THE CODEBREAKERS

by DAVID KAHN

  The nonsecrecy of the State Department codes inhibited negotiations with the moderates in Japan. Noted Grew in his diary on August 1, 1941: “Prince Konoye [the Japanese premier] knows that I would like to talk with him oftener, just as the President does with Admiral Nomura, but it is the fear of leakages and publicity which has prevented such interviews. It was indicated that any reports which our Embassy might send to Washington would of course become known to the Japanese authorities, although our informant said that he understood that we did have ‘one confidential code’ (highly significant, but I feel perfectly safe in the use of the one confidential code referred to).”

  On December 6, 1941, President Roosevelt dispatched his personal appeal for peace to the Emperor of Japan. He sent it to the State Department accompanied by a handwritten note on White House stationery: “Dear Cordell Shoot this to Grew—I think can go in gray code—saves time—I don’t mind if it gets picked up FDR.” His message was delivered to the embassy in Tokyo ten hours after it had been received in the Communications Ministry in Tokyo, and it is interesting that Grew, though he did not know of Roosevelt’s note, long thought that the use of the GRAY code had not saved but cost time because Japanese militarists had picked up the message, solved it, and deliberately detained it to frustrate any peace efforts. However, this was not so; the delay resulted from an embargo placed by the military on all incoming diplomatic messages.

  Among the interested readers of coded American diplomatic messages was the Reich Foreign Minister. Pers Z served him well. As early as 1925, it had studied the American system of superencipherment. The codewords were only of the CUCUC and CUCUC types; to encipher them, the code clerk split them into a single consonant and two CU or UC groups, then replaced these segments with substitutes from the appropriate tables. This superencipherment left the CUCUC and CUCUC configuration of the codegroup unchanged, and this regularity enabled the Pers Z mathematicians to break first into this original system and, in 1940, into a modification of it. Ironically, changes of super-encipherment within a message, intended to provide greater security, furnished the German cryptanalysts with isomorphic repetitions that helped them reconstitute the superencipherment substitution. With the superencipherment stripped off, the linguistic group solved a big 72,000-group code with not too much trouble. Dr. Hans-Kurt Müller was instrumental in this; he had an uncanny gift for seeing the outlines of the whole plaintext in the murk of the partial solutions. Miss Friedrichs assisted.

  President Roosevelt prescribes to the Secretary of State the code to be used to speed his personal appeal for peace to the Emperor of Japan the night before Pearl Harbor

  Pers Z solution of an encoded message of Robert Murphy to the State Department dealing with highly secret negotiations with General Weygand in North Africa in 1941

  They were greatly helped in their work by their knowledge of the activities of diplomat Robert Murphy, who in 1941 and 1942 was in North Africa, handling delicate negotiations with the Vichy French and paving the way for the Allied invasion of North Africa. Murphy insisted upon using the State Department codes to preserve his autonomy, even though American officers in Eisenhower’s command pointed out their insecurity. He was certain that the Germans had not broken his codes. In fact, however, the Pers Z cryptanalysts had broken them enough to recognize the groups meaning For Murphy or From Murphy that recurred at the head of so many telegrams. “We knew what he was interested in, and this gave us clues,” Miss Friedrichs said. These rapidly helped complete the solution of the big code. Murphy’s communications so facilitated her work, she said, that when she saw him drive by one day after the war while she was interned in Marburg, “I wanted to stop him and shake his hand.”

  Thus, as early as August 12, 1941, the state secretary of the Foreign Office could hand to von Ribbentrop fully solved copies of Murphy’s telegrams of July 21 and August 2. The first reported that Murphy had transmitted Roosevelt’s views on French North Africa to General Maxime Weygand, commanding there. The second transmitted a Weygand aide’s request for an American promise of military assistance. The Nazis knew Weygand was no friend of theirs, but it was not until they had what a Vichy source called “documentary proof” of his dealings with the United States that they forced Vichy to dismiss him. Thus the solution of an American diplomatic code cost the United States much valuable time and work that it was forced to recommence with the new leaders of French North Africa, and it may ultimately have prolonged the war and cost the lives of American soldiers who fought in that theater.

