowledge.
He had been born in London in 1912, the son of an English administrator in the Indian Civil Service and a mother who rejoined her husband when Alan was a year old, leaving him and his older brother to be raised by a retired army colonel and his wife on the south coast of England. Despite an undistinguished career at his public school, Sherborne, Turing was bright enough to win a mathematics scholarship to King’s College, Cambridge. Upon his graduation, his brilliance was recognized by the college, which made him, at twenty-two, one of its forty-six fellows, enabling him to pursue his studies in mathematics.
That field was then in turmoil. A few years earlier, in 1931, the Czech mathematician Kurt Gödel had proved that contradictions would eventually arise in certain self-referential statements that would prevent some problems from being solved; mathematical knowledge would forever remain incomplete. Turing took this idea a step further in a paper published in 1936, when he was twenty-four. He began by envisioning a mechanism that could move to right or to left an infinitely long tape marked into squares and that could read and change or read and leave unchanged the blank or the mark—the 0 or the 1—in each square. He demonstrated that this machine could compute anything that could be calculated. But then he proved that even this device could not tell whether the potentially solvable problems could be solved.