The contention that that broken Japanese codes could have alerted us won’t go away. Were key messages overlooked? Deliberately withheld? Or is there a simpler explanation?
FIFTY YEARS LATER, AMERICANS STILL WONDER HOW Japan’s surprise attack on the pride of the Pacific Fleet could have succeeded. The joint congressional committee that in 1945 and 1946 investigated the attack put the question sharply:
Why, with some of the finest intelligence available in our history, with the almost certain knowledge that war was at hand, with plans that contemplated the precise type of attack that was executed by Japan on the morning of December 7—Why was it possible for a Pearl Harbor to occur?
The “finest intelligence” came from code breaking. Solving the secret messages of a hostile power is like putting a mirror behind the cards a player is holding, like eavesdropping on the huddles of a football team. It is nearly always the best form of intelligence. It is faster and more trustworthy than spies, who have to write up and transmit their reports and who are always suspected of setting up or falling for a deception. It sees farther into the future than aerial reconnaissance, which detects only what is present. It is broader in scope than the interrogations of prisoners, who know little more than what they have experienced. And it is usually cheaper and less obtrusive, hence more secret, than all of these. But it has a serious double-barreled failing: It cannot provide information that a nation has not put onto the airwaves, and its apparent omniscience and its immediacy seduce its recipients into thinking they are getting all the other nation’s secrets.
This is one of the lessons of Pearl Harbor. American code breakers performed prodigies, giving remarkable insight into Japanese thinking. But that insight was not total, and so even the extraordinary U.S. cryptanalysis could not warn policymakers of Japan’s secret intentions.
The nations of the world learned the value of code breaking during World War I. Radio—used extensively in that conflict for the first time— gave them their opportunity. Messages were easily intercepted, so armies and navies sheathed them in codes and ciphers. But linguists and mathematicians on both sides learned to crack them, and the information thereby obtained provided victory after victory to generals, admirals, and political leaders. Cryptanalysis substantially helped France to block a supreme German offensive in 1918, Germany to defeat Russia, Britain to bring the United States into the war, the United States to convict a German spy. When hostilities ended, the powers refounded these agencies to retain in peace the benefits won in war.
The United States was one of these nations, and its main target was Japan. Before World War I, Japan had defeated China and then Russia to become mistress of the western Pacific. Now it was building a fleet to match that of the United States and, under a League of Nations mandate, had occupied islands that enabled it to menace the ocean routes to the Philippines. It was generally felt that Japan constituted the greatest danger to the United States.
The State and War departments jointly set up the Cipher Bureau in 1919 under the inspiring leadership of Herbert O. Yardley, a 30-year-old who had created and run a code-breaking unit for military intelligence in World War I. The Cipher Bureau scored the first great achievement of American code breaking while working out of a narrow brownstone at 141 East Thirty-seventh Street in Manhattan. Despite only a rudimentary knowledge of Japanese, Yardley and his associates cracked Japanese diplomatic codes. A bewhiskered missionary then turned the messages into English. Sent to the State Department, they informed American negotiators at the Washington naval disarmament conference of 1921—22 about Japan’s fallback position on capital ships. Armed with this knowledge, the negotiators drove Japan to promise to build such ships in a U.S.—Japan tonnage ratio not of 10 to seven, as Japan had wanted, but of 10 to six—the equivalent of three fewer battleships.
Although the navy was more concerned about Japan than was any other element of government, it had no code-breaking unit. Then, early in 1923 naval intelligence came upon a 1918 Japanese naval codebook while rifling the steamer trunk of a Japanese naval officer visiting New York. This impelled the navy to create a code-breaking agency—called, for security reasons, the Research Desk—within the Division of Naval Communications. Its first head was Laurance F. Safford, a lieutenant with a flair for mechanics and mathematics. He set up shop with four civilians in Room 1621 of Main Navy, a temporary wooden building on Constitution Avenue near the Lincoln Memorial. One of the first things he did was to set up radio intercept stations in the Pacific, to furnish more material for code breaking than was obtainable through haphazard monitoring by ships and the naval radio station in Shanghai.
