THE CODEBREAKERS

by DAVID KAHN

  Thanks to Commander Laurance F. Safford, head of OP-20-G and father of the Navy’s communications-intelligence organization, the United States had, upon its entrance into the war, an Atlantic arc of high-frequency direction-finders to exploit the U-boat garrulity. Stations reported their bearings to their net control center in Maryland, whence they were flashed to the naval communications-intelligence organization at 3801 Nebraska Avenue, North West, in Washington. Commander Knight McMahon and his staff combined them into fixes and flashed these to the Atlantic Section of the Navy Commander in Chief’s Combat Intelligence Division. From here they sped to antisubmarine forces.

  How fast this net—called “huffduff” from the HF/DF abbreviation of “high-frequency direction-finding”—could work was shown by the episode of June 30, 1942. That morning, U-158 went on the air to report to Dönitz that he had nothing to report. Huffduff stations at Bermuda, Hartland Point, Kingston, and Georgetown heard him. McMahon plotted his position as latitude 33 degrees north, longitude 67 degrees 30 minutes west. This information raced down through channels until it reached Lieutenant Richard E. Schreder, U.S.N., flying an antisubmarine patrol out of Bermuda. Ten miles from the spotted location he found U-158 loafing on the surface, its crew sunbathing. One of Schreder’s depth charges landed on the submarine’s superstructure just as it was trying to dive. It went down all right, but it never came up.

  In another case, huffduff hounded a U-boat to death. The net first heard a transmission of U-66 on April 19,1944, and followed her successive messages in her attempts to rendezvous with a supply submarine. Allied ships, told where to go by huffduff, repeatedly frustrated these efforts, and on May 5 her commander wirelessed home: “Refueling impossible under constant stalking. Mid-Atlantic worse than Bay of Biscay.” Her “spurt” transmission—made by tape-recording the message and then radioing the tape at high speed—lasted less than 15 seconds, but no fewer than 26 huffduff stations got bearings on it, probably as a result of improved equipment that scanned the horizon 20 times a second and zeroed in accurately and semiautomatically on any emission. Three hours later, an American plane spotted the U-boat; an hour after that an American ship began to attack it, and within 25 minutes the submarine had gone down.

  In addition to huffduff, an intercept network eavesdropped on the text of the German messages. The Navy monitors could often tell one U-boat from another by the sending characteristics of their radio operators, and sometimes could ascertain the number of U-boats in a wolf pack. They grew so familiar with the submarine signals that they sometimes knew simply from external characteristics that a given message was a convoy contact report or a signal that attack had begun.

  Help was obtained from the most exciting code theft of World War II. It took place on the high seas with lightninglike speed under conditions of great peril.

  Early in 1944, Captain Daniel V. Gallery, U.S.N., commanding the antisubmarine Task Group 22.3, conceived a daring plan for boarding a U-boat and capturing it if, as sometimes happened, it surfaced after depth-charge damage to allow its crew to escape. Even though the plan as a whole might fail, he might pirate the submarine’s cryptographic equipment, which alone would make such a venture worthwhile. So he trained a team of volunteers in dismantling booby traps, closing sea cocks, and handling a U-boat.

  Messages in plaintext from Radio Logbook No. 6 of the captured U-505

  On May 31, 1944, he began tracking U-505, which huffduff had discovered was apparently heading for its home port at Brest. At 11 a.m. Sunday, June 4, a clear day with a light breeze, he made sound contact with the U-boat about 150 miles west of Cape Blanco, French West Africa. Its captain was at lunch when a salvo of depth charges slammed the peacefully gliding vessel, holing the outer hull and convincing him that his ship was mortally stricken. He blew his tanks and surfaced, and as his crew boiled out of hatches and the conning tower and leaped into the sea, U.S.S. Pillsbury was lowering a whale-boat carrying the boarding party.