  A year later, the still-continuing German reading of coded American dispatches endangered the work of Allen W. Dulles, nominally a diplomat attached to the Bern legation but actually chief American spymaster in Europe. Dulles, who had recently begun to plot with anti-Nazis in Germany to overthrow Hitler, was unusually sensitive to the possibility of broken codes. In his cables, he referred to agents by codenames, which he changed frequently. He called the conspirators planning to assassinate Hitler BREAKERS. In February of 1943, one of them, Hans Bernd Gisevius, an official in the German consulate at Zurich, told Dulles that Germany had broken one of the American codes. Producing a little black notebook, he recited the gist of numerous telegrams from Bern to Washington. “Fortunately,” Dulles wrote, “it was not my own code and I had not used it for sending any operational messages, but as I was then short of code clerks I occasionally had fallen back on this particular code to send general political reports.”

  One of the messages that Gisevius read off had contained a fairly accurate portrait of the anti-German group in Italy. Even early in 1943, this was solidifying around Marshal Pietro Badoglio, who in July deposed Mussolini and formed a new anti-Fascist government, and around the prolific diarist Count Galeazzo Ciano, Mussolini’s son-in-law and Foreign Minister. Gisevius told Dulles that the American telegram had been laid on Hitler’s desk and then sent by the Führer to Mussolini with his compliments. A few days later Ciano was dismissed as Foreign Minister. “I never was able to discover,” Dulles wrote, “whether this was coincidence or whether this cable was the cause.”

  After Gisevius’ disclosure, Dulles, displaying great delicacy in intelligence operation, used the code “only for messages which we were quite willing or even anxious to have the Germans read, and over the months we discarded it entirely. To have stopped using it immediately would have told the Germans that we knew they had broken it.” Gisevius had feared that the intelligence the Germans might have obtained from the cryptanalysis would spoil the anti-Hitler program. But Dulles convinced him that he took every precaution in handling this information. The incident actually strengthened their collaboration, which culminated in the bomb plot of July 20, 1944, in which Hitler narrowly escaped with his life.

  Dulles thought that, after abandoning use of the broken code, “the Germans never succeeded in deciphering any of the messages I sent, and I had the satisfaction of knowing that no one who worked with me was ever jeopardized through deciphered telegrams. It was worrisome business, however, and I never put a cipher message on the air which gave specific facts about the underground without a feeling of apprehension.” His apprehensions were justified. For although the Germans may not have solved his cryptograms, the Hungarian Army unit under Major Bibo had. It fed its information to the R.S.H.A. through Wilhelm Höttl, but no concrete anticonspirator results seem to have been achieved, perhaps because of Dulles’ care in always using codenames.

  During these midwar years, President Roosevelt, still distrusting State Department codes, continued to rely on naval cryptosystems for his most secret messages. He exchanged hundreds with Churchill, who, signing himself “Former Naval Person” in recollection of his having been First Lord of the Admiralty (and no doubt with Roosevelt’s former assistant secretaryship of the Navy well in mind), said that “I sent my cables to the American Embassy in London, which was in direct touch with the President at the White House through special coding machines.” These machines were naval; they were very probably on
e-time tape devices manufactured by the Teletype Corporation.

  As the war progressed, the State Department gradually took the old solved codes out of service and replaced them with new cryptosystems. It thus choked off the German sources of information. To get them flowing again, Pers Z launched, in 1944, a major effort to break the M-138. The work was primarily mathematical, with Hans Rohrbach, a 37-year-old doctor of mathematics, playing a leading role. Rohrbach and Müller first divided the messages into “families” enciphered with the same strip arrangement, using repetitions as family resemblances. This meant that, in a given family, the first strip was always the same, the second was always the same, and so on. Stereotyped beginnings gave the cryptanalysts many plaintext assumptions—Müller was as adept at spotting words here as with the code. On each strip, the plaintext stood an unknown distance from the ciphertext. By comparing many such equivalents, both within a single strip and with the help of information from neighboring strips, the cryptanalysts mapped the letters on the strips to reproduce the original alphabet. Collaboration among the half-dozen cryptanalysts was extremely close. Each man looked after his own families, but they conferred frequently so that each could try on his own sequence of strips the possibilities found by others. Helping them in their work was a mechanism that moved the strips up and down to align them quickly. Eventually Pers Z recovered all the M-138 strips and read nearly all the messages. But by then they had lost much of their intelligence value, and any hopes that the solution would help in the future vanished when the strips were changed.