In August, Safford took one of his most important strides forward when he hired Agnes Meyer Driscoll, 32, as a cryptanalyst. A onetime mathematics teacher and a former employee of the Code and Signal Section, under which the Research Desk came, she soon proved to be an outstanding code breaker. Among her first assignments was to work on the photographed code from the rifled steamer trunk. The Research Desk had found that not only was the “plaintext,” or the original message, encoded; its code groups were themselves enciphered. “Miss Aggie,” as she was called, had to remove that encipherment. Incessantly turning the pages of the reproduced codebook with the rubber tip of her eraser, she completed that job after two to three years of work. A husband-and-wife team of translators turned the Japanese into English. By then, in 1926, Safford had returned to sea. He was succeeded by Lieutenant Joseph J. Rochefort, one of the first American naval officers to have studied Japanese in Japan. He was a “mustang”—a former enlisted man who had won a commission. This had made him tough and independent in a world dominated by Annapolis graduates; he neutralized his caustic speech with a conciliatory smile. Rochefort became one of the very few Americans with aptitude both in the Japanese language and in code breaking.
A subordinate, probably solving one of the Imperial Japanese Navy’s subsidiary codes, described the work:
Hours went by without any of us saying a word, just sitting in front of piles of indexed sheets on which a mumbo jumbo of figures or letters was displayed in chaotic disorder…Wel gave ourselves to cryptography with the same ascetic devotion with which young men enter a monastery.
The hardest part of breaking a code is the beginning. Rochefort explained it in colorful terms:
It first off involved what I call the staring process. You look at all of these messages that you have, you line them up in various ways, you write them one below the other, and you’d write them in various forms and you’d stare at them. Pretty soon you’d notice a pattern; you’d notice a definite pattern between these messages. This was the first clue. You notice a pattern that when you follow through, you say this means so-and-so; you’d run that through, and it doesn’t work out. Then you’d proceed on some other effort and eventually, if you’re lucky and the other fellow makes mistakes, which he invariably will, then you come up with a solution that will stand up under test, and this gives your first lead-in.
ROCHEFORT SAID HE FELT GOOD WHILE DOING THIS WORK “because you have defied these people who have attempted to use a system they thought was secure, that is, it was unreadable. It was always somewhat of a pleasure to defeat them or challenge them.” But the work took its toll. While engaged in the actual cryptanalysis, he said, he generally felt frustrated. The tension was so great that after work he had to lie down for two or three hours before he could eat anything; he developed ulcers anyway, and this, together with the fact that duty in communications intelligence hurt a man’s career, drove him to get out of the work when his tour at the Research Desk ended in 1927.
The translation of the photographed Code No. 1 originally had been put together in 10 “volumes” with metal-strip Acco office binders. When Safford returned from sea duty to the Research Desk in June 1929, he had the material retyped in four copies on huge 12-by-18-inch forms and bound in two volumes in red buckram McBee binders, far more convenient to use. This gave the code its more common name, the Red Code.
On December 1, 1930, the Japanese replaced it with a new code. But Miss Aggie by then had learned the ships, communications patterns, and frequently used phrases of the Japanese fleet, and she solved its transposition encipherment and then reconstructed the entire 85,000-group, two part code. It was later called, from the color of its binding, the Blue Code. Her work was a remarkable feat of cryptanalysis, and for years it gave the U.S. Navy insight into Japanese forces and tactics.
Two events in 1929 led the army to expand its own code-breaking activities. In May, after giving the new secretary of state a little time to understand the realities of the job, Yardley passed him some solved messages. Henry L. Stimson was shocked at what he regarded as a dishonorable and counterproductive activity—”Gentlemen do not read each other’s mail,” he said later, maintaining that “the way to make men trustworthy is to trust them.” He withdrew State Department support from the Cipher Bureau. In the meantime the army decided that Yardley was not doing what it needed most: training cryptanalysts for immediate use in case of war. These events doomed the unit, which was dissolved on October 31, 1929—two days after the great stock-market crash. Its papers went to the army’s Signal Corps.
