THE CODEBREAKERS

by DAVID KAHN

  So far the Japanese had sighted no American ships. They had not been diligently looking for them because, according to their expectations, no major enemy forces should have been in the vicinity: they should have been in Pearl, waiting to find out where the Japanese would strike. Admiral Chuichi Nagumo therefore struck below the 93 planes he had prudently held to counter even the highly unlikely enemy naval attack and ordered them rearmed for land bombardment. Thirteen minutes later he was dumfounded to receive a report of the sighting of enemy ships to the northeast. What should he do? For a precious quarter of an hour he mulled it over. Finally he canceled his order and directed the planes readied to attack ships. The incendiary and fragmentation bombs that the crews had just sweated into the bomb bays had to be replaced with the original torpedoes and armor-piercing bombs. Before this work was completed, his airplanes began returning from Midway, and his carriers had to recover these before launching the others.

  It was at this most vulnerable of moments—with all planes aboard, with fueling in process and bombs and ammunition stacked in the open on the hangar and flight decks—that American planes attacked. Three waves of torpedo-bombers, one each from Hornet, Enterprise, and Yorktown, swept in, suffered heavy losses under Zero attacks or antiaircraft fire, and scored not a single hit. The last plane zoomed away at 10:24 a.m. This moment marked the high tide of Japan’s fortunes in World War II. Jubilant officers cheered what they thought was victory at Midway, and in the war. Within six minutes, the tide was ebbing.

  Dive-bombers from Enterprise screamed down on Akagi, Kaga, and Soryu. One hit set off Akagi’s torpedo storage, another exploded amid planes being rearmed on her flight deck; flames swept her, and within 24 hours she had been sunk. Kaga took four hits in rapid succession and sank that evening. Yorktown dive-bombers pummeled Soryu with three half-ton bombs; within 20 minutes she had to be abandoned, and a few hours later was torpedoed by an American submarine. The work of December 7 had not been completed, but avenged.

  The rest was anticlimax. Later in the day Hiryu was sunk, and the Japanese in turn got Yorktown. Yamamoto next day realized that he was beaten. He called off the invasion of Midway and retreated, keeping close to his cabin on the homeward voyage. The samurai chieftains canceled plans for further advances and shifted from offense to defense. The failure to destroy the American Navy knocked the keystone from Yamamoto’s strategy, and his words to Prince Konoye before the war haunted him: “I must also tell you that, should the war be prolonged for two or three years, I have no confidence in our ultimate victory.” And not only did American industrial strength rise up like a specter. Japan’s lack of it meant that she would never recover from the loss of four big carriers. The 4th of June had doomed her.

  “Midway was essentially a victory of intelligence,” Nimitz has written. “In attempting surprise, the Japanese were themselves surprised.” General Marshall was even more specific. As a result of cryptanalysis, he declared, “We were able to concentrate our limited forces to meet their naval advance on Midway when otherwise we almost certainly would have been some 3,000 miles out of place.” The surprise, the concentration, were engineered days before in a basement office a thousand miles from the scene of the action, where the solution of messages in JN25b (abetted by the recoveries of the other cryptanalytic units) and its internal time and place ciphers forged effects more crucial to the course of history than any other solution except that of the Zimmermann telegram. The codebreakers of the Combat Intelligence Unit had engrossed the fate of a nation. They had determined the destinies of ships and men. They had turned the tide of a war. They had caused a Rising Sun to start to set.

  There was no single moment when the Battle of Midway was suddenly and decisively won, and so there was no burst of wild cheering in the basement office. The cryptanalysts reacted prosaically. The unit went on a watch in three instead of a watch and watch. It was also expanding rapidly. By the next year, it had changed its name to Fleet Radio Unit, Pacific Fleet—FRUPAC, in the Navy’s interminable list of acronyms. Rochefort had departed in October, 1942, for two years of noncryptologic duties. He was replaced by Captain William B. Goggins, 44, a 1919 Annapolis graduate with long communications experience. Goggins, who had been wounded in the Battle of the Java Sea, remained as head of FRUPAC to January, 1945. Dyer continued to head cryptanalysis. Eventually FRUPAC comprised a personnel of more than 1,000. Much of the work was done in the new Joint Intelligence Center, housed in a long narrow building across Midway Drive from Nimitz’ headquarters perched atop a cliff overlooking Pearl Harbor. Fabian, in Melbourne, directed a field unit similar to FRUPAC. He was on the staff of the Commander in Chief, 7th Fleet, which was attached to MacArthur’s South West Pacific Area command.