  A few moments later, it reached the abandoned sub, rocking gently in the long Atlantic swells. Lieutenant (j.g.) Albert L. David, leading the boarding party, and petty officers Arthur K. Knispel and Stanley E. Wdowiak slipped through the hatch, raced forward to the radio room, smashed open a couple of lockers, and grabbed the cryptographic equipment—the current codebook with superencipherments, the cipher machine and its list of keys, and hundreds of messages with parallel plaintexts and ciphertexts. The Germans had apparently never considered the possibility of a boarding and so had not bothered to jettison the material. The three Americans hastily passed the items up on deck so that the team would have something to show for its efforts even if it lost the sub.

  But within fifteen minutes, the team had disconnected demolition charges and shut off an eight-inch stream of water, and U-505 had become the first enemy warship captured by a U.S. Navy boarding party since the War of 1812. Gallery put a line on her and towed her back to the United States, where she eventually wound up as a permanent display outside Chicago’s Museum of Science and Industry. David received a Congressional Medal of Honor for his heroism; his helpers were awarded Navy Crosses. The cryptographic material went to Nebraska Avenue. The crews of Task Group 22.3 maintained a discreet silence about their feat, and U-boat command, thinking that U-505 had been sunk, since no contact had been made after June 3, never suspected the truth and did not change its ciphers. Writer Ladislas Farago called the seizure “the climactic single episode of the American antisubmarine effort in the Atlantic.”

  The Allies now read U-boat operational traffic. For they had, more than a year before the theft, succeeded in solving the difficult U-boat systems and—in one of the finest cryptanalytic achievements of the war—managed to read the intercepts on a current basis. For this, the cryptanalysts needed the help of a mass of machinery that filled two buildings.

  What all this did to the submarines was graphically described by the German naval officer Harald Busch: “In the latter half of 1944 no U-boat commander would incur the ordeal of refueling if he could possibly avoid it…. on a suspiciously large number of occasions, enemy aircraft had made their appearance at the very moment when the pipeline was stretched between the two boats and neither was able to dive, with the result that many U-boats had been destroyed in the act of refueling…. Evidently U-boat commanders were right in their suspicions: the enemy could and did decipher the signals transmitted by Admiral Dönitz’ headquarters in Berlin.”

  In the eleven months remaining before the end of the European war, the Allies, greatly aided by the information that told them where to send their now powerful air and naval forces, sank nearly 300 U-boats—almost one a day—and greatly reduced their shipping losses. “Battles might be won or lost,” Churchill wrote, “enterprises might succeed or miscarry, territories might be gained or quitted, but dominating all our power to carry on the war, or even keep ourselves alive, lay our mastery of the ocean routes and the free approach and entry to our ports.” These the Allies mastered. “Reduced to the simplest terms,” wrote Farago in his study of the Battle of the Atlantic, “the Allies won the U-boat war and Germany lost it because Dönitz talked too much.”

  Final victory over the Nazi evil could come only by driving a military stake through its heart, and in this mission communications intelligence played an important role. The march actually began in North Africa in 1942 under the pressure for a “Second Front Now.” Communications-intelligence units were there—though not exactly in the role assigned them. Radio-intelligence companies of the American Army charged ashore as assault troops! They soon resumed their proper duties, however, and, equipped with intercept receivers and direction-finders, began to eavesdrop on the Axis messages. During the Tunisia campaign, the 128th, 117th, 122nd, 123rd, and 849th Signal Companies (Radio Intelligence) tracked the Germans all over North Africa and, by monitoring American communications, plugged leaks in Allied radio security. The 128th first discovered that the Germans were withdrawing from the Kasserine Pass, which they had taken
a few days earlier in America’s first blooding in Europe. Later the 128th gave advance warning of several enemy attacks. In Italy, the VI Corps intelligence officer said that his radio intelligence platoon had done “outstanding” work during the march on Rome and had supplied information second in value only to battle reconnaissance. Thus, even though the manning and equipping of radio intelligence companies did not get under way until relatively late in the war, officers in the field soon declared their product to be “of material value … at times vital” and praised the units as among the “most constantly profitable sources” of intelligence on German plans and movements.