This body had set up a small cryptologic group of its own in 1921, hiring a 29-year-old who was on his way to becoming the world’s greatest cryptologist. William F. Friedman—natty, uptight, brilliant—had written some theoretical treatises of landmark importance and had solved German codes in France during World War I. His new job was nominally to improve the army’s own codes and ciphers, but doing this properly required him to test cryptographic systems offered to it. This gave him experience in cryptanalysis and expanded the army’s knowledge of it. With the closing of Yardley’s agency, it was logical for the Signal Corps to add code breaking to its responsibilities, and Friedman became the head of a new Signal Intelligence Service (SIS). He hired three young men who knew languages and mathematics to be junior cryptanalysts. The first to report was Frank B. Rowlett, 21, a former teacher from Virginia with an all-American look to him. At 8:00 A.M. on April 1, 1930, Rowlett found himself entering Room 3406 of the Munitions Building—next door to the main Navy Department building, which housed the navy code breakers—on Constitution Avenue near the Lincoln Memorial.
Two months later Rowlett and his colleagues were excitedly combing through the secret files of Yardley’s defunct organization. This most clandestine and most valuable form of intelligence thrilled them. They went on to study basic cryptography and the solution of machine ciphers, clearly the wave of the future. In 1932, their training completed at last, they attacked Japanese diplomatic cryptographic systems, working on messages provided by the army’s new intercept service.
They first cracked a simple code, the LA. That code did little more than replace the syllables of the plaintext with pairs of code letters listed in a codebook. In fact, the system resembled simple cryptograms found in Sunday newspapers. First the Japanese words of the message were transliterated into romanized letters (so that Western telegraphic systems could be used in sending them). This was done by using the katakana (literally, “borrowed words”), a syllabary that expresses Japanese words phonetically. The cryptographer then took the phonetically spelled words and encoded them, syllable by syllable, looking each of them up in the code list.
When the young cryptanalysts discovered that LA encrypted only insignificant messages, such as expense or vacation reports, and when they had gained experience and knowledge of Japanese diplomatic language and communications practices, they turned their attention to the more important messages. These were protected by electromechanical machines that enciphered messages at one end and deciphered them at the other. The machines rendered more complicated codes than a codebook did because they constantly changed the enciphered letters as the cipher clerk typed the message. Only a counterpart machine, properly set and advancing at the same pace as the sender’s, could decipher the message. This system served two main Japanese diplomatic communication networks—one covering the Far East, the other linking Tokyo with major world capitals.
As difficult as machine systems were, however, study of the cryptograms did yield clues. Vowels, for instance, had a remarkably higher frequency than consonants. It appeared that the machine divided the romanized alphabet (used in the katakana transliteration) into two subsets, the six vowels and the 20 consonants. Working with one of the less garbled intercepts, and perhaps with some help from the navy’s solution of another Japanese cipher machine, Rowlett and Solomon Kullback, one of the other original junior cryptanalysts, struck gold one day: Among their tentative recoveries of plaintext were three letters followed by an unknown and then another letter: oyo?i. They knew then that they had cracked the system, because oyobi is romanized Japanese for and. They named this machine system RED (not related to the Red Code).
By 1937, for the first time in American history, solutions of foreign messages began going to the White House, probably to President Franklin D. Roosevelt. The gentlemen were once again reading someone else’s mail. It revealed, for instance, advance information about Italy’s possible adherence to the German-Japanese Anti-Comintern Pact. This was in March 1937, six months before American diplomats began reporting on it. Later it provided part of the text of the treaty.
The next year, messages began to appear suggesting that a new machine would supplement and probably eventually replace the older one, which was physically wearing out. On February 20, 1939, three messages in the new system were intercepted, and over the next three months messages in RED gradually disappeared. Japan’s major diplomatic messages had become unreadable. Faced with the loss of the nation’s paramount intelligence, the SIS mounted a concentrated attack to solve the new machine. Friedman put Rowlett in charge and exercised overall supervision himself. The Americans called the new machine PURPLE, perhaps in part because its deeper hue fit its deeper mystery.
In the absence of American spies almost anywhere in the world, these half-dozen cryptanalysts were providing the United States with its best secret intelligence on Japan as relations with that nation, which was persisting in its aggression against China, deteriorated. The cryptanalysts plunged into their work in Rooms 3416 and 3418 in the Munitions Building. Room 3418, about 25 feet square with a steel door secured by a combination lock and with barred windows, was known as the vault. As additional cryptanalysts were assigned to the PURPLE problem, the group moved into larger quarters, finally occupying about eight rooms.
Rowlett worked in Room 3416. His desk was usually neat, for he spread out his worksheets on a nearby table. He was extremely focused on the work, arriving at 7:00 A.M., an hour early, and leaving at 5:00 P.M., an hour late. He never hummed or chewed his pencil or muttered to himself; he looked out the window only when something distracted him; he never drank coffee at work, though he did puff on a pipe. His mind didn’t dwell on the cryptanalytic problems during the 15-minute drive to work from Arlington County, but he would awaken each night after a few hours of sleep and, lying there in the dark, review the day’s work and think of ways of improving it. In the morning, at work, he would exchange ideas with the other cryptanalysts—Robert O. Ferner, Albert W. Small, Genevieve Grotjan, and Mary Jo Dunning, assisted by Leo Rosen, Sam Snyder, Kenneth D. Miller, Glen S. Landig, and Cyrus C. Sturis, Jr. (their names deserve to be remembered). After the conference, they all would return to their desks. Quiet reigned as they pored over the intercepts, most of which had been teletyped in from the monitoring stations; sometimes they puzzled over statistical and alphabetical tables compiled by hand from the intercepts. Only the rustling of papers and the scratching of pencils disturbed the brooding silence, although for a time the banging and hammering of workmen on another floor proved frustrating.
Reconstructing a cipher system is like solving an immensely complicated scientific problem, with this difference: Nature does not deliberately conceal her secrets. The researchers concoct hypotheses and test them. If x stands for e, will the other cipher-to-plain equivalents that it entails make sense? Or will they merely yield gibberish, or lead to a self-contradiction? Can one recovered alphabet be linked with another? There’s no clear way to the answer, as there is in the algebra problems posed in math classes. Particularly in the early stages of a difficult cryptanalysis, the work is one of the most excruciating, agonizing, tantalizing, compelling mental processes known to humans—and, when successful, one of the most satisfying.
PURPLE had carried over from RED the division of letters into groups of six and 20 letters. But this time the six were not exclusively vowels. Nevertheless, within a few weeks the cryptanalysts ascertained how they were enciphered. This enabled the team to recover the plaintext for those letters. The process was slow and painstaking. Assigned to devise a way to mechanize this pencil-and-paper method, Leo Rosen hit upon the idea of using telephone selector switches, employed in dialing. They worked like a dream, and the solution process was considerably speeded up.
Despite Rosen’s remarkable advance, the totality of PURPLE still resisted the Americans. Friedman, who had been supervising the work rather loosely, was asked by his bosses—all extremely supportive, financially as well as psychologically—to participate personally. His genius helped considerably. The navy also lent a hand temporarily, organizing its files the same way as the army’s to facilitate cooperation. After about four months, however, the navy returned to its main effort, Japanese naval codes. The Signal Intelligence Service pushed ahead. Within Rowlett’s group, teamwork was extremely close; determination was pervasive. No one complained that a task was too menial. Rowlett was confident from the start that they would reconstruct the PURPLE mechanism the way he and others had reconstructed the RED. He never got depressed, even as months went by without a solution.