  FRUPAC’s growth mirrored that of all American cryptanalytic agencies. This expansion compelled OP-20-G to reorganize as early as February, 1942. The workload had become too heavy for one man (Safford). The outfit was split up into sections for its three major cryptologic functions: (1) the development, production, and distribution of naval cryptosystems, headed by Safford; (2) policing of American naval communications to correct and prevent security violations; (3) cryptanalysis, headed by Commander John Redman. In September the development function was separated from the production. Safford retained control of the development work until the end of the war, devising such new devices as call-sign cipher machines, adapters for British and other cryptographic devices, and off-line equipment for automatic operation. About June, the Navy ceded Japanese diplomatic solutions to the Army, giving over its files as well as its PURPLE machine. So rapidly was the workload expanding, however, that this diminution of its responsibility did not prevent the cryptologic branch from bursting the seams of its Navy Department building offices. In 1942, it moved into the brick buildings of a former girls’ school at 3801 Nebraska Avenue, at the corner of Massachusetts Avenue, in a quiet section of northwest Washington. In the fall of 1941, about 700 persons, including 80 officers, had been doing communications intelligence in the entire Navy; two thirds of them were intercepting, direction-finding, or training for that work; the others, including most of the officers, were solving and translating. By the end of the war, there were 6,000.

  The Army’s growth was even more spectacular. It multiplied its communications-intelligence manpower thirtyfold from its strength December 7, 1941, of 331-44 officers and 137 enlisted men and civilians in Washington, and 150 officers and men in the field. Ever-growing requirements quickly dwarfed early estimates, such as the one early in 1942 that a staff of 460 would suffice, and kept up a relentless pressure for more and still more workers. Yet the agency faced stiff competition for them in manpower-short Washington. Moreover, the necessity for employees to be of unquestioned loyalty and trustworthiness, because of the sensitive nature of cryptanalytic results, and the importance of their being temperamentally suited to the highly specialized nature of the work, greatly reduced the number of prospects. To fill its needs, the agency launched a series of vigorous but discreet recruiting drives. It snatched people out of its school even though they were only partially trained: during the school’s entire time at Fort Monmouth, New Jersey, not one student completed the full 48-week course. It brought in members of the Women’s Army Corps—almost 1,500 of them. These measures enabled the agency to grow to a strength of 10,609 at its peak on June 1, 1945—5,565 civilians, 4,428 enlisted men and W.A.C.’s and 796 officers. (This figure excludes cryptologic personnel serving under theater commanders overseas.) Nevertheless, the personnel supply never caught up to the demand. In April, 1944, for example, the agency had more than 1,000 civilian positions empty.

  But its growth soon made more space necessary. Like the Navy, it found a former girls’ school ideal for its purposes. During the summer of 1942, it moved from the Munitions Building to Arlington Hall, whose brick buildings stood on 58 wooded acres fronting on Glebe Road in Arlington, Virginia, about three miles from downtown Washington and away from the eyes of enemy agents. The agency s
oon outgrew even this, and in the late fall of 1942 began expanding into Vint Hill Farms, an old estate in the Virginia horse country about 50 miles from Washington. Giant intercepting towers and half a dozen ugly barracks-like buildings soon disfigured the lovely Blue Ridge foothills, and here, in rooms filled with desks with tilted tops, most of the Army’s traffic analysis was done. In addition, the agency taught most of its cryptology here, with the removal of its school from Fort Monmouth in October, 1942.

  In June of 1942, owing to a reorganization in the Office of the Chief Signal Officer, the outfit shed its old name of Signal Intelligence Service and gained and lost three new ones within two months. Then from July, 1942, to July, 1943, it was called the Signal Security Service, and from July, 1943, to the end of the war, the Signal Security Agency. Lieutenant Colonel Rex Minckler, chief since before Pearl Harbor, was replaced in April, 1942, by Lieutenant Colonel Frank W. Bullock. In February, 1943, Lieutenant Colonel W. Preston (Red) Corderman, tall, husky, quiet, pleasant, who had studied and then taught in the S.I.S. school in the 1930s, became chief. He remained in the post to the end of the war, rising to a brigadier general in June, 1945.

