On a late summer morning in 1940, Frank B. Rowlett, a 32-year-old civilian employee of the U.S. Army, climbed into his Ford sedan in Arlington, Virginia, and drove to his job in Washington, D.C. Though his work was all but obsessive and tempted him to revert to it during his commute, Rowlett was disciplined and kept his mind on the traffic. He parked in a lot behind the Munitions Building, the army's offices on Constitution Avenue, arriving at 7 a.m., an hour early, as was his custom. He walked down one of the wings that stretched out the back of the building like teeth on a comb. A steel gate and an armed guard blocked the entrance to Rooms 3416 and 3418. They were among the most secure in the entire structure, and the work that went on in them among the most secret in the U.S. government.
Rowlett was a codebreaker; he had charge of the team trying to crack the most secret diplomatic cipher of the Empire of Japan, a machine that American cryptanalysts called PURPLE, and within hours on that day, Friday, September 20, he would be celebrating one of the greatest moments in American cryptology.
Tension with Japan had begun when the United States seized the Philippines in 1898. Within the Imperial Japanese Navy a vocal faction saw the westward march of the United States as squeezing and poisoning Japan. Friction intensified at the Washington disarmament conference of 1922, when the United States forced Japan to accept a lower warship ratio than it would have liked. This American diplomatic victory was achieved with the help of the charismatic cryptanalyst Herbert O. Yardley and his assistants, whose solution of coded Japanese diplomatic messages told American negotiators just how far they could push the Japanese.
But in 1929 Secretary of State Henry L. Stimson, believing that mutual trust worked best in international affairs and that therefore "gentlemen do not read each other's mail," refused to expend State Department funds for cryptanalysis. When Yardley, out of work in the Depression, wrote an indiscreet book in 1931 revealing the inside story, Japanese officials lost face, the Japanese press fulminated, relations with the United States deteriorated-and Japan improved its diplomatic cryptosystems. Tokyo adopted machine ciphers more complex than the system employing simultaneous use of multiple codebooks that Yardley and his team had cracked.
But the dissolution of Yardley's agency had not killed American codebreaking. Both the army and the navy had their own units. The navy's began almost by accident, after naval intelligence, having rifled the trunk of a Japanese naval officer in New York in 1923, found itself with a codebook but without means to exploit it. In January 1924 the navy established a four-man unit to gain information from Japanese communications. Its head was Lieutenant Laurance F. Safford, ranked fifteenth in the Annapolis class of 1916. Since the codebook was Japanese, and since the navy regarded Japan as its most likely enemy, Safford started setting up listening posts in the Pacific.
Safford also hired a brilliant, 32-year-old woman who proved to be an outstanding cryptanalyst. Agnes Meyer Driscoll, who had taught mathematics and music and had worked for Yardley, attacked the messages in the photographed code. The book listed Japanese words, syllables and phrases opposite five-digit codegroups, such as 48771, which replaced them. The codegroups were not used plain, however. They were themselves enciphered. It was the job of "Miss Aggie" to break through this armor and recover the original codegroups. Incessantly turning the pages of the photographed book with the rubber tip of her eraser, she eventually succeeded. The codegroups then yielded Japanese text, which a husband-and-wife team translated.
But in December 1930 the code, which by then was 12 years old, was replaced by a new one. The hardest part of breaking a code is getting started. As one of the naval codebreakers explained: "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; 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."
Miss Aggie, who by then had learned the ships' names, the communications patterns and the frequently used phrases of the Japanese fleet, no doubt utilized a form of this process first to dissolve the codeword encipherment and then to reconstruct the new 85,000-group code. For years it gave the U.S. Navy insight into Japanese forces and tactics, as expressed during fleet exercises.
And it was only one source of communications intelligence. During the 1930s, the navy attacked and often solved other Japanese naval codes-for administration, merchant ships, logistics, intelligence-and several cipher machines. When codebreaking failed, traffic analysis provided much information. Traffic analysis infers an organization's structure and operation from message routing, message volume, call signs and operators' chatter. A prime tool, direction-finding, locates a radio transmitter. Much as a portable radio can be rotated to bring in a station most strongly, a sensitive antenna can find the direction from which a signal is coming. If two antennas take bearings on the source of a signal, a control center can draw the lines of direction on a map; their intersection marks the position of the transmitter. Such a fix can tell where, for example, a ship is. Successive fixes can plot its course and speed. A study of call signs, combined with direction-finding, can show which stations are talking to which and which of those are communications centers and, therefore, presumably headquarters. Finally, increased volume in one area may suggest increased activity. The intercept and direction-finding stations that Safford swung in a vast arc across the central Pacific furnished both the intercepts that the cryptanalysts solved and the data for the traffic analysts. All of these sources gave the U.S. Navy a fairly complete picture of the Imperial Japanese Navy's forces, organization and movements throughout the 1930s and into 1941.