As they sought a breakthrough, the cryptanalysts spent much of their time trying to match possible plaintext—guesses, often educated, as to the text of the original message—to the cipher text, or encoded language and numbers. Early in the effort, for example, many identical Japanese telegrams were sent to multiple addresses; some of the telegrams were composed using the RED machine, some the PURPLE. The cryptanalysts could read RED, which then gave them the text of the same PURPLE messages. They knew, too, that many diplomatic dispatches began “I have the honor to inform Your Excellency that . . . and they often tried that as the start of the plaintext. In a very few cases, the State Department gave them the text of notes to or from the Japanese ambassadors, which the code breakers used as cribs.
The code breakers had to make all sorts of guesses. They theorized that the PURPLE machine would have to advance in some regular fashion, that its mechanism would have to click forward at some prescribed rate. Suppose, for example, that the probable plaintext word Japan was guessed. If the probable a’s were represented in the ciphertext by, say, x and z, then the cryptographers could hypothesize that the encoding machine had simply moved forward one space with each new letter: x for a, something for p, and z for the next a.
More than a year of painstaking trial-and-error work passed. Then, about 2:00 p.m. on a warm Friday, September 20, 1940—in the middle of Roosevelt’s campaign for an unprecedented third term, as Britain anxiously awaited a German invasion from occupied France—Albert Small noticed that Genevieve Grotjan, a dignified, 26-year-old statistician, seemed to be concentrating extremely intently. When he asked, she told him that she had just discovered a couple of the needed intervals and was looking hard for others. He took her in to see Rowlett, who was conferring with Bob Ferner. Grotjan showed the men her discoveries; then a third interval leaped out at the code breakers. They saw at once that these intervals proved that their concept of PURPLE was correct. The ebullient Small dashed around the room, hands clasped above his head. Ferner, normally phlegmatic, shouted, “Hooray!” Rowlett jumped up and down. “That’s it! That’s it!” Everybody crowded around. Friedman came in. “What’s all the noise about?” he asked. Rowlett showed him Grotjan’s findings. He understood immediately. Grotjan’s discovery constituted the decisive breakthrough in the solution of PURPLE—it was the greatest moment in the history of American code breaking. And what did the egghead cryptanalysts do? They sent out for Coca-Colas!
When the euphoria and the effects of the colas had worn off, the cryptanalysts drifted back to work. Grotjan, who seems to have gotten excited about the breakthrough mainly because everybody else did, soberly regarded it as just one step in a series of steps. A week later—the day after Japan started to occupy French Indochina and the very day the Tripartite Pact establishing the Rome-Berlin-Tokyo axis was signed—the Signal Intelligence Service handed in its first two solutions of PURPLE messages. This did not mean its work was done. The settings for the machine changed each day, and the cryptanalysts had to recover these. But this work was facilitated by Rosen’s construction of two American analogues of the Japanese PURPLE machine, at a cost of $684.65. Later, additional copies of the machine were built, several at the Washington Navy Yard; some of them were given to the navy, which had rejoined the PURPLE work to help with the heavy volume of solutions, and some to the British, so they could read the messages without having to wait for American solutions to be forwarded to them.
Soon navy lieutenant Francis A. Raven discovered a pattern to the daily setting changes. With this knowledge the Americans were able to read reports from and instructions to Japan’s ambassadors on average within a day or so, sometimes within hours. They had gained access to the most secret diplomatic dispatches of the empire of Japan as relations continued to worsen, with an American embargo on the export of iron and steel scrap and, later, with the movement of Japanese forces toward Thailand.
The beginning of wisdom was to know that there was no such thing as “the” Japanese code. For PURPLE was not the only cryptographic system of Japan’s Foreign Ministry, much less of the empire. The Foreign Ministry employed a hierarchy of systems, of which PURPLE was the apex. Under it came several codes that—unlike PURPLE, which served embassies exclusively—were used at both embassies and consulates. LA, the simplest, lay at the bottom. Above it rested a two-part system, PA-K2. More complex still was the Foreign Ministry’s J series of codes. The K transposition key for these codes changed daily; the code breakers had to undertake a fresh analysis with each day’s messages. Some 10 to 15 percent were not solved at all, and those that were took an average of a week from interception through translation to distribution.
