THE CODEBREAKERS

by DAVID KAHN

  Much of Bell Labs’ expertise was developed in testing Allied scrambles, particularly a highly regarded combination of band-splitting and T.D.S. This was called a 2-dimensional, or 2-D, scramble because it operated along both the vertical (frequency) and the horizontal (time) axes. With its trusty spectrograph, the Bell Labs defeated even the best of these, the British 2-D privacy system. The reconstitution of speech patterns from the many little rectangles in the spectrogram was a tedious job but surprisingly quick—a crew of six men would take from two to three hours to do it. The experience suggested some humble but effective ways of increasing privacy: speaking in a low-pitched monotone to diminish speech patterns, varying the length and cycle of T.D.S. elements to eliminate periodicity, and adding noise after scrambling the speech. The noise would not bother the ear, but it would spatter the spectrogram with false lines that would mislead the solver. In fact, noise of the right kind might make T.D.S. or 2-D spectrograms appear more continuous in their scrambled order than in the order that unscrambles the speech but scrambles the noise!

  Ciphony never attained the security of written communications. Cryptologic terminology reflects this difference by calling scramblers “privacy” systems and not “secrecy” systems. Late in the war, Roosevelt and Churchill recognized this gauzelike security and switched from the telephone to teletype talks, enciphered by a little box called “Telekrypton” that was almost certainly the one-time tape. Nevertheless, ciphony made such great strides during the conflict that General George C. Marshall, whose fear of using a scrambler had had such dire effects on December 7, 1941, could say three years later, “We have the very finest equipment now.”

  * At the start of the war in September, 1939, the Allies prohibited the use of any codes at all. But pressure of business houses and realization that commercially coded messages were only a step up from plaintext forced them to relent, and at the end of December they permitted the use of Bentley’s Complete Phrase Code, Bentley’s Second Phrase Code, the ABC Code (6th edition), and Peterson’s Code (3rd edition). In April, 1940, they admitted five more codes: Acme Code and Supplement, Lombard General Code, Lombard Shipping Code and Appendix, New Standard Half Word Code, and New Standard Three Letter Code. These were the nine later allowed by the United States and most of the Latin American nations. Under pressure from the Allies, Argentina, which had not severed diplomatic relations with the Axis, halted all code communications—but the first code message stopped was one from the Vatican! During the war, even neutrals such as Spain and Sweden demanded copies of the codes used and prohibited the use of (secret) cipher. Only Switzerland placed no restrictions on either code or cipher communication.

  * The R.S.H.A., whose Section IV E was charged with domestic counterespionage and frequently clashed with the Abwehr’s Section III F over their differing views of their almost indistinguishable responsibilities, called it the ENGLANDSPIEL, the name by which it is perhaps more widely known. The R.S.H.A. felt it more important to uproot an underground network than to play back its afu and this led to several conflicts with the Abwehr. Giskes finally agreed with his opposite number in Section IV E-Netherlands, Sturmbannführer Joseph Schreieder, that the Abwehr would handle matters in the ether while the R.S.H.A. would make arrests on the ground. The rivalry between the two, which hampered German counterespionage, reflected the high-level power struggle between Himmler, the R.S.H.A. overseer, and Admiral Wilhelm Canaris, chief of the Abwehr, which Himmler eventually won when the R.S.H.A. absorbed the Abwehr.

  * Apparently they never noticed that neither the third captured message nor the previously intercepted ones fitted this pattern, or perhaps they thought that Lauwers had made some errors.

  * It does not seem possible to devise a scrambler that distorts the sound itself (i.e., the vibrations in the air) because, once the waves were degraded by, say, some kind of baffle, they could not be restored to their original form. Transposition systems, on the other hand, might be possible in a very crude form by means of mechanical phonographs. From a practical point of view, however, nonelectrical scramblers may be ruled out. None ever seems to have been constructed.

  17

  THE SCRUTABLE ORIENTALS

  FROM THE SUNDAY MORNING when Commander Mitsuo Fuchida, in his bomber high over Pearl Harbor, radioed “TORA, TORA, TORA!” to indicate that his attack force had achieved complete surprise, the gods of war had smiled without surcease upon the armed forces of imperial Japan. The strike at Pearl Harbor had decimated the United States fleet. Unhindered, the Greater East Asia Co-Prosperity Sphere expanded rapidly and uninterruptedly. Guam was captured on December 10, Wake on the 23rd. Two days later Hong Kong fell. Japanese aircraft sank Prince of Wales and Repulse, giving Winston Churchill his worst shock of the war and leaving the whole western Pacific, the Indian Ocean, Oceania, and even Australia virtually undefended by naval forces. Tojo’s armies overran Singapore and Malaya with its rubber plantations, then the Dutch East Indies with its great oil fields. Siam and the Solomons were in their hands. China was under blockade. In May the Philippines surrendered. Within six stupefying months, the Rising Sun shone upon nearly a tenth of the globe’s surface. Nippon’s enemies had been wiped from the seas. Her troops raped and pillaged from bustling Rangoon to the languorous South Sea islands. It was the most rapid conquest in history.

  It amply fulfilled the Japanese war plan. Japan did not intend to invade the United States. Rather, she planned to feed upon the riches of the conquered territories behind a ring of impregnable defense positions, from which she would beat off any attacker. But the high command, bedazzled by success and greedy for more, decided instead to continue the sweep before its momentum was lost. The admirals and generals pointed out that naval losses, which they had anticipated at 25 per cent, had been infinitesimal. The largest ship sunk had been a destroyer, and so more than adequate forces remained for the new drive. Furthermore, they reasoned, the defense perimeter would be protected as much by greater depth as by greater consolidation. They therefore set in motion two ambitious plans. One was an amphibious assault southward to Port Moresby, a town on the southeastern tip of New Guinea only 400 miles from Australia. The other pivoted on Midway, a tiny atoll in the middle of the Pacific that stood as a sentinel to Hawaii.

  This second plan had two parts. The first part aimed at the atoll’s capture. Its two coral islets—the larger barely two miles long—possessed no intrinsic worth but great strategic value, for whoever held them controlled the central Pacific and hence the approaches to either end of the oceanic basin. The second and more important part of the plan sought to lure out the remainder of the American fleet and destroy it. Admiral Isoroku Yamamoto, Commander in Chief of Japan’s Combined Fleet, appreciated America’s industrial might and realized that Japan had to win quickly—before America could bring it to bear. He also knew that the United States could not let Midway go by default, as it had Wake and Guam. When the Pacific Fleet, enfeebled by the losses at Pearl Harbor, steamed out to defend the atoll, he would fall upon it with his vastly superior forces and annihilate it. This final disaster would convince Americans that Japan could not be beaten. They would therefore quit a pointless struggle and leave Japan master of the western Pacific. Or so the warlords purposed.

  They did not know that the United States had fashioned a secret weapon of such potency that it could alter the balance of power in the Pacific. It was located in the long, narrow, windowless basement of the 14th Naval District’s Administration Building in the Navy yard at Pearl Harbor. Vaultlike doors protected its secrets; steel-barred gates at the top and bottom of the stairs kept out visitors; guards stood a round-the-clock watch. This office was staffed, when the war broke out, with about 30 officers and men. It was equipped with International Business Machine Corporation tabulators, which were partitioned off in a separate section because of the racket they made. Its raw material came in by courier from the radio intercept station at Wailupe. This was the so-called Combat Intelligence Unit, the radio intelligence orga
nization that served the Pacific Fleet.

  Lieutenant Commander Joseph John Rochefort had commanded it since May of 1941. Before Pearl Harbor, the bulk of its personnel worked on interception, direction-finding, and traffic analysis; the unit fed these results to the fleet intelligence officer. Though one of its young cryptanalysts, Chief Radioman Farnsley C. Woodward, had attacked the Japanese diplomatic systems in use by the Honolulu consulate as a favor for counterintelligence, the unit’s main cryptanalytic duties before Pearl Harbor involved the solution of the Japanese flag officers’ system and miscellaneous administrative, personnel, and meteorological codes. It had only three real cryptanalysts to handle this workload, Rochefort and Lieutenant Commanders Thomas H. Dyer and Wesley A. Wright. The others were trainees, aides, clerks, and translators. Since August of 1941 it had been working a seven-day week; in October it went to a night watch as well—the only unit in Pearl to do so.

  Three days after Pearl Harbor the unit was given a major change in assignment. It was to discontinue work on the flag officers’ system (which was to be analyzed in OP-20-G in the Navy Department in Washington) and to join in the attack and breakdown of the Japanese fleet cryptographic system, dubbed JN25 by OP-20-G. This most widely distributed and extensively used of Japan’s naval cryptosystems, in which about half her naval messages were transmitted, was already the target of three other cryptanalytical units—a 16th Naval District group under Lieutenant Rudolph J. Fabian on Corregidor, a British group at Singapore, and OP-20-G. They had determined that it was a two-part code of about 45,000 five-digit groups, enciphered by two volumes of 50,000 five-digit additives each. The b, or second, edition had come into force on December 1, 1940, and by the following November messages in it were partly readable. At 6 a.m. on December 4, 1941, new additive books came into effect, together with new indicators. Fabian’s group broke into this new encipherment four days later, and by Christmas messages were again being read as before. But these readings were tantalizingly fragmentary, and much remained to be done.

  The commencement of hostilities generated an enormous increase in radio traffic and consequently in the workload of the Combat Intelligence Unit. To handle it, the unit dragooned personnel from every possible source. It first acquired the band of the U.S.S. California, which had been badly damaged in the first few minutes of the air attack. Dyer threw up his hands when he heard about it, but music and mathematics and cryptanalysis seemed to go together,* and nearly all the bandsmen proved above average and some exceptional in their new tasks. By May, the basement office contained about 120 persons. Of these, perhaps half a dozen were by then fairly competent cryptanalysts, 50 were beginning to get the feel of the work, and the remainder were clerks. Work went on round the clock in the air-conditioned basement, but the unit was woefully understaffed.

  Rochefort virtually lived in that cellar for the first three months. He supervised the entire operation—interception, traffic analysis, cryptanalysis, translation. Dyer, his immediate subordinate, was in charge of the cryptanalytic section. A slender man just turning 40, with a mild, friendly personality but a tough and unrelenting mind, Dyer had come to the Islands in 1936 and had begun cryptanalytical work largely on his own initiative. He had become interested in the field soon after his graduation from Annapolis in 1924. Assigned to New Mexico as an assistant radio officer, he began doing the cryptograms in the naval communications bulletins, which intrigued him, and then read Friedman’s Elements of Cryptanalysis, which hooked him. In 1931, he succeeded Safford as head of the Research Desk in the Code and Signal Section of Naval Communications, commanding the entire U.S. Navy cryptanalytical group of four people, clerks included.

  The following year, Dyer became the father of machine cryptanalysis when he installed I.B.M. machines to speed up solution. (The Army did not begin using the machines for cryptology until 1936.) In 1937, after he had been in Hawaii for a year, the Navy sent some I.B.M. machines out to him and assigned him a yeoman to expand, in a modest way, the cryptanalysis that he had been doing. Those machines were his baby. While other cryptanalysts used pencil and paper to test assumptions, Dyer tried them out directly on the machines—and worked more quickly that way than he could have by hand. He stayed in cryptanalysis all during the war, winning the Distinguished Service Medal, and even afterward, rising to a captaincy. On his retirement from the service in 1955, he started teaching mathematics at the University of Maryland.

  His chief assistant was Wright, who handed out the work that Dyer wanted done and then pitched in himself. In 1929, three years after he graduated from Annapolis, he found himself with his crew on a rifle range shared by Safford, a fellow officer in a destroyer division. Like Dyer, he had solved the ciphers in the communications bulletins, and Safford, in a sales campaign that began to the crack of musketry, convinced him that he should specialize in cryptology. But it was not until June of 1933 that Wright began his first tour in communications. Sea duty alternated with cryptologic work until, in March of 1941, he went to Pearl Harbor with Admiral Kimmel as the cryptanalyst in a fleet security unit. He immediately began working with the Combat Intelligence Unit and in February of 1942 was formally transferred to it. He was then 39, a broad-shouldered redhead with craggy features and big hands whose strong resemblance to a tugboat captain—his nickname is “Ham”—belies his gentle manner and his courtesy. He too remained in cryptology throughout the war, winning the Legion of Merit; like Dyer he stayed in it afterwards, winning a gold star to his Legion of Merit. He retired in 1957.