  Churchill expresses amazement at Roosevelt’s cryptographic subtlety in proposing to call himself “Admiral Q”

  The M-138 was, by then, no longer the topmost American diplomatic cryptosystem. The armed services had given the State Department cipher machines, including Vernam-system machines called SIGTOT, which the diplomats used for their most secret messages, though they still used codes for economy. American Army and Navy cryptanalysts taught State some of the practical lessons in cryptology that they had learned through their solutions. On June 3, 1944, Captain Lee W. Parke, the Navy cryptanalyst who had been senior watch officer in OP-20-GY at the time of Pearl Harbor, was detailed to the State Department, and on November 1 he became chief of the new Division of Cryptography. Under the direction of an expert cryptologist, the State Department communications at last took on a strength commensurate with that of the nation itself. The days of easy-to-break codes were ended. The era of American diplomatic cryptosecurity had begun.

  Among the characteristic features of World War II was the extensive use of codenames to designate important operations or secret projects. Codenames had been used before—the words “tank” and “blimp” themselves derive from World War I codenames—but never so frequently. They aimed both at security and brevity: obviously it was easier to say “Operation TORCH” than “the Anglo-American invasion of North Africa,” and solvers of any messages would still have to determine the meaning of the codenames.

  Selection and assignment of the codenames was, in the United States, a duty of the Current Section of the Army’s Operations Division. Men of the unit culled the unabridged dictionaries for suitable words—chiefly common nouns and adjectives that did not imply operations or localities. They avoided, as confusing, personal and ships’ names and geographical terms. Of the dictionaries’ 400,000 words, they compiled about 10,000 in scrambled order in a classified book. They cross-checked these to eliminate any conflicts with British codenames. Then they assigned blocks of codenames to theater commanders.

  In theory the codenames bore no relation, either by denotation or connotation, to what they stood for. In the majority of cases this held in practice. FLINTLOCK meant the Allied attack on the Marshall Islands in 1944; AVALANCHE, the amphibious attack on Salerno; ANVIL, later DRAGOON, the Anglo-American landings in the soft underbelly of France. Even relatively small operations were dubbed: the relief of Australians trapped in Tobruk was SUPERCHARGE, the occupation of the Canary Islands was PILGRIM. Some codenames were written in blood: OMAHA, UTAH, GOLD, SWORD, and JUNO, for the Normandy beaches of D-Day.

  Five of the codenames assigned for that cross-channel operation—themselves highly secret, as were their referents—inexplicably appeared in the crossword puzzles of the London Daily Telegraph in the month before June 6. Alarmed counterintelligence officials, fearing a Purloined-Letter type of concealment in what might be a warning to Germany, investigated—and found that the cause was merely an incredible coincidence.

  For minor operations the Germans usually selected Decknamen (“cover-names”) that did not suggest the operation: MERKUR (“Mercury”) for the seizure of Crete and FISCHREIHER (“Heron”) for Stalingrad. But in major operations they violated this precept. The Decknamen SEELÖWE (“Sea Lion”) for the invasion of England and HERBSTREISE (“Autumn Pleasure Voyage”) for the simultaneous feint of seaborne troops from Norwegian ports to northern England hardly obscured the secrecy of the operations they named. Least subtle of all was BARBAROSSA for the invasion of Russia. To be sure, “Barbarossa” was the nickname of the great medieval German king, Frederick I, but not only does it mean “Red Beard” in Italian, it also calls to mind one of Frederick’s greatest achievements in extending German authority over Slavs to the east.