By contrast, most PURPLE messages were solved within hours, and all but 2 to 3 percent of the keys were recovered. Did the Japanese err in assessing the security of their cryptographic systems? Yes and no. PURPLE was a much more difficult system to solve in the first place, but once solved it was easier to keep up with.
While the army was concentrating on Japan’s diplomatic systems, the navy’s code-breaking agency—except for its occasional help to the army— focused on Japan’s naval systems. This agency, again under Safford, now a commander, was called OP-20-G. During the 1930s it continued to read messages in what it called the Blue Code, gaining considerable knowledge about Japan’s naval maneuvers. This code was replaced on November 1, 1938. But the paucity of intercepts in the new code, which the Americans called the flag officers’ code, meant that almost no progress was made in reading it.
On June 1, 1939, the Japanese introduced yet another code. Called JN25 by the Americans, as it was the 25th Japanese naval code they attacked, it encoded messages dealing with naval operations. Miss Aggie, greatly helped by Lieutenant Prescott Currier, attacked the new code. About a year and a half later, in almost the very week that the army was producing its first PURPLE solutions, the first JN25 solutions emerged. But the navy’s satisfaction did not last long. On December 1, 1940, the Imperial Navy substituted a new version, which the Americans called JN25b. But the Japanese foolishly kept the prior version’s latest encipherment in force for the first two months of JN25b’s service. This bared the underlying code and permitted the quick determination of the meaning of 2,000 code groups. When the new encipherment went into effect, OP-20-G’s IBM tabulators were able to strip it off, enabling the Americans to read bits and pieces of the coded messages. But progress continued to be slow.
So in March 1941, OP-20-G requested that the radio intelligence unit on Corregidor Island in the Philippines help with its search for a solution. A British code-breaking unit in Singapore exchanged recoveries of JN25b code groupings with Corregidor. Despite all these efforts, the code remained readable only to a very small degree. By December 1941, only 10 to 15 percent of each message could be understood.
This, then, was the cryptanalytic situation with Japan on December 6, 1941: The main diplomatic system could be read rapidly and completely; other diplomatic and consular systems could be read with a few days’ delay; the main naval system could be read only slightly.
Nearly all intercepts came from army or navy stations listening to commercial frequencies such as RCA’s that radioed messages from Japan. The intercepts were sent to the SIS or OP-20-G by teleprinter, airmail, courier, or radio, re-enciphered in an American system. Translation was a bottleneck because of the difficulty of finding enough qualified people who understood Japanese. On the other hand, not all of the PURPLE messages were in Japanese: Some of the notes, intended to be handed to the State Department, were in English.
Fifty to 75 intercepts were solved and translated each day. The most important of these were selected for distribution to a handful of high level officials: the president; the secretaries of state, war, and the navy; the chief of staff and the chief of naval operations, the war plans chiefs of the two services; and some intelligence officers. Fourteen copies of each were typed up, some for the files, and intelligence officers carried them in locked briefcases to these officials, calling attention to some of the more critical dispatches and explaining obscure references. Then they took the papers back with them and burned them. As a cover, this intelligence was called MAGIC.
What did these messages say? Many revealed the empire’s reaction to world events and American policies. They included the reports of and the instructions to the Japanese emissaries. “If the United States expresses too many points of disagreement to Proposal A,” Tokyo cabled its ambassadors in Washington on November 5, in a PURPLE message that the navy solved the same day, “and if it becomes apparent that an agreement cannot be reached, we intend to submit our absolutely final proposal, Proposal B (contained in my message #727).” That other message had been intercepted and solved the day before. Messages to and from the consulates frequently dealt with the movement of U.S. Navy warships into and out of harbor. On November 15 Tokyo told Honolulu in the J19 code, “As relations between Japan and the United States are most critical, make your ‘ships in harbor’ report irregular, but at a rate of twice a week.” The navy cracked that message on December 3.