  With the entrance of the Rochefort group into the fray against JN25b, the three Allied cryptanalytic units in the Pacific and OP-20-G in Washington began working in the closest possible cooperation. Positive or tentative codegroup recoveries were flashed from unit to unit via the COPEK channel for MAGIC. Each unit intercepted messages that the others might not have picked up, and so could make new assumptions or confirm or disprove old ones. Washington, which had the most equipment and the largest staff, seems to have led in the work of stripping the additive groups. The Singapore and the Philippines units had made the difficult initial entries, but their work was interrupted when the British had to move to Colombo and Fabian was evacuated by submarine from Corregidor in February, 1942, several weeks before MacArthur. Aside from a few such generalized observations, it is almost impossible to say which group, much less which individual, deserves the major share of credit for solving the edition of the fleet cryptographic system then in force. Collaboration was too intimate. A possibility raised in a discussion between Dyer and Wright might be developed into a probability by a check of messages in Washington and verified by a new intercept at Colombo.

  Meanwhile, the Japanese—who had no suspicion of all this activity—felt a vague unease at the extreme length of service of this code. A new edition, which would be called JN25c by the Americans but was called the Naval Code Book D by the Japanese, was to be placed in service April 1. But administrative confusion in the Navy libraries, which had custody of the codebooks, plus difficulties in physically distributing the books by destroyer and airplane to moving ships and widely dispersed installations, forced a postponement to May 1. Consequently, the American cryptanalysts could tunnel ever more deeply into JN25b.

  Gradually the isolated fragments of plaintext that they were recovering grew denser, enlarged, touched, made sense. Parts remained unread, but the large patches of coherence offered clues to Japanese thoughts and plans. Hence it was that by April 17 the cryptanalysts smoked out the gist of the Japanese plan to seize Port Moresby and threaten Australia. The new Commander in Chief of the Pacific Fleet, Admiral Chester W. Nimitz, dispatched two carriers, Lexington and Yorktown, to spoil it.

  This task force, commanded by Rear Admiral Frank Jack Fletcher, began cruising the lovely waters of the Coral Sea off the northeast coast of Australia in search of the enemy. At 8:15 a.m. May 7 a message from a Yorktown search plane was decoded as reporting the discovery of “two carriers and four heavy cruisers” 175 miles northwes
t of the American force. Fletcher thought that this was the main Japanese force covering the amphibious landing and flew off two deckloads of planes to attack it. When the search pilot returned, it was discovered that the “two carriers and four heavy cruisers” had resulted from a disarrangement of his codepad; they should have been reported as “two heavy cruisers and two destroyers.” But another contact report alerted the fliers to the presence nearby of the landing force itself, escorted by the light carrier Shoho. They swarmed over Shoho and sank it in ten minutes—a record for the war. “Scratch one flattop!” exulted one pilot. The transports, shorn of their air cover, retired to the northward. This accidental attack on the wrong force thwarted the main Japanese objective and, since the transports never again entered the Coral Sea, lifted the threat of invasion from Australia.

  Fletcher could hardly foresee this, however, and next day he located the main Japanese force of two big carriers and attacked them at the same time that they spotted and attacked him. It was the first naval battle in history which was fought entirely by air and in which the opposing ships never even sighted each other. One Japanese carrier was put out of action; the other had its flight deck bent so that it could not recover all its planes, many of which had to be jettisoned. But Yorktown was scarred and the beloved Lexington so badly damaged that, after futile attempts to save her, she had to be torpedoed by an American destroyer. Though this gave the Japanese a tactical victory in the Coral Sea, they had lost strategically. More important, their two damaged carriers would not be present at the Midway battle. For the check at the Coral Sea had not altered Japan’s grandiose plans for winning the war against America.