  Strategic communications intelligence about German intentions in the European war mainly came, however, from Japanese sources. This should not be surprising. The Wehrmacht had the advantage of interior communications throughout occupied Europe and so could use wire networks, which offer very little opportunity for interception. But the Japanese diplomats in Berlin, Rome, Madrid, Lisbon, Sofia, Budapest, and Moscow had no way of getting messages back to Tokyo but by radio. These the Allies intercepted.

  The messages of the Japanese military attaches, whose code the United States had broken, yielded quantities of information. This source was lost to the Allies in 1943 in an ironic development that demonstrates the superiority of cryptanalysis over theft as a secret source of information. The Office of Strategic Services, America’s new spy outfit, in a laudable attempt at espionage, penetrated the offices of the Japanese embassy in Portugal. They did not disclose their plans to the Army, whose Signal Security Agency (formerly the Signal Intelligence Service) had broken the code; nor did the Army warn the O.S.S. against doing anything that would jeopardize its cryptanalyses. The upshot was that the Japanese discovered traces of the search, decided that their military attaché code might have been compromised, and changed it. The Allies, who had been comfortably reading the messages without benefit of espionage, still had not broken into the new code by the fall of 1944. Thus the attempt to gain information by cloak-and-dagger methods deprived the United States of information that it had been obtaining by the traceless means of communications intelligence.

  Bulky cipher machines such as the Japanese diplomats used could not be shipped or smuggled into blockaded Europe very easily, and so PURPLE remained in service throughout the war. Quite probably the Japanese considered the system secure. But even before Pearl Harbor American cryptanalysts were reading Japanese PURPLE messages from Berlin, and they preyed upon them even more avidly after the United States entered the war. Thus William F. Friedman’s solution of PURPLE reverberated throughout the war, leading to major effects and making it one of the world’s great cryptanalyses not only in technique but in importance as well.

  The Germans granted the Japanese ambassador, Baron Hiroshi Oshima, the intimacies of an ally, and, as a former military attaché, he took considerable interest in the military sphere. Toward the end of October, 1943, when it became evident that the Allies would invade Europe and the Wehrmacht had begun to stiffen its defenses, Oshima toured the Westwall and the Siegfried Line. He reported on these preparations in great detail in a long radiogram of between 1,000 and 2,000 words.

  As a powerful German station pumped it into the ether for the 5,000-mile leap to Tokyo, a new American intercept post at Asmara, in the former Italian colony of Eritrea bordering the Red Sea, picked it up. Back the cryptogram went to the Signal Security Agency. It proved to be in PURPLE, which the American cryptanalysts read with relative ease. The solution went to General Dwight D. Eisenhower’s headquarters, where its intelligence helped shape basic strategy for the conquest of Germany.

  The success of the invasion stemmed in part from its geographical surprise, and that surprise stemmed largely from an elaborate deception by the Allies, in which radio played the major role. To distract German attention from the real landing area in Normandy, Eisenhower’s headquarters cooked up a complete cover-plan codenamed FORTITUDE. Parts of it had gone into operation more than a year before the invasion. Field Marshal Montgomery’s radio messages were not broadcast from his actual location in the south of England, but were led by land line to a spoof headquarters near Dover and transmitted from there. Dummy ships were concentrated in the Cinque Ports to help the illusion. A very busy signals staff contrived, by sending out the right sort of dummy wireless traffic, to “assemble” a fictitious 4th Army in Scotland. The “wireless training” of this army contained some purposeful indiscretions. By these furtive, impressionistic, and devious indirections, FORTITUDE sought to let the Germans convince themselves of what they had always wanted to believe anyway—that the invaders would pour across the Channel at the narrowest point, from Dover to the Pas de Calais; the build-up in Scotland suggested a preliminary feintlike assault on southern Norway. ‘The final result was admirable,” Churchill wrote. “The German High Command firmly believed the evidence we obligingly put at their disposal.” In fact, so conclusive did the evidence seem to be that more than a month after the invasion in Normandy, Hitler declared that “the enemy will probably attempt a second landing in the 15th Army’s sector”—which was the Pas de Calais!