By the fall of 1941, high levels in the U.S. government had become almost addicted to MAGIC. Secretary of State Cordell Hull, who looked upon MAGIC “as I would a witness who is giving evidence against his own side of the case,” was “at all times intensely interested in the contents of the intercepts. ” The chief of army intelligence regarded MAGIC as the most reliable and authentic information that the War Department was receiving on Japanese intentions and activities. General George C. Marshall, the army chief of staff, called it a “priceless asset. ” And when the president was not given MAGIC for a few days in November, through a bureaucratic mixup, he specifically asked for it. This was the situation as the Pearl Harbor strike force slipped out of Japan’s naval bases to assemble in the foggy Kuril Islands north of the main islands of Japan, far from any prying eyes, thence to sail in utter silence across the empty wastes of the North Pacific toward its unsuspecting target: a palm-fringed inlet in Hawaii.
SOME PEOPLE HAVE CONJECTURED THAT THIS FABULOUS decoded information made it clear to Roosevelt and his advisers that Pearl Harbor was going to be attacked. They say the president, wanting to bring the United States into the war on the side of Great Britain, traitorously suppressed this information and sacrificed American ships and American lives to achieve his goal. Various theories have been put forth to support this notion.
Safford himself, by then a captain, agreed. He based his argument upon the so-called winds code. Japan had notified its diplomatic posts in a J19-K10 circular telegram on November 19 that if diplomatic relations and international communications were likely to be cut off, it would warn these posts with a fake weather forecast in the middle of the Japanese shortwave news broadcast. If Japanese-American relations were in danger, the forecast would predict “east wind rain.” American code breakers solved this message on November 28. Immediately a frantic effort was made to pick up this broadcast. Safford insisted that the “winds execute”—the forecast— was heard on December 4 and that the intercept was subsequently removed from the files as part of a cover-up. Virtually no one has supported this contention. But even assuming that an execute had been transmitted, it would at best confirm that relations were strained. In no way could it point to Pearl Harbor.
Rear Admiral Edwin T. Layton has argued that the lack of a PURPLE machine in Hawaii prevented Admiral Husband Kimmel and General Walter Short, the commanders there, from using MAGIC-provided information on international affairs to illuminate their situation. This would have enabled them to predict the attack, Layton has claimed. But this is speculation, supported only by hindsight. Moreover, the presence of a PURPLE machine in the Philippines did not prevent the American forces there from being surprised.
Author John Toland found several former radio operators who had been listening in San Francisco or at sea. They said that in the week before December 7, they had heard a cacophony of radio signals from northwest of Hawaii—presumably the strike force heading for Pearl Harbor. They said they reported this, to no effect. But this story founders because, according to the Japanese, the strike force maintained absolute radio silence throughout its voyage. And American naval intercept units, straining to pick up whatever they could on the Japanese naval circuits, heard nothing. (U.S. radio-intelligence operators knew that several carriers had dropped out of the traffic picture. They thought that the ships were in home waters, covering a movement to the south—the Philippines or the oil- and rubber-rich Dutch East Indies. Carrier communications had likewise vanished in February and July 1941, and naval radio-intelligence operators hypothesized then that the carriers had been held near Japan—a hypothesis later determined to be factual. But what happened then was not what was happening in December.)
Several writers have suggested that the solution of the many messages dealing with ship movements in and out of Pearl Harbor should have alerted the authorities to the impending attack. But similar messages were transmitted about the Philippines, the Panama Canal, San Francisco, San Diego, and Seattle. In fact, from August 1 to December 6, 59 such intercepts dealt with the Philippines and only 20 with Hawaii. The writers have pointed to one intercept, instructing the consulate in Hawaii to divide the Pearl Harbor anchorage into smaller areas for more precise reporting of ship locations, as a clear indication of a forthcoming attack. This is hindsight. At the time, the authorities viewed it merely as evidence of the thoroughness of Japanese intelligence or of the need to abbreviate communications.
