by Ian W. Toll
BASED ON THE EARLY INTELLIGENCE READINGS, Nimitz had gained the confidence to send Admiral Halsey and the aircraft carriers into the Marshalls on February 1. As predicted, there had been no major enemy fleet units in the vicinity. It was Hypo’s first major achievement of the war. But the greatest benefit of Halsey’s raid, from the cryptanalysts’ standpoint, was the accompanying outpouring of fresh Japanese radio traffic. Hundreds of messages were intercepted by Allied listening posts throughout the Pacific, and rushed to Hypo for decryption. Raw intercepts of any kind were potentially useful to the codebreakers, but transmissions linked to known events in a known geographic location were invaluable. In such messages one found the same recurring externals: geographic designators obviously pointing to the Marshall Islands; radio call signs for Japanese ships, units, and commanders known to be present at the time of the raid; and code groups referring to enemy aircraft carriers.
Hypo had entered a virtuous circle. Nimitz, emboldened by intelligence reports that grew ever more tangible, authorized riskier and more ambitious carrier raids into Japanese waters. Each new raid prompted the Japanese to spew forth an immense volume of raw radio traffic, which in turn accelerated Hypo’s progress. By the end of February, the codebreakers were fully in the game. Rochefort’s team had correctly determined that all two-letter geographic designators beginning with “A” referred to American (or formerly American) islands in the central Pacific. “P” stood for the Japanese mandated islands of the central Pacific; “R” were British-held islands in the South Pacific; “M” was the Philippines. The second letter identified specific places within these spheres. In March, a message to “AA” requested reconnaissance on ships currently in the harbor of “AK.” The string of intercepts, linked to corroborating data, left little doubt that these referred to Wake Island and Pearl Harbor.
On March 9, Japanese air group commanders in the central Pacific were radioed a routine forecast of wind force and direction at “AF.” Based on weather patterns, Rochefort surmised that “AF” might be Midway.
Intelligence, as Layton observed, was “a perishable commodity,” and Rochefort agreed: “It is useless to obtain intelligence . . . unless you use it.” Their field was yet unproven, and the debacle of Pearl Harbor had not enhanced its reputation in the eyes of the fleet. Communications intelligence was a new and untried field, and it was simply unfamiliar to many officers. Acting upon predictions derived from intelligence sources even seemed to contradict Mahan’s dictum that a commander should arrange his forces according to the enemy’s capabilities, and not his inferred intentions. These were formidable barriers to the proper use of intelligence. Both Layton and Rochefort were aware of the need to “market” their product to policy makers and senior commanders who could act on it in a timely fashion. But how to disseminate their findings without alerting the Japanese that the codes had been breached?
A bedrock tenet of communications intelligence was that the enemy must always be encouraged to “feel safe,” and never given cause to suspect that his radio transmissions were less than impenetrable. That the Americans were tunneling into the Japanese naval code was one of the most closely guarded secrets of the war. “Ultra” was the name given to intelligence based on decrypted Japanese messages, and only a small, lofty circle of officers were “cleared for Ultra.” When information derived from Ultra had to be distributed outside the privileged circle, it was sanitized. For example, a predicted Japanese fleet movement derived from radio decrypts might be described as a submarine or aircraft scouting report. The trick was to protect the vital secret while also ensuring that fresh intelligence was finding its way to commanders who had the power to act upon it. “This business of secrecy, you see, is a sort of self-defeating thing,” Rochefort later said. “If you don’t tell anyone about this, how can it be used?”
The shroud of mystery surrounding Hypo aroused enmity throughout the Navy Yard. Whenever Rochefort needed more men, supplies, or square footage, he was inevitably told to explain his need in greater detail. This he could not do. Instead, he appealed directly to Admiral Bloch, commander of the Fourteenth Naval District; and Bloch always decreed that Rochefort should get whatever he wanted, and with no questions asked. Supply officers were not accustomed to being manhandled by a mere commander who headed a shadowy unit that they were not even permitted to ask questions about, but they quickly learned not to deny Rochefort’s requests.
Hypo was one unit within a widely dispersed radio intelligence sector that was managed from Naval Headquarters in Washington. The largest unit in the system was “OP-20-G,” the Washington-based Navy Radio Intelligence Section. Before the war, there had been additional intercept stations on Corregidor in the Philippines (Station “Cast”), on Guam, and on Bainbridge Island in Puget Sound. The British had an excellent cryptanalytic team at Singapore. (The Japanese onslaught swallowed up the unit at Guam. Later, the Singapore unit evacuated to Ceylon, and Station Cast absconded in a submarine to Melbourne.) Throughout the Pacific, smaller subsidiary D/F stations measured radio direction-finding data on enemy transmissions. When the system functioned properly, information flowed more or less freely between them. Freewheeling collaboration was vital, as breaks made by one unit could lead directly to breaks made by another. All built on the collective effort.
But competition and dissension were in the wind. The surprise attack on Pearl Harbor had been a cataclysmic failure of communications intelligence. The debacle had cast a long shadow over the entire intelligence-gathering branch of the military. Being located on the spot at Pearl Harbor, Hypo was the target of much innuendo, though history would eventually show that the most damaging failures had occurred in Washington, as a result of a hard-fought bureaucratic turf battle between the offices of Naval Intelligence (ONI) and Naval Communications (ONC). In any case, the high priests of the navy agreed that a shakeup was in order. Admiral King and several of his chief subordinates, including war planning chief Admiral Richmond Turner, believed that radio intelligence should be streamlined and centralized in Washington, and that the satellite units of Hawaii and the Pacific should be brought under the firm control of the commander in chief of the U.S. Fleet (COMINCH).
In February 1942, the prewar chief of OP-20-G, Commander Laurence Safford, was removed from his post and replaced by Captain John R. Redman, Naval Academy class of 1919, who had no experience in codebreaking or communications intelligence. The captain’s elder brother, Rear Admiral Joseph R. Redman, was named director of naval communications that same month. The brothers, along with Admiral Redman’s deputy, Commander Joseph N. Wenger, would form a powerful axis in Washington. Safford and Rochefort had enjoyed a relationship based on personal trust, with the understanding that the different units in the system would not vie for credit. With Safford purged, the trust and rapport were gone, and relations between Hypo and Washington quickly turned sour. Wenger argued that Washington should “assume active coordinating control . . . of all intercept stations, DF nets, and decrypting units,” and set up at the headquarters “a central coordinating authority for all communications intelligence activities.”
Admiral King, who was favorably inclined to consolidating power and authority in his own headquarters, gave this plan his peremptory approval on February 6. With a stroke of his pen, King had relegated the team at Hypo to a subsidiary role. But the prickly Rochefort was determined to preserve the independence of his unit, which he correctly identified as the best codebreaking outfit in the service. (Indeed, he went further: “I would say with all modesty that this was the best communications intelligence organization that this world has ever seen.”) Rochefort and the Redmans were rivals from the start, then enemies—and their feud would very nearly spoil the American victory at the Battle of Midway.
In February, as the Redmans took the reins in Washington, tension was already building between Rochefort and his new bosses. Everyone in the system had access to the same message decrypts and radio traffic reports, but they were seeing different patterns
and drawing different conclusions. Even as JN-25 began yielding up its secrets, the decrypts were mostly a series of blanks interspersed with a few tantalizing fragments of broken code. Radio traffic readings were subject to conflicting interpretations. Rarely did all the available evidence point in one direction. More often there were red herrings, and the analysts had to sift through huge amounts of data and develop theories based on deduction and inference. Often their predictions were little more than sophisticated guesses, which for the sake of decorum they called “estimates.” In Rochefort’s view, Washington’s analytic work was shoddy and its conclusions too often flew in the face of common sense. Nor did he trouble to disguise his contempt for the Redmans. In his eyes, they were empire builders, in the game for their “own personal purposes, not for what they could offer.” They had foreseen that communications intelligence was a burgeoning field, and hoped to ride its coattails to higher rank, pay, and “personal glory.”
Repeatedly throughout February and March, Admiral King predicted Japanese carrier attacks on Hawaii, Midway, the west coast, the Panama Canal, and Allied-held islands on the Hawaii-Australia line. On February 6, in a dispatch to Nimitz and several other Pacific subcommanders, he reported that the Japanese operations in the ABDA theater “may well be accompanied by strong raids against Midway, Oahu, New Hebrides, Northeast Australia, and possibly west coast or Canal.” Later the same day, again, King predicted carrier strikes against New Hebrides and New Caledonia. Where else? A skeptical Rochefort replied that the offensives were aimed at Java and Sumatra (they were). On March 11, the COMINCH warned that “recent enemy air and submarine activities may well indicate another full-scale effort against the Hawaii-Midway line with the likely principal objective of crippling or destroying our vital base at Pearl Harbor.” Hypo countered that the Japanese were pushing south, and lacked the means to mount simultaneous large-scale operations in the central Pacific. On March 18, Nimitz gave his positive endorsement to Layton’s reply, stating, “No indication of immediate major offensive action except in Malaya area.”
It appeared that OP-20-G analysts were nourishing King’s anxieties by sending him cherry-picked data that could be interpreted as pointing to such attacks. The men who had King’s ear were unduly alarmist, and their impulsive theories might incite the fleet to chase its own tail. In the best case, this would waste time and fuel; in the worst, it would leave the fleet out of position to repel a concerted enemy thrust. Rochefort and his team, having lived through the rumor-plagued aftermath of the raid on Pearl Harbor, had learned to filter the evidence with hearty skepticism. Rochefort concluded that Washington was lost in the fog of war, and he was not afraid to say so. That he was often vindicated after the fact did not endear him to Wenger or the Redman brothers.
HYPO EXPANDED STEADILY, and by April, the basement was overcrowded. Rows of desks stretched from wall to wall, occupied by men wearing green eyeshades to protect their eyes from the overhead fluorescent lights. They hunched over their notes and decrypts in an attitude of intense concentration. Several of the key cryptanalysts rarely left the basement, sleeping on field cots shoved up against the walls, or in their chairs with their heads laid down on their desks. Rochefort grew concerned that Thomas Dyer, perhaps the only other indispensable man on his team, would buckle under the strain. In March, he began pushing Dyer out the door every few days, with orders to catch a night of sleep in his own bed. Rochefort himself was usually found at his desk chair, behind a barricade of stacked decrypts, wreathed in a haze of blue smoke from his pipe. One officer estimated that Hypo did a full year’s worth of peacetime work in the first three months of the war, and the pace did not slacken in the least until after the Battle of Midway in June 1942.
Jasper Holmes updated the position of all known ships—American, Allied, and Japanese, including warships, merchant ships, and submarines—on a big plotting board on one of the basement’s walls. Stacks of decrypts and cardboard boxes of IBM punch cards covered all the horizontal surfaces, until finally the yeomen began stacking them on the floor, all around the desks and along the walls. Paperwork management was provisional and makeshift. Rochefort and his principal analysts knew they ought to devise a proper filing system, with cross-indexing of archived messages, but they never found the time for that. Somehow, through the blizzard of decrypts and IBM cards, order prevailed over chaos. “This is one reason why these people are mostly crazy,” Rochefort later recalled. “We’d have no problem at all.”
You’d mention something and you’d say, “Now wait a minute. Back here when they were around Halmahera on their way down to a landing at Port Something-or-other, there was a message like this. Let’s have it.” And they’d look in this pile of junk and they were able to locate it. . . . And then of course, you’d get a new one here and this leads to another thing over here and this leads to another thing and this is how you fill the whole works up. One letter leads to another and that leads to a third one and so on. Then that’s when your memory comes in very handy.
Holmes added that a cryptanalyst “needs only time, patience, an infinite capacity for work, a mind that can focus on one problem to the exclusion of everything else, a photographic memory, the inability to drop an unsolved problem, and a large volume of traffic.”
At first, decrypted externals were like plaintext islands in a sea of indecipherable code, but even a small island supplied a foothold for logical deduction. With time, the sea fell and the islands began to merge with their neighbors. Theories were checked against traffic analysis and sighting reports. There were fewer blanks and more comprehensible phrases. Then there were near-complete messages with a few mulishly indecipherable fragments embedded in them. When a newly decrypted code group confirmed the deductions they had made earlier, the codebreakers rejoiced. Each new triumph bolstered their conviction in the deductions they could not yet prove. Dyer compared the sensation of cracking one of the Japanese code groups, after many weeks of frustration, to a sexual orgasm. “Physiologically, it’s not the same,” he added quickly, “but the emotional feeling is pretty much the same.”
Not every break brought instant enlightenment. Japanese is a language rich in nuance and subtext. Not infrequently, a fully decrypted plaintext message made no sense to the analysts because they could not properly translate the Japanese words. Even the most competent and seasoned Japanese-language officers might disagree with a colleague’s translation. Japanese naval idiom was particularly multifarious, and it seemed to have evolved rapidly even since the late 1930s. Moreover, Pacific place names gave the Hypo analysts a lot of trouble. Over the centuries, the theater’s innumerable islands, towns, bays, and shoals had accumulated native names, European names, and Japanese names. Many places were known by multiple names, and even the same name might have various spellings. Analysts could tap into a large in-house library of charts, piloting guides, and gazetteers that had been published in various languages and eras. They spread old charts out on tables and bent over them with magnifying glasses. Even when the Japanese used a Western name that could be found on an old chart, that name was rendered in kana (Japanese syllabary)—and the Japanese sequence of syllables might sound very different from the original. It took Jasper Holmes some time to recognize “WO-DO-RA-KU” as “Woodlark Island.”
The first inklings of the impending Japanese offensive against Port Moresby in the south Pacific came in the last week of March, when intercepted messages revealed that Japanese naval air units had received orders to attack a target identified as “RZP.” At first the geographic designator remained a mystery, but the volume of traffic and the repeated appearance of the term left no doubt that a major push was in the works. Further parsing of related messages from all stations revealed that the operation, whatever it was, was code-named “MO.” The drive to break these identifiers was lifted to highest-priority status.
Intelligence culled from many different sources pointed toward the Coral Sea. In early April, Japanese air units based in the Solomons were heavily
reinforced. Australian coast-watchers confirmed that enemy air patrols in the region were being extended by length and density. Movement of forces through Truk suggested a shift of forces into the South Pacific. British codebreakers confirmed Rochefort’s hunch that Nagumo’s foray into the Indian Oceans was merely a raid, and not the beginning of a sustained operation. The Japanese carriers were heading back to Truk, from whence they would depart for the South Pacific. Rochefort informed Nimitz that “an offensive in the southwest Pacific is shaping up.” On April 3, he forecast that the offensive would stage out of Rabaul and would be aimed at the southeastern end of New Guinea. But when? Station Belconnen in Melbourne (the group that had evacuated from Corregidor) forecast April 21, but Rochefort believed that was too early. The first week of May seemed more likely.
At about this time, Admiral King reached down from on high and tapped Joe Rochefort on the shoulder. In a message addressed to the Hypo unit commander, the COMINCH asked for an overall assessment of Japanese intentions. The query short-circuited the chain of command, and came as a bit of a shock. “We were a little surprised that he would ask us what our views were,” Rochefort said. “I personally felt that he was not even aware of our existence.” Rochefort and his lieutenants dropped everything else and focused on their response, which was on its way back to King in less than six hours. It contained four substantive points. First, the Japanese had completed their current operations in the Indian Ocean, and the carrier groups were on their way home. Second, an offensive into the Coral Sea area was imminent, and would be aimed at securing the southeastern end of New Guinea. Third, there was no evidence that the Japanese intended an invasion of Australia. Fourth, another major operation was in the planning stages, to take place not long after the Coral Sea offensive. Details of time and place had not yet come into focus, but it would be a very large operation, involving “most of the units of the Japanese fleet.” The response was a fine balance of logical deduction and acknowledged uncertainty. Even then, Rochefort suspected the as-yet-undetermined offensive was aimed at Midway, but he lacked sufficient evidence to back up his hunch with a concrete forecast.