While the navy was expanding its communications intelligence organization, so was the army. The Signal Corps had hired in 1921 a 29-year-old who was even then one of the world's great cryptanalysts. William F. Friedman had written some theoretical treatises of landmark importance and had solved German codes in France during World War I. Friedman's job now was to improve the army's cryptosystems. With the closing of Yardley's agency, the Signal Corps added foreign codebreaking to its responsibilities, and Friedman became the head of a new Signal Intelligence Service. The first new hire was Rowlett, who was 21 at the time. After a couple of years of training in cryptanalysis under Friedman, Rowlett and two colleagues began attacking Japanese diplomatic cryptosystems, mainly because diplomatic intercepts were available and Japanese army intercepts were not. The three cracked a simple code that served for personnel matters and expenses of Japanese diplomats. Then they turned their attention to what appeared to be a machine system that seemed to protect the more important messages. The army cryptanalysts reconstructed the machine and named it RED.
That solution made codebreaking the nation's premier source of secret intelligence. It easily surpassed espionage: except for a few local agents who provided observations to naval attachés, the United States had no spies anywhere in the world. And codebreaking outperformed the diplomats, gaining information long before they did and in a highly specific form, namely, word-for-word instructions and reports. In March 1937, for example, the intercepts revealed advance information about Italy's possible adherence to the German-Japanese anti-Comintern pact-six months before American diplomats began reporting on it. Later the intercepts revealed part of the text of the treaty. In 1937, for the first time, solutions of intercepted foreign messages began flowing to the White House.
Then, on March 20, 1939, three messages in a new cryptosystem were intercepted. Within three months, the new system had displaced RED, whose mechanisms were wearing out. Japan's major diplomatic messages had become unreadable. Faced with the loss of the nation's paramount intelligence, the Signal Intelligence Service mounted a concentrated attack on Japan's new machine. Friedman put Rowlett in charge. The Americans called the new machine PURPLE.
Rowlett's team was assigned to two rooms of the Munitions Building. Rowlett worked in Room 3416, his mind focused. He never hummed or chewed his pencil or muttered to himself; he looked out the window onto the neighboring wing only when something distracted him; he never drank coffee, but he did puff on a pipe. Most mornings he held a conference with the other cryptanalysts. Afterwards they collectively scrutinized the intercepts. Most of these had been teletyped in from army intercept stations that monitored the commercial radio circuits used by the Japanese Foreign Ministry for most of its messages. The cryptanalysts' work consisted of matching proposed plaintext-a guess as to the text of the original message-against the text of the cryptograms. The proposed plaintext came from circular telegrams sent simultaneously in the readable RED and in PURPLE during the three months that both systems served, from formulaic diplomatic language, and in a few cases from the State Department's supplying of the text of notes to or from the Japanese ambassadors.
Friedman, who had assembled the team, supervised. He did not engage in any of the actual cryptanalysis but made sure the unit had all the support it needed. Determination pervaded Rowlett's co-workers, who remained confident that they would break PURPLE the way RED had been broken. The summer of 1940 arrived. Rowlett drove his Ford to work. Belgium, the Netherlands, France fell. England braced for invasion. Japan, persisting in its three-year-old aggression in China, seemed now to cast a covetous eye on orphaned French Indochina and the Dutch East Indies. In July President Franklin D. Roosevelt embargoed the export of aviation fuel and high-grade scrap metal to Japan.
Though the cryptanalysts followed these developments in the newspapers, their inspiration came less from the pressure of events than from fascination with the cryptologic problem. They struggled to envision what kind of electromechanical device could produce these groups of incoherent letters from Japanese plain language. A hypothesis emerged. As they sought to test it, construction workers began to build an additional floor above them. Hammers pounded; men shouted. If the cryptanalysts closed the windows to reduce the noise, they sweltered in the non-air-conditioned offices; if they opened them, they could barely think. Still they persisted. Just as the new structure was finished, a fire destroyed it-and it had to be built all over again.
Then on September 20, cryptanalyst Genevieve Grotjan spotted a particular pattern among the letters she was scrutinizing. The 26-year-old codebreaker looked for another. Soon another member of the PURPLE team noticed that she seemed to be concentrating extremely intently. He spoke to Grotjan about it and then he, Rowlett and another cryptanalyst came to her desk. At once they grasped the significance of what Grotjan was showing them: the pattern confirmed the team's theory of how the PURPLE machine worked. One of the team members dashed around the room; another shouted "Hooray!" Rowlett jumped up and down, crying: "That's it! That's it!" Grotjan's discovery capped one of the greatest cryptanalyses of all time: enciphered Japanese diplomatic cables were now comprehensible. And how did the codebreakers celebrate? They sent out for Cokes!
A week later, the day the Tripartite Pact established the Rome-Berlin-Tokyo axis and a day after Roosevelt prohibited the export of all iron and steel scrap to Japan, the Signal Intelligence Service handed in its first two solved PURPLE messages. These two drops of intelligence marked the trickling start of what, a year later, would grow into a stream. By late 1941 solutions of messages in PURPLE and in lower-level Japanese diplomatic cryptosystems soared to 50 to 75 messages a day. The most important of these, selected by army and navy intelligence officers, went to the president, the secretaries of state, war and the navy, the chief of staff, the chief of naval operations and a handful of top-level officials. (The secretary of war was Stimson, who had closed Yardley's unit in 1929 but now welcomed the intercepts. He was not inconsistent: he believed that codebreaking was a legitimate function of a military service but not of a diplomatic one.) These intercepts provided insight into the thoughts and activities of Japan's Foreign Office and corroborated the evidence from negotiations and events, such as Japan's occupation of French Indochina, that matters were approaching a crisis.
For example a PURPLE message on July 31, 1941, from the foreign minister in Tokyo to the ambassador in Washington declared: "There is more reason than ever before for us to arm ourselves to the teeth for all-out war." The occasional instruction to seek a peaceful solution, such as one on November 15, was overwhelmed by belligerent indications. On November 19, the Foreign Office arranged with its embassies and legations that "in case of emergency (danger of cutting off diplomatic relations), and the cutting off of international communications," the phrase "east wind rain" would be added in the middle of the daily Japanese language shortwave newscast to indicate Japanese-American relations were threatened. "When this is heard please destroy all code papers," the Foreign Office instructed. On November 30, Tokyo told its ambassador in Berlin to "say very secretly to them [Adolf Hitler and Foreign Minister Joachim von Ribbentrop] that there is extreme danger that war may suddenly break out between the Anglo-Saxon nations and Japan through some clash of arms and add that the time of the breaking-out of this war may come quicker than anyone dreams."
But the Japanese diplomatic PURPLE and other intercepts did not reveal military or naval plans. The army had not solved any Japanese army codes because it could not intercept enough messages. The navy had made scant progress on the main Japanese operations code, JN25, whose second, enlarged edition, JN25b, had been introduced on December 1, 1940. With Miss Aggie and reserve Lieutenant Prescott Currier leading the attack, the navy had stripped off most of the additive groups that concealed the underlying codegroups, but had recovered only a small number of these, so that by December 1941 only about 10 percent of the text of an average JN25b message could be read. This was due less to Japanese cryptographic superiority than to the navy's insufficiency of cryptanalysts, in part because it was helping the army decipher PURPLE messages (once the machine had been solved) while also helping the British break U-boat messages in the German navy's Enigma cipher machine.
Nor did other sources provide much insight into the specifics of Japanese military or naval planning. U.S. military attachés furnished 90-page reports on weapons, tactics, personalities and organization but had no solid information on the intentions of the Japanese armed forces. The diplomats concentrated on the political situation and, likewise, had little hard information on military developments. In January 1941 Ambassador Joseph Grew duly reported that a Peruvian colleague had heard that Japan was planning an attack on Pearl Harbor. The rumor was false, for at the time no attack was being planned. Washington, in any event, filed the report and forgot about it. William J. Donovan, named in July 1941 to head what became the Office of Strategic Services, had no secret agents and thus no independent source with which to make predictions.
Only one form of intelligence appeared to offer relatively solid information about Japanese naval matters: traffic analysis. For years traffic analysts watched with precision the ships and squadrons of the Japanese fleet and their maneuvers. Two tense situations in 1941 reconfirmed their ability. During the war scare of February, when Japan was moving on French Indochina with the apparent intention of attacking Singapore (which might have brought the United States into the war), various forms of intelligence observed the southward movements of Japanese warships. Traffic analysis added to this. When its monitors picked up no communications either to or from the aircraft carriers, its analysts concluded that these vital units were standing by in home waters as a covering force in case of an attack on Japan's main islands. Later intelligence confirmed this analysis. In July, when Japan occupied French Indochina, carrier communications again went dead. Again traffic analysis decided that the carriers had been in home waters. Twice, then, a complete blank of communications with the carriers, together with indications of a strong southward thrust, had meant the presence of the carriers in Empire waters. A pattern seemed to have emerged.
Throughout the fall tension between Japan and the United States escalated. Negotiations all but ended after an American démarche on November 26, the acceptance of whose demands would have required Japan to pull out of China and in other ways to reverse its foreign policy. Japan then finalized its decision for war with a nation whose population was twice as large and whose industrial output nine times as great as its own. Why did Tokyo make this apparently irrational move? Premier Hideki Tojo told the Privy Council on December 1 that, if Japan were to submit to American demands, "Japan's existence itself would be endangered. . . . Japan has now no other way than to wage war against the United States." Japan had never intended to invade the United States. It planned only to cripple the main American instrument of war-the U.S. Pacific Fleet-and then wait behind a ring of impregnable defenses until the Americans wearied of the struggle and quit, leaving Japan to wax fat on its conquests and to reign as master of East Asia.
American officials did not think Japan would attack their country. To start war with so superior a power would be to commit national hara-kiri. To Western modes of thought, it made no sense. This rationalism was paralleled by a racism that led Americans to underrate Japanese abilities and will. Such views were held not only by common bigots but by opinion-makers as well. These preconceptions blocked out of American minds the possibility that Japan would attack an American possession.
Yet on December 3, 1941, an intercept revealed Tokyo instructing its Washington embassy to burn codes and destroy cipher machines. How else could it be interpreted but as a preparation for war? On the evening of Saturday, December 6, Roosevelt, who had excused himself from a large dinner party, and his aide Harry Hopkins read an intercept in his lamplit White House study as the naval lieutenant who had delivered it watched. "This means war," the lieutenant remembered the president saying in effect. What could he have meant? Perhaps eventual or accidental hostilities, perhaps hostilities following a Japanese conflict with Britain through a move against the Malay Peninsula, at whose tip lay Singapore.
An attack on Pearl Harbor was seen as all but excluded. Though senior army and navy officers knew that Japan had often started wars with surprise attacks, and though the naval air defense plan for Hawaii warned of a dawn assault, officials also knew that the base was the nation's best defended and that the fleet had been stationed that far west not to attract, but to deter, Japan.
The evidence available early in December seemed to confirm American preconceptions. Reports from consuls, attachés, and ships' masters told of Japanese forces moving southward. American reconnaissance airplanes saw Japanese submarines, cruisers, transports and destroyers in Camranh Bay and elsewhere along the French Indochina coast. On December 5, British aircraft spotted three Japanese convoys rounding Cambodia Point and entering the Gulf of Siam. American traffic analysis further indicated an advance to the south. Many Japanese units were heard as they headed that way. Yet no communications went to or came from the carriers. As in February and July, the carriers were apparently in home waters. So certain were American officials of this southward move that, in a staff meeting on the morning of December 7, as Chief of Staff General George C. Marshall was about to warn Pacific commands that Japan was breaking off negotiations at 1 p.m. Washington time, one general shouted out to the communications officer that if there were a question about priority, the first message should go to the Philippines.
As he was saying this, six of the carriers thought by traffic analysis to be lying in home waters were in fact plowing the seas north of Hawaii. They had launched airplanes for a Sunday morning attack on the Pacific Fleet at Pearl Harbor. Japan's strategic plan would prove unsuccessful. The attack so enraged and unified the American people that they would never tire of the struggle but would battle on to total victory. Yet Japan's tactical plan worked to perfection: the raid achieved complete surprise.
American intelligence had failed. Evidence warning of an attack could have overcome American preconceptions, but intelligence-which relied almost solely on the diplomatic transmissions via PURPLE-had found no such evidence. Japan had sealed all possible leaks. The ambassadors in Washington were not told of the attack. Knowledge of it was limited in Tokyo to as tight a circle as possible. Plans for it were distributed by hand to the ships of the task force. No reference to a raid on Pearl Harbor ever went on the air, even coded. The February and July situations misled traffic analysts. JN25b messages intercepted before the attack, but solved after the war, show that even if that naval code had been fully solved and those messages read before December 7, they would not have foretold the attack. And though war with Japan was indeed expected, that expectation did not-could not-imply knowledge of an attack on Pearl Harbor, for it is impossible in logic to leap from a general belief to a specific prediction.
Some historians, pointing out that officials saw the intercepts only day by day, claim that had someone sat down and looked through them all at one time, he or she would have seen a pattern indicating the attack. This is extremely unlikely. There was nothing in either the diplomatic or the naval intercepts about an attack on Pearl Harbor. Indeed such a collation might have pointed away from Hawaii. Of solved messages reporting ship movements between August 1 and December 6, 1941, 59 dealt with the Philippines, 23 with the Panama Canal and only 20 with Pearl Harbor.
In her 1962 study, Pearl Harbor: Warning and Decision, Roberta Wohlstetter argues, "We failed to anticipate Pearl Harbor not for want of the relevant materials, but because of a plethora of irrelevant ones." In the terms of information theory that Wohlstetter uses, this means that the noise was too great for the signal to be picked out. But she errs. There was a dearth of intelligence materials. Not one intercept, not one datum of intelligence ever said a thing about an attack on Pearl Harbor. There was, in Wohlstetter's terms, no signal to be detected. Intelligence officers could perhaps have foreseen the attack if the United States, years before, had insinuated spies into high-level Japanese military and naval circles, flown regular aerial reconnaissance of the Japanese navy, put intercept units aboard ships sailing close to Japan to pick up naval messages that a greatly expanded codebreaking unit might have cracked, or recruited a network of marine observers to report on ship movements. The intelligence failure at Pearl Harbor was one not of analysis, as Wohlstetter implies, but of collection.
For many people, Pearl Harbor remains an enigma. "We were breaking the codes," they cry. "Roosevelt must have known!" Or, as Congress's Joint Committee on the Investigation of the Pearl Harbor Attack put it: "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 simple answer is that the intelligence, good though it was in certain areas, was not good enough.
To some people the intelligence failure was deliberate. They contend that President Roosevelt provoked the attack by his intransigence toward Japan and ensured its success by suppressing intelligence and withholding information from the commanders at Pearl Harbor. His purpose, the theory goes, was to trick a reluctant United States into the war. After the war, when the cryptologic details became public, the conspiracy theories blossomed.
The naval commander at Pearl, Admiral Husband Kimmel, complained that he was taken by surprise in part because he had not been allowed to receive the diplomatic intelligence obtained from PURPLE. But its additional details of the deteriorating situation could not have alerted him more than the November 27 message of the chief of naval operations, which began, "This dispatch is to be considered a war warning."
Safford, who was head of the naval codebreaking and codemaking unit on December 7, insisted that the "winds" code had been executed and that evidence of it had been suppressed. On December 3 or 4, he said, Japan had broadcast "East wind rain," meaning war was imminent with the United States. The intercept with the serial number JD-1:7001 was not in the files; this was obviously the "winds" execute, Safford charged, and it was missing because higher authority had ordered the removal of this evidence. The first problem with this theory, which incriminates Roosevelt and his aides, is that interception of a such an order would not have told anybody anything more than he or she already knew. The second problem is that it raises the incredible picture of Admiral Harold Stark, the chief of naval operations, rummaging through file drawers in an office with which he was unfamiliar to find and abstract a piece of paper. No one ever reported his presence in the office or said they had received orders from him to destroy a government record. Finally, naval officers reported that serial numbers were sometimes canceled for legitimate reasons, such as duplication.
More recently a book by intelligence writer James Rusbridger and former Australian codebreaker Eric Nave, Betrayal at Pearl Harbor, claims that a British codebreaking unit at Singapore, where Nave worked, solved JN25 before the Pearl Harbor attack. This revealed the forthcoming Japanese strike to Prime Minister Winston Churchill, who said nothing about it to Roosevelt because, according to the authors, he wanted the United States in the war with him. A massive coverup, they assert, has kept this information from coming out.
Several flaws destroy this theory. Churchill wanted the United States in the war against Germany, not Japan. A British codebreaker's diary states that the prime minister was surprised by the attack. The code in use from December 1, 1940, was not JN25, but the quite different JN25b. The British and the Americans exchanged codegroup recoveries in this system, so the Americans would have been able to discover the same things the British had. Finally, none of the presumed JN25b solutions the authors cite even mention Pearl Harbor.