James Rusbridger, in his recent book Betrayal at Pearl Harbor, claims that the British code-breaking unit at Singapore solved enough of JN25 to reveal the plan to attack Pearl Harbor and that this information was passed on to Prime Minister Winston Churchill, who withheld it from Roosevelt to ensure American entry into the war, thereby enabling the attack to succeed. Citing the Official Secrets Act, British authorities have denied Rusbridger any access to the records of the unit, and the U.S. Navy reports that it cannot find any of these JN25b solutions, partial or complete, from before December 7. His thesis rests on the memory of an Australian code breaker working in Singapore. This is a slender reed upon which to base so heavy a charge. Moreover, Rusbridger does not distinguish between the a and b editions of JN25—and does not make clear why Churchill would try to get the Americans to fight Japan instead of Germany.
A retired communications intelligence analyst, Fred Parker, has dredged through the files of Japanese messages intercepted before Pearl Harbor but not solved until afterward. Though he has found no smoking gun, no message referring specifically to an attack on Pearl Harbor, he believes that the messages he has found point clearly to an impending attack there. He cites, for example, the presence of an oiler on what became the strike force’s homeward route and the broadcast of the message “Climb Mount Niitaka 1208.” The oiler obviously was put there, he says, to refuel the returning ships. The 1208 in the message meant December 8, the attack date on the Tokyo side of the International Date Line. Mount Niitaka (Hsin-kao, in Taiwan) was the highest peak in what was then the Japanese Empire. Parker contends that a solution of these messages would have suggested an attack on Pearl Harbor. Like the other theories, however, this is hindsight.
Some historians have contended that if only the army and navy intelligence officers, and perhaps State Department officials as well, had found the time to analyze all the intercepts as a group, they would have discerned a pattern that pointed to Pearl Harbor. This argument resembles one that Roberta Wohlstetter made in her book Pearl Harbor: Warning and Decision, in which she holds that the noise of the false evidence drowned out the indications of the true signals: “We failed to anticipate Pearl Harbor not for want of the relevant materials, but because of a plethora of irrelevant ones.” This is wrong. There were no true signals, no clear indications of the attack.
The fact is that code-breaking intelligence did not prevent and could not have prevented Pearl Harbor, because Japan never sent any message to anybody saying anything like “We shall attack Pearl Harbor.” The ambassadors in Washington were never told of the plan. Nor were any other Japanese diplomats or consular officials. The ships of the strike force were never radioed any message mentioning Pearl Harbor. It was therefore impossible for the cryptanalysts to have discovered the plan.
What then is the answer to the joint congressional committee’s question? What about that “finest intelligence”? The simple answer is that fine though it was, it was not fine enough. Perhaps if the United States had established intercept operators in the U.S. Embassy in Tokyo to obtain enough messages to make a solution of JN25b more likely, or had been able to buy a spy in the top circles of the Japanese government, or had been able somehow to fly aerial reconnaissance regularly above the island empire— then perhaps there might have been a chance that the Pearl Harbor attack would be detected in advance. None of these things could have been easily done. Even if they had, discovery of the plan would not have been certain. Japan had successfully closed all openings through which foreigners might gain information about its intentions. The real reason for the success of the Pearl Harbor attack lies in the island empire’s hermetic security. Despite the American code breakers, Japan kept her secret. For Americans, the Rising Sun rose in eclipse.
DAVID KAHN is the author of The Codebreakers: The Story of Secret Writing (Macmillan, 1967) and of Seizing the Enigma, (Houghton Mifflin, 1991).
This article originally appeared in the Autumn 1991 issue (Vol. 4, No. 1) of MHQ—The Quarterly Journal of Military History with the headline: “Why Weren’t We Warned?”
Want to have the lavishly illustrated, premium-quality print edition of MHQ delivered directly to you four times a year? Subscribe now at special savings!