  Its population explosion and its voluminous output strained its administrative structure, and this was realigned several times. As of Pearl Harbor it was divided into four sections: the A, or administrative; the B, or cryptanalytic; the C, or cryptographic, and the D, or laboratory.

  The C, or cryptographic section, devised hundreds of codes and ciphers during the war and produced thousands of key lists. It printed 5,000,000 classified documents, some running to many pages each, and distributed them in a carefully guarded manner throughout the world, accounting for each one. It tested the security of Army cipher machines (mainly Friedman’s M-134 SIGABA) by attempting to solve them—and found that they generally proved impregnable. It supervised the mechanization of Army cryptography—the increasing replacement of strip cipher and M-209s and similar slow systems with typewriter-keyboard cipher machines, often with an on-line capacity. Only such mechanization enabled Army cryptographers to keep up with the ever-rising flood of traffic: the 23,000 codegroups a day that the 5th Army headquarters processed during its Sicily campaign strained even the machines to their limit—and by the time that army was marching on Rome, its headquarters was handling 40,000 groups a day. Traffic volume passed belief: in Hollandia, a million groups a day in November, 1944; at the Army’s European Theater of Operations headquarters even before OVERLORD, 1,500,000 to 2,000,000 groups a day, or the equivalent of a shelf of 20 average books. The biggest message center of all, the War Department’s in Washington, handled its peak load on August 8, 1945: nearly 9,500,000 words, the equivalent of almost one-tenth the total of French intercepts in all of World War I.

  In August of 1942, subsection 6 (traffic) of the cryptanalytic section was upgraded to an E, or communications, section, to disseminate the solutions and to send directives to the field intercept units. In December, the shop of the cryptographic section was set up as the nucleus of the F, or development, section, for cryptographic equipment. In March of 1943, all six sections were elevated to branches, and by the following year two more had been added: the machine and the information and liaison branches. The Army had begun to use machines for cryptology in 1936, when Hollerith tabulating machines facilitated the compiling of codes. Their cryptanalytic potential had also been noted in that same year. By Pearl Harbor, 13 I.B.M. machines tended by 21 operators were working on S.I.S. projects. The personnel-short agency converted as many tasks as possible to mechanical operation, and the G, or machine, branch grew to enormous proportions. The 407 machines and 1,275 operators that it had by the spring of 1945 handled accounting and cryptologic tasks that would otherwise have required the hand labor of impossible numbers of clerks.

  The cryptanalytic branch, then headed by Solomon Kullback, one of the three original cryptanalysts hired by Friedman in 1930, was much the largest, with 2,574 people in July of 1944, 82 per cent working on Japanese Army messages. To balance the agency and reduce the number of branch chiefs reporting to its commanding officer, the agency was reorganized the following month into four divisions: intelligence, which did traffic analysis and cryptanalysis; security, which handled cryptography and radio counter-measures and formulated and executed policy and technical doctrines; operating services, which provided services for the intelligence and security divisions and ran the secret-ink laboratory; and personnel and training.

  Though this set-up held until the war ended, operational control of the agency passed on December 15, 1944, to G-2, the military intelligence section of the War Department General Staff, which was the agency’s major customer and which, as such, for many months had indirectly guided its activities. The Signal Corps merely retained administrative control. This confusing arrangement—complicated further by the agency’s having both staff and command functions—ended in August, 1945, when the War Department transferred all signal intelligence units to agency control. On September 6, four days after the war ended, the War Department ordered the creation within G-2 of a new cryptologic organization by merging the Signal Security Agency, the field cryptanalytic units, and Signal Corps cryptography. This was the Army Security Agency, which came into existence September 15, 1945.