Upon similar close examination, the other anti-Roosevelt conspiracy theories likewise fall apart. Most spring from a wish to defend Kimmel or the army commander at Pearl, General Walter Short, or from a hatred of Roosevelt. They demonstrate their fundamental illogic by forgetting, for example, that had the army lieutenant who headed a radar unit at Pearl Harbor, which had spotted the incoming flight of Japanese bombers, not told the radar operators to "forget it," the Roosevelt plot would have been frustrated. They ignore that both Stimson and Frank Knox, the secretary of the navy, were Republicans (Knox had, in fact, run in 1936 as GOP vice presidential candidate against the Roosevelt ticket). They demonstrate their total misreading of character by implying that General Marshall would let people die in so cynical a way and then lie about it. They seem not to realize that, even if Roosevelt had wanted war, he would not have wanted to enter it with his fleet badly weakened. The concocters of these theories are unable to accept that humans sometimes do things wrong or do not do them at all, that accidents happen, that in the complex system that is the world improbable events occur.
Pearl Harbor was the greatest shock ever sustained by the United States. It destroyed the national myth of isolation and invulnerability. This was one of the reasons the United States "unencapsulated" itself after World War II and reached out to other nations in a network of alliances. And Pearl Harbor reverberated throughout the Cold War as a fear of surprise attack. The nation spent billions on early-warning radar lines and intelligence satellites and, for decades, kept its missiles ready to fire within minutes.
These general legacies of Pearl Harbor were joined by two specific ones. Because it appeared that lack of coordination between the army and the navy commands had contributed to the disaster, the joint congressional committee recommended that "immediate action be taken to ensure that unity of command is imposed at all military and naval outposts." Some moves toward this had already been made. During the war each theater had a single commander, who controlled both army and navy units in that theater. Thus Admiral Chester Nimitz commanded all forces within the Central Pacific theater, General Douglas MacArthur all in the Southwest Pacific theater. After the war, and despite bitter objections from the navy, the nation forged a unified Defense Department, which remains today.
The second major development was the centralization of intelligence. It had begun, however feebly, even before the war. In July 1941 Roosevelt created the post of Coordinator of Information "with authority to collect and analyze all information and data, which may bear upon national security . . . and to make such information and data available to the president." The independence and centrality that this implied was, however, curtailed 11 months later when Roosevelt converted the COI into the Office of Strategic Services and subordinated it to the Joint Chiefs of Staff. An espionage and evaluating agency, the OSS served through the war but was abolished by President Harry S. Truman less than a month after Japan surrendered. At the same time, however, and probably with the surprise Japanese attack in the back of his mind, Truman ordered that plans be drawn for "a comprehensive and coordinated foreign intelligence program for all federal agencies concerned with that type of activity." A few months later Congress's Pearl Harbor committee recommended "that there be a complete integration of army and navy intelligence agencies." This gave powerful impetus to Truman's establishment a year later of the Central Intelligence Agency. As former President Herbert Hoover wrote in 1955 in a report of his Commission on Organization of the Executive Branch of the Government: "The CIA may well attribute its existence to the attack on Pearl Harbor."
But the OSS's spying had not produced intelligence comparable to that from codebreaking, which became the Allies' prime source of secret intelligence. The solution after Pearl Harbor of Japanese naval codes engendered three critical American victories: the battle of Midway, the midair assassination of Japan's leading strategist and architect of the Pearl Harbor attack, Admiral Isoroku Yamamoto, and the strangling of the island empire through the sinking of its merchant marine. In the Atlantic, the British-American exploitation of the German Enigma cipher machine, called "the most important sustained intelligence success in the history of human conflict," helped defeat both the U-boats and German land forces. The breaking of PURPLE, though unable to prevent Pearl Harbor, later yielded astonishing insights into Hitler's plans, gleaned from the messages of the Japanese ambassador in Berlin. All of this noticeably shortened the conflict.
This priceless result was achieved by cryptologic organizations enormously enlarged from the prewar handful of analysts. The army's comprised 10,000 men and women in the Washington headquarters alone, excluding theater operations; the navy's, 6,000 likewise. Stimson, believing that one cause of Pearl Harbor was inadequate analysis of the intercepts, established a special unit to correlate the diplomatic solutions with other information and present finished intelligence to the president and other high officials. But codebreaking was divided between the two services, and the cooperation that had begun before Pearl Harbor with a joint assault on PURPLE, and later with the sharing of the work of daily recovery and intercept reading, declined during the war to a mere division of labor and exchange of information.