At Dawn We Slept: The Untold Story of Pearl Harbor

By 1941 Shivers had a staff of about twenty-five, including clerical employees. His Honolulu field office was responsible for “all cases of subversive activity (including espionage) involving the general civilian population.” In cases of Japanese subjects, the FBI shared concurrent authority and responsibility with the Navy District Intelligence Office (DIO). The DIO consisted of a main office in Honolulu, three zone offices on outlying islands, and ten Intelligence units located within naval stations on Oahu, Maui, and Midway. This organization could investigate all counterespionage affairs in which the subjects were Navy personnel and employees and naval contractors’ employees, and it shared counterespionage responsibility with the FBI in cases of Japanese subjects.³

Mayfield doubted that Japanese espionage centered on the Japanese consulate on Oahu:

. . . I felt that the consulate would perhaps be advised of the existence and would cooperate with the net, but that the consulate . . . itself was not the head of the net nor [sic] necessarily an important part of the net, as the consulate might expect to be closed in similar fashion to the German and Italian consulates, and that therefore they must have prepared a plan which could be carried on without any assistance from the consulate.⁷

The Japanese had indeed prepared such a plan, and they did use a few outside agents. But regardless of Mayfield’s opinion, the consulate was the center of Japanese espionage in Hawaii.

Mayfield longed to lay his hands on the cablegrams which the consulate sent to Tokyo. He tried his luck with the commercial companies, but they refused to violate Section 605 of the Federal Communications Act of 1934, which explicitly prohibited wiretaps or interception of messages from and to foreign countries.⁸ One of Mayfield’s enlisted men, Theodore Emmanuel, did manage to tap a number of the consulate’s telephone lines, recording as many as fifty or sixty calls each day over an extended period.⁹ Of course, neither Kita nor anyone else in the consulate would discuss important classified matters over the telephone. The best Mayfield could hope for was some general idea about consular personnel and the names of their frequent contacts in the islands.

The chief investigative officer for the Hawaiian Department was Lieutenant Colonel George W. Bicknell, an imposing man who stood a hefty six feet four inches. He had an attractive freckled face with twinkling eyes. He usually worked in civilian clothes. A Reserve officer, able, alert, and intelligence-minded, Bicknell had come to Hawaii in October 1940 as assistant to the G-2. Herron had a very high opinion of Bicknell and in his initial briefing of Short informed the new commander that he had planned to make Bicknell his G-2. But Short did not take the hint.¹⁰ Bicknell therefore remained as assistant G-2 and was known as the contact officer. Generally speaking, he was Mayfield’s counterpart.

Bicknell’s main duties were to keep the department commander thoroughly informed of the civil population’s activities on the Islands. He also met and kept in touch with all visiting officials and businessmen returning from the Orient, in order “to obtain any information which they might have on the general situation in the Pacific area.” Then, too, he was responsible for the internal security of the Islands and for observations of “all counter-intelligence measures necessary. . . .”¹¹ And he had investigative responsibility in counterespionage if the subjects were in or employed by the Army or had access to an Army reservation.¹²

Bicknell worked out of the Federal Building, which also housed Shivers and his small band of FBI people. Every Tuesday Shivers, Bicknell, and Mayfield met to exchange information, and a cordial working relationship existed among their offices.¹³ Both the FBI and DIO kept a partial watch on activities at the Japanese consulate and the Nippon Yusen Kaisha (NYK) steamship line.¹⁴ But either of them could have done little to cut off Yoshikawa, Seki, or Kotoshirodo. A host nation usually bends over backward to let accredited diplomatic and consular personnel go their ways so long as they operate within the letter of the law. And Kita and Okuda were very careful to have their people carry on “legal espionage.”

To the best of our knowledge, no one from the consulate ever entered a restricted military area unless at the invitation of the American authorities. There is no record that the Japanese ever stole or photographed classified information. They violated no law in pausing to look at the imposing spectacle of men-of-war moored in Pearl Harbor. In fact, the base was an open book by its very nature. And the individual services could do nothing about this. Pearl Harbor was too big, and in too open a position, to be hidden or camouflaged from either sea or land view. The only way to cut off observation would have been to make the entire island of Oahu a restricted military reservation—something no American government could tolerate. The very law the Americans swore to uphold and protect tripped them up. This law guaranteed the privacy of the airways, and the local companies very properly refused to give the FBI, Army Intelligence, or Navy Intelligence copies of the consulate’s messages—until early December 1941, when it was too little and too late.