  On the Normandy beachhead, the solution of a German message enabled General Omar Bradley’s 12th Army Group “to meet a very strong attack against one of our weaker positions,” he said. His cryptanalytical unit was the 849th Signal Intelligence Service (formerly the radio intelligence company of the same number). Though attached to the army group, it worked not at headquarters but in the field, intercepting and cryptanalyzing German messages from the level of the army group down to company. The material was tactical and of great value in day-to-day operations.

  In the fall the 849th moved into winter quarters in a building in Luxembourg. About the first of December it began to read German messages indicating the movement of armored divisions behind the Ardennes forest. These increased from day to day, but G-2, the recipient of this information, appeared to take no notice whatever. Finally, on the Sunday morning of December 17, the unit solved messages that confirmed what the cryptanalysts had long dreaded: the German Panzer attack. Why had the army commanders not heeded the intelligence—which included a great variety of other indications? They did not believe that the Germans would or could attack with armor through so heavily wooded, so hilly, and so generally unfavorable a terrain for tanks. The Battle of the Bulge has long been cited as a failure of intelligence. It was not the intelligence that failed, however. The Bulge was a failure of evaluation.

  Attached to Lieutenant General George S. Patton Jr.’s 3rd Army was a similar radio intelligence unit, commanded by Major Charles Flint, regarded by one fellow officer as “a young, trigger-smart expert.” Flint’s outfit was particularly valuable in fluid situations when the Germans were on the move and had to use radio. At Bastogne, it solved a message that enabled Patton to inflict heavy losses on the redoubtable 5th Para Division. Like other S.I.S. units, Flint’s monitored American traffic; it once warned a mechanized-cavalry transmitter of a communications-security violation that might have revealed important information to the enemy.

  How highly the Allies regarded communications intelligence was demonstrated in a left-handed fashion in the closing days of the war by the case of the missing SIGABA. Also known as the M-134-C, the SIGABA, which had been devised by William Friedman, was a rotor machine like the German Enigma and British TYPEX; like them, it protected top-level communications—an interesting case of parallel evolution and an indication of the cryptographic success of the rotor. And it protected them exceedingly well. The branch of the Army’s Signal Security Agency charged with testing American crypto-systems had failed in all its efforts to break down messages enciphered in M-134-C. And, though the United States did not know it at the time, German cryptanalysts had, despite prolonged efforts, likewise found it impossible to read these cryptograms.

  But all the cryptographic ingenuity built into the machines would have been expended in vain had only one of them fallen—even briefly—into the ha
nds of the enemy. As a result, probably nothing in the war zones were guarded as closely as the ABA’s, as they were called for short. Some units close to the front moved them to the rear each night. Heavy safes protected them when not in use—one safe for the basic mechanism, another for the rotors (for additional key changes, the machine came equipped with more rotors than the five it used at a time), a third, apparently, for the key lists. Armed guards watched over them continuously. The precautions seemed to be paying off, for to the Army’s knowledge, not one ABA had ever been compromised.

  But on the night of February 3, 1945, two ABA guards, sergeants of the 28th Division, parked the truck in which they were transporting the three safes of a divisional ABA outside a house in Colmar, France, and paid a brief visit to some friendly ladies. When they emerged, the truck and its safes were gone. Counterintelligence began at once to search for it. On the side of a road near Colmar, agents found a trailer that had been attached to the truck—but no truck, no safes, and no ABA.

  Panic struck. At Eisenhower’s headquarters, security and cryptologic officers went frantic. Ike himself was concerned. Colmar, which had just been liberated, was still close enough to the front for German collaborators or agents planted by the retreating Wermacht to have stolen the ABA’s. Somehow they might have gotten through the fluid front lines and so to German cryptanalysts. These would then be in a position to do what they had not been able to do before. For with a knowledge of the wiring of the ABA rotors, the heart of the system, and with a working mechanism to complete their understanding, all the German cryptanalysts would have to do would be to determine which rotors in the set had been used for a particular message and their initial setting. This would not be easy, but it certainly could be done.