  Throughout the war, most of the intercept material for Signal Security Agency headquarters was supplied by the 2nd Signal Service Battalion. It had been created as the 2nd Signal Service Company on January 1, 1939, by Major General Joseph Mauborgne, the chief signal officer, out of the 1st Radio Intelligence Company at Fort Monmouth, plus the radio intelligence detachments of signal companies in the Canal Zone, Fort Sam Houston, Texas, the Presidio, San Francisco, Fort Shafter, Hawaii and Fort McKinley, Philippine Islands. Commanding its 101 enlisted men was First Lieutenant Earle F. Cooke. It grew rapidly—in October, 1939, a detachment under First Lieutenant Robert E. Schukraft arrived at Fort Hunt, Virginia, to install and operate a new Army intercept station. With the onset of war, the imperative demands for manpower compelled the Army, on April 2, 1942, to increase the company to battalion strength. Eventually it expanded to an enormously oversized company of 5,000 men. From April, 1942, to the end of the war, its commanding officer was the Signal Security Agency chief. When G-2 took operational control, the battalion was redesignated the 9420th Technical Service Unit, which at the end of the war became part of the Army Security Agency. By that time, the original four radio circuits on which it was sending intercept material back to Washington at the time of Pearl Harbor had swollen to 46 full-time radioteletypewriter channels.

  The Army, like the Navy, established cryptanalytic units in the several theaters of war. Their organization varied from one theater to another. The South West Pacific Area, under MacArthur, had at its headquarters a communications-intelligence unit called the Central Bureau and in the field a number of subordinate units. Central Bureau, or simply C.B., had been founded in August of 1942 by Lieutenant Colonel Joe R. Sherr, who had been head of the 18-man 2nd Signal Service Company detachment in the Philippines and who had accompanied MacArthur to Australia. Later, Abraham Sinkov, who had been another of Friedman’s original crypt-analysts, went out to take charge. C.B. was quartered in a rambling wooden house—which local legend said was a former whorehouse—close to the Ascot racetrack in Brisbane. A guard stood in front. A small air-conditioned brick building at the track itself housed the I.B.M. machinery. Sinkov worked wonders: when a downed Japanese bomber yielded an air-to-ground codebook, it was discovered that Sinkov had already recovered nearly all of it. His title at the end of the war was Cryptanalytic Officer, Signal Intelligence Service, U.S. Army Forces, Far East; his rank by then was colonel. A sweet and unmilitary man who seemed slightly embarrassed by the eagles on his shoulders, he was unable to return a salute without blurting out a “Good morning.” He was awarded a Legion of Merit and an Oak Leaf Cluster to it for his work.

  Some elements of the Central Bureau were—despite the name—attached to widely scat
tered units. MacArthur’s chief signal officer, Brigadier General Spencer B. Akin, who enjoyed more authority than any other theater signal officer, attached communications-intelligence units to major headquarters so that the intelligence would be promptly available to officers who could act upon it. He even assigned one such detachment to Admiral William F. Halsey, Jr.’s flagship, while Admiral Spruance found the Army service so valuable, when he took command of the 5th Fleet, that he kept the communications-intelligence specialists with him.

  In addition, Signal Corps radio intelligence companies provided tactical, combat-level communications intelligence. One of the first, the 101st Signal Company (Radio Intelligence), replaced Hawaii’s old Monitor Post 5 in July of 1942, vastly improving the quantity and quality of the work. Typical, perhaps, of these companies was the 138th. Trained in Spokane for Europe and then transported to the East Coast, it was loaded aboard a transport and promptly shipped through the Panama Canal to Australia, landing there in June of 1943. The 299-man company was mobile and self-contained so that it could operate in isolation: it was mountable within two hours and had its own truckdrivers, cooks, repairmen, and so forth. The men lived in tents.

  The company’s mission was to determine the Japanese order of battle and ascertain military concentrations and movements. Most of its work involved air-to-ground messages. To pick up these low-power transmissions, it had to move forward from island to island as the Allies advanced. Its first position, early in 1944, was at Nadzab, an airstrip in the Markham Valley of New Guinea. One subordinate direction-finding group was over a hump at Gusap; another was on an abandoned ranch near Darwin, Australia, where it enjoyed fresh meat daily. In the middle of the year it advanced to Biak, a small island north of New Guinea, where it was nearly strangled by the thick jungle, and it went ashore on Leyte about five days after the first wave of invasion troops. By then its direction-finding groups were scattered all over the South Pacific.