But irony of ironies, Washington was scooping up these and other Japanese diplomatic messages by the bucketful. The basic story is as follows: The Japanese used several diplomatic codes, the most secret of which was an exceedingly complicated cipher system known as Purple. Tokyo had a childlike faith in the complete infallibility of its diplomatic codes. It never credited the Americans with the ability to crack the Purple system.

In fact, the Signal Intelligence Service (SIS) under the tireless direction of Lieutenant Colonel William F. Friedman, succeeded in doing so as early as August 1940, after eighteen to twenty months of the most intense labor.¹⁵ Rated “the world’s greatest cryptologist,” Friedman, though quiet and unassuming, possessed a drive and tenacity that refused to recognize the word “impossible.” The decrypting of Purple and its brother systems earned the name Magic. Friedman paid a high price for his magnificent gift to his country. In December 1940 he suffered a nervous collapse from overwork, and as 1941 opened, he was under treatment in Walter Reed General Hospital in Washington.¹⁶ The “individual genius . . . of Harry Larry Clark, one of the younger civilian cryptanalysts,” triggered the breakthrough. The Navy assisted throughout 1939 and 1940 by furnishing the intercepts and taking over all other Japanese diplomatic systems so that the Army could concentrate on Purple. “The Army provided the solution and wiring diagram; the Navy provided the funds and manufacturing facilities.”¹⁷

From the summer of 1940 on, therefore, U.S. Intelligence had been reading Japan’s diplomatic messages. This meant that the U.S. government had full knowledge of virtually all the traffic which passed between the Foreign Office in Tokyo and its most important embassies and consulates abroad. So Washington knew Tokyo’s instructions to Nomura and his reports from the embassy. U.S. cryptanalysts were also reading lower-grade Japanese diplomatic ciphers, notably the so-called J codes, the current one being J-19. These were mainly in use between the Foreign Ministry and many consulates, including Honolulu. Thus, the United States also picked up the traffic between Tokyo and Honolulu about the U.S. Pacific Fleet.

By the fall of 1941 American policy makers actually knew more than Nomura about his country’s intentions, for Tokyo was by no means candid with its ambassador. The United States Army, Navy, and State departments acknowledged the enormous worth of the Magic data and leaned heavily upon them for command decisions.

But Magic was not a cure-all or an enchanted key to the mazes of all Japanese thinking. Its messages revealed only what the Foreign Office gave its own diplomats. And the Foreign Ministry itself was not omniscient. The Army and Navy dictated Japanese foreign policy, and they did not always clue in the foreign minister and his associates until matters had proceeded well along—sometimes too far. So Magic could not answer all the questions the United States wanted to ask.

For instance, in 1941 U.S. Intelligence had not yet broken through the chain of Japanese naval codes. Generally speaking, military codes are more difficult to break than diplomatic, and in addition, the Japanese Navy prudently changed some of its codes several times during 1941. Hence Washington did not know of the orders Yamamoto sent to the ships of the Combined Fleet or the messages which the Naval General Staff radioed to the Pearl Harbor task force as it sailed across the northern Pacific to Hawaii.

The main mission of the Communications Intelligence Unit of the Fourteenth Naval District at Pearl Harbor was to break down the “Japanese flag officers system.” From approximately 1926 to late 1940 this code and cipher had provided most of the U.S. Navy’s information about its Japanese counterpart. Unfortunately for the United States, on about December 1, 1940, the Japanese changed their flag officers’ code. Despite the best efforts of both the Washington and Pearl Harbor units, they had not succeeded in cracking the new version,¹⁸ although Commander Joseph J. Rochefort, who took over the Pearl Harbor Intelligence Unit on about May 15, 1941, was one of the highest qualified men in the business.

In April 1941 a machine went to Cavite; it was transferred in August to Corregidor, where the Communications Intelligence Unit had been assigned the Purple, Red, and J codes. Stark approved sending this machine because the Philippines were “the best place to intercept Japanese traffic and receive information during that time. . . .” Any benefit to Admiral Hart “was a secondary consideration.”²² A copy of all of this unit’s diplomatic translations went daily to the Army locally. In addition, all Purple and some Red and J-19 “were immediately enciphered and sent to Washington.” These cryptologists also maintained liaison with their British opposites at Singapore and furnished Washington with anything of interest from that source.²³

London received two Purple machines in January 1941. By July of that year Pearl Harbor could have had one, “but only at the expense of Washington.” Then the question arose of a third for the British. The “best compromise” was to send the machine to London “and at the same time order parts of more machines.” So around “September or early October” London had its third apparatus, and a requisition for stepping switches for four more machines was “bogged down in the War Production Board. . . .” Thus, Hawaii did not receive a Purple machine.²⁴ Whether Kimmel and Short would have derived much benefit from one is doubtful because the information of most concern to them came over the J system between Tokyo and Honolulu.

In Washington the Army’s Signal Intelligence Service, headed by Lieutenant Colonel Rex W. Minckler and under the general supervision of Colonel Otis K. Sadtler, worked in the closest cooperation with the Navy’s Communications Security. The latter’s chief, Commander Laurence F. Safford, answered to Rear Admiral Leigh Noyes, director of Communications. One year younger than Friedman, his close co-worker, Safford, like Cassius, had a lean and hungry look. A quiet and gentle person, yet a dynamo of controlled energy, he pursued his goals with almost fanatical obstinacy. His long association with cryptology, combined with a native genius for the work, made him the Navy’s recognized authority in the field.

Communications Security had a twofold mission: first, to furnish the United States with codes and ciphers; secondly, to supervise American communications security and intelligence on foreign nations, “particularly Japan—in fact, almost exclusively Japan.”²⁵

Magic decoding and translation lagged for a variety of reasons. Enough radio circuits and facilities did not exist to transmit all intercepts from station to the center at Washington by radio, so airmail was the usual route. If anything interfered with the normal airmail schedules, the data might go by train or ship. At this time only one air clipper a week plied between Hawaii and the mainland, and if the weather held it up, deliveries went by ship to the West Coast.²⁶ Once the data reached Washington, they had to take their turn in seriously undermanned offices.

SIMPLIFIED CHART OF NAVY DEPARTMENT AS OF DECEMBER 7, 1941

Japanese is very difficult to translate into English. To make matters worse, the messages came in as phonetic syllables. One such sound could have a variety of unrelated meanings. Even translators highly qualified in Japanese needed “considerable experience in this particular field before they could be trusted to come through with a correct interpretation. . . .” Moreover, they worked with diplomatic material, where a shade of phraseology carries a vital significance. No wonder Kramer often put in brutal hours of overtime.²⁹

Much the same situation existed on the Army side. SIS received and decoded the messages. Then they were sent to Colonel Rufus S. Bratton, chief of the Far Eastern Section, a dedicated officer and West Pointer who played a vital role in Magic. He determined what to distribute for top-level consideration. He knew Japan, its language and its people, far better than his superiors, for he had been a language student there and attended the Imperial War College in 1932.

All during 1941 Bratton believed that Japan would expand its Asian war and that eventually the United States would be sucked into the whirlpool. His intelligence training fortified his natural ability, so that he developed almost a sixth sense for spotting developments which others might not recognize. Once he decided that a course was right, Bratton would stick with it undaunted.³⁰

The Pearl Harbor inquiries did not bring out the exact number of translators available to Bratton, but it is doubtful if he had any more than Kramer. Inevitably, therefore, certain items received priority, while others piled up. The diplomatic exchanges between Washington and Tokyo received first choice. Nevertheless, Magic was usually translated the same day SIS passed it on. As Miles observed, “The astonishing thing . . . was not that these messages were delayed in the process of translation from Japanese to English, but that we were able to do it at all.”³¹

Yet it is not enough for facts to be gathered and for these facts to be accurate. They must be timely, and they must be acted upon, either at the central agency or by dissemination to the most interested parties. Otherwise, their harvesting becomes a mere exercise in accumulation. Having brought off one of the most astonishing coups in the history of intelligence, the United States failed to take full advantage of it. The top brass in Washington reasoned this way: The shadow of a hint reaching Tokyo that the United States could read its diplomatic mail would trigger an immediate change in the entire chain of systems which would set American Intelligence back for months, perhaps years. Thus, as Stark testified, “anybody who was let in on that had to sign a paper never to disclose it, practically as long as he lived, or ever to talk about it.”³²

SIMPLIFIED CHART OF WAR DEPARTMENT AS OF DECEMBER 7, 1941

Originally Military and Naval Intelligence had prepared summaries of the information or paraphrases of the messages for distribution. In November 1941, when American-Japanese relations were mounting to a climax, the President insisted on seeing the original messages “because he was afraid when they tried to condense them, someone would change the meaning.”³⁶ Actually Intelligence had already largely discontinued summarizing. Kramer’s early screening had been intended to weed out the more important items from “material covering the whole world.” By mid-1941, however, the sheer volume of traffic kept him busy checking the many references to preceding messages contained in the body of the incoming intercepts. He would dig up these citations and attach them to the current document, then place the bundle in each day’s folder for distribution so that the reader had the complete picture. In the autumn the greater percentage of the traffic dealt with the Berlin-Tokyo circuit or the Japanese-American negotiations.³⁷

Poco Smith later testified: “To my mind there was no danger in transmitting messages from Washington to Pearl Harbor over our system. If not safe, then it was unsafe to send our own messages back and forth between Washington and Pearl Harbor. . . .”³⁸ Evidently a certain amount of confusion existed at the Navy Headquarters as to just how much information Kimmel had available to him. Turner testified that he believed “at that time, and it was Admiral Stark’s belief, that all of these major diplomatic messages, at least in the Pacific, were being decrypted by both Admiral Hart and by Admiral Kimmel, and I did not know that Admiral Kimmel did not hold the code for those dispatches until I was so informed at the time of the Navy court of inquiry on Pearl Harbor.”³⁹

Perhaps the real idea behind Washington’s attitude lay in a cautious letter which McCollum sent to Layton on April 22, 1941, in reply to the latter’s request for diplomatic intelligence such as he had received in February concerning Japanese designs on Vichy:

It does not seem to me to be very practical to build up an organization afloat which will merely duplicate the efforts of the Intelligence Division in the Department. I appreciate that all this leaves you in rather a spot as naturally people are interested in current developments. I believe, however, that a sharp line should be drawn and a distinction continuously emphasized between information that is of interest and information that is desirable to have on which to base action.

In other words, while you and the Fleet may be highly interested in politics, there is nothing that you can do about it. Therefore, information of political significance, except as it affects immediate action by the Fleet, is merely a matter of interest to you and not a matter of utility.⁴⁰

This exchange between McCollum and Layton did not touch on the traffic to and from the Japanese consulate in Honolulu. This used the J system until early December 1941, when they adopted the PA-K2 code. Even to the most inexperienced eye these intercepts gave every evidence of military espionage. Again, Washington had an answer: Information of this type poured out of Japanese consulates on the West Coast, the Philippines, Panama, indeed from all over the world, and there was nothing to distinguish the Honolulu messages as being different in essence from those originating in any other city.⁴¹ This explanation does not quite hold water. Granted, Communications and Intelligence both were inundated with items similar to those moving between Tokyo and Honolulu. But Honolulu was not just any other city. It was the home base of the U.S. Pacific Fleet, the pivot of American Asian strategy in that vast sea. Therefore, anything even remotely hinting at undue Japanese interest in that location deserved priority handling and prompt transmission to the organization concerned.

The Honolulu-Tokyo traffic, although dispatched in diplomatic code, did not deal exclusively with diplomatic matters. Often these messages contained strictly military information. The United States authorities knew very well that in Japan the military called the tune, so that the key to their probable moves would more likely rest in their military messages than in the high-level diplomatic channels.

To complicate the situation further, the all-important function of estimating enemy intentions had recently been the subject of a struggle between Naval Intelligence and War Plans. Early in 1941 Turner came to Captain James’s office and requested that “ONI make no estimate of prospective enemy intentions for CNO but furnish information to War Plans who would make the required estimates.” James resisted this foray into his territory, informing Turner that “existing printed organization instructions of CNO required Intelligence to make these estimates.” For the time being that settled the matter.

The net effect was to reduce Naval Intelligence to a collecting and distributing clearinghouse. Even more serious, Stark’s decision placed the responsibility for evaluating Japanese intentions in the hands of officers who did not know Japan, its language, or its armed forces, as did those in ONI.

The intelligence hassle placed undue emphasis on the concept of estimating enemy intentions. Those engaged in this power struggle would have been better advised to devote that energy and time to estimating enemy capabilities. And whoever decided to withhold the Honolulu intercepts—or at least information based thereon—from Kimmel and Short must accept part of the blame for the Pearl Harbor tragedy.

CHAPTER 9

“IN RATHER A SPOT”