  The Allies never were as obvious as that, but their selections were sometimes constrained by principles that that master of English, Winston Churchill, laid down in a memorandum of August 8, 1943:

  I have crossed out on the attached paper many unsuitable names. Operations in which large numbers of men may lose their lives ought not to be described by code-words which imply a boastful and overconfident sentiment, such as “Triumphant,” or, conversely, which are calculated to invest the plan with an air of despondency, such as “Woebetide,” “Massacre,” “Jumble,” “Trouble,” “Fidget,” “Flimsy,” “Pathetic,” and “Jaundice.” They ought not to be names of a frivolous character, such as “Bunnyhug,” “Billingsgate,” “Aperitif,” and “Ballyhoo.” They should not be ordinary words often used in other connections, such as “Flood,” “Smooth,” “Sudden,” “Supreme,” “Fullforce,” and “Full-speed.” Names of living people—Ministers or Commanders—should be avoided, e.g., “Bracken.”

  2. After all, the world is wide, and intelligent thought will readily supply an unlimited number of well-sounding names which do not suggest the character of the operation or disparage it in any way and do not enable some widow or mother to say that her son was killed in an operation called “Bunnyhug” or “Ballyhoo.”

  3. Proper names are good in this field. The heroes of antiquity, figures from Greek and Roman mythology, the constellations and stars, famous racehorses, names of British and American war heroes, could be used, provided they fall within the rules above. There are no doubt many other themes that could be suggested.

  4. Care should be taken in all this process. An efficient and a successful administration manifests itself equally in small as in great matters.

  Churchill himself had always manifested an interest in these matters, particularly where they involved nuances of meaning. “The name ‘Round-up’ has been given to the 1943 operation [proposed invasion of Europe],” he cabled to Roosevelt on July 6, 1942. “I do not much like this name, as it might be thought overconfident or overgloomy, but it has come into considerable use. Please let me know whether you have any wishes about this.” He complained to the chiefs of staff that it was “boastful, ill-chosen,” and hoped that “it does not bring us bad luck.” The codeword died a natural death (for a codeword) when the plan it designated was replaced by Operation GYMNAST, whose possible variations were indicated by appropriate modifications of the codeword. Churchill had strongly urged this operation, which was the invasion of North Africa, and after the Allies decided to go ahead with it, he “hastened to rechristen my favourite. ‘Gymnast,’ ‘Super-Gymnast’ and ‘Semi-Gymnast’ vanished from our code-words. On July 24 in an instruction from me to the Chiefs of Staff
‘Torch’ became the new and master term.”

  The Americans demonstrated a like sensitivity when they codenamed the crowning operations of the Pacific War, the invasion of Japan, CORONET and OLYMPIC. But it remained for Churchillian eloquence to find the great code-name of the war for the greatest operation of the war. The name evoked a sense of majesty and patriarchal vengeance and irresistible power for the supreme Allied effort to enter the continent of Europe and crush forever the wicked Nazi conspiracy. The master wordsmith himself consecrated that crusade with the codename Operation OVERLORD.

  Before that vast offensive could be mounted, the Allies had to win the Battle of the Atlantic. In this, communications intelligence played a role of high importance. Indeed, in some respects the Battle of the Atlantic might be viewed as a duel between the Axis and the Allied cryptanalytic organizations. And while Dönitz’ B-Dienst had its successes, the Allied communications-intelligence agencies enjoyed the advantage of access to the extremely heavy traffic of the U-boat fleet.

  In part, this stemmed from Dönitz’ insistence on maintaining tactical control of his submarines so as to concentrate them in wolf packs on the richest prizes. He was aware of the danger in all the talk, but, he contended, “The signals from the U-boats contained the information upon which was based the planning and control of those combined attacks which alone held the promise of really great success against the concentrated shipping of any enemy convoy.” His encouragement of communication led to an almost complete relaxation of radio discipline. U-boats went on the air to report a toothache on board or to congratulate a friend at headquarters on a birthday. U-boat command became “the most gabby military organization in all the history of war.”