by John Prados
Philip Jacobsen, who retired as a lieutenant commander after a full career in signals intelligence, came away from Guadalcanal convinced the Allies had penetrated the secret of the enemy withdrawal. He points to the numerous position and mission reports put out by FRUMEL and FRUPAC as evidence. Movement reports, however, absent knowledge of Japanese intentions, could fit many interpretations. More striking was a partial Ultra decrypt of a message to one of the destroyer units assembling for the Tokyo Express. The Washington office that circulated this information commented that the KE Operation might be intended for evacuation. That very word appeared in a late-January dispatch regarding submarine movements. Jacobsen also recalls quiet rumblings among codebreakers at FRUMEL, but Duane Whitlock, who was on the scene at Melbourne, remembered no great excitement there at the time, the opposite of what one would expect if codebreakers held the view that the Japanese were escaping their grasp.
Whatever doubts existed never made their way into the intelligence summaries. Indeed, on February 1 in Washington, COMINCH released a cable for transmission to Nimitz, Halsey, and MacArthur: “INDICATIONS ARE THAT JAP OFFENSIVE OPERATION NOW IN FULL SWING ON MAJOR SCALE PRIMARILY DIRECTED AGAINST SOUTHERN SOLOMONS.” With the Tokyo Express in motion, Pearl Harbor intelligence, which had also predicted an offensive, and observed that it appeared more and more probable, reported no change in its estimate, an observation repeated throughout the withdrawal. On February 6 the doubters in Washington finally managed to get out a cable that asked, “ARE THERE ANY INDICATIONS THAT RECENT TOKYO EXPRESSES MAY HAVE BEEN FOR THE PURPOSE OF EVACUATING NIP FORCES FROM GUADALCANAL?” Halsey’s SOPAC intel people answered the next day: “AS YET NOTHING DEFINITE.”
The second Tokyo Express, also under Admiral Hashimoto, left Shortland the morning of February 4. Its combat air patrol was again pushed aside by a big Cactus Air Force formation, some thirty-three Dauntlesses and Avengers. They damaged the destroyer Maikaze. Two ships in the Koyanagi unit suffered minor impairment. Once more Hashimoto’s flagship was crippled, this time by engine failure. But the armada voyaged to Cactus and loaded about 4,000 more men, including General Hyakutake and his staff. The third Express took place on February 7, with the strongest air escort yet—but was again overcome by the Cactus Air Force. Fifteen SBDs bombed Hashimoto. Lieutenant Commander Soma Shohei of the Akigumo startled his sailors with his bad mood. Soma had a premonition of doom. In fact, she came through fine. A different destroyer was knocked out, and another damaged insufficiently to stop her. Hashimoto closed the last Cactus evacuation beach, and Koyanagi led his second unit to pull out the men holding the barge station in the Russell Islands. The Japanese were gone. Estimates of the sum total of Japanese rescued from Starvation Island range from 10,652 to some 13,030. The Imperial Navy escaped with a single destroyer sunk and others damaged.
When Admiral Koyanagi returned to Truk, he reported to the Combined Fleet chief. Yamamoto confessed that he too had feared when he heard, early on, that Hashimoto had lost a destroyer to damage, but that he had consoled himself with the thought that Koyanagi Tomiji was on the scene. Yamamoto congratulated Koyanagi on a job well-done.
The date-time group on the SOPAC message replying to Washington’s query regarding a Japanese evacuation indicates that dispatch was sent twenty-nine minutes after the final Tokyo Express reached Shortland safely. Bull Halsey had kept his big ships for news of Yamamoto’s carriers and their expected offensive. Now it was too late. On February 5, a B-17 at the limit of her search actually caught sight of the Kondo fleet. Halsey strung his bow to shoot it, but Kondo stayed beyond his reach.
Several days later, according to Lieutenant Commander Ito Haruki of Imperial Navy radio intelligence, the Japanese extended their deception, using the frequency of a Catalina patrol plane, whose own communications had become scrambled, to send a carrier sighting report in its name. Ito, who survived the war without participating in any other major battle, observed that “the fake message which helped the total evacuation of Guadalcanal will be my only consolation.” Senior officers reprimanded Ito for violating Imperial Navy security regulations with this gambit.
Americans who researched this claim found that a PBY had been in the area but could not verify that any Allied command had circulated the phony report. The CINCPAC intelligence summary for that day actually noted that the major enemy fleet appeared to be returning to Truk. The summary for February 9, twenty-four hours later, is worth quoting:
The return of the Advance Force to Truk along with the comparatively rapid advance of U.S. Army forces as far as Visale from the southwest and the Doma Cove area from the east may indicate that the enemy is indeed evacuating from Guadalcanal and that the major operational stage [for] the present…is completed. If this be true it shows that the tide of war in the Pacific has changed and that the Nip is on the defensive at last.
At 4:25 p.m. that day, General Patch’s GIs completed clearing Cactus. He messaged Halsey: “‘TOKYO EXPRESS’ NO LONGER HAS TERMINUS ON GUADALCANAL.” Admiral Chester W. Nimitz’s observations are also noteworthy:
The end was as abrupt as the beginning of the struggle for Guadalcanal. Until the last moment it appeared that the Japanese were attempting a major reinforcement effort. Only skill in keeping their plans disguised and bold celerity in carrying them out enabled the Japanese to withdraw the remnants of the Guadalcanal garrison. Not until all organized forces had been evacuated on February 8 [East Zone date—at Pearl Harbor] did we realize the purpose of their air and naval dispositions. Otherwise, with the strong forces available to us…and our powerful fleet in the South Pacific we might have converted the withdrawal into a disastrous rout.
POLISHING APPLES
Quite close to the end, an event took place that had a huge impact on the war. This was the loss of the submarine I-1 off Kamimbo Bay on January 29. Fresh from the dockyards, where the I-1 had gone to repair balky engines, this I-boat represented a more thoughtful approach to the “mole” (mogura) submarine supply duty. The boat’s after deck gun had been removed and substituted with a waterproofed Daihatsu landing barge. After tests in Empire waters and at Truk, Lieutenant Commander Sakamoto Eichi’s I-1 went on a mole run to Guadalcanal with supplies and sixty soldiers.
Ultra intervened. A series of late-January messages had mentioned I-boat cruises to Cactus on several succeeding days. The I-1 appeared in the traffic. Though only one decrypt confirmed a mogura mission, and two others actually spoke of cancellation, SOPAC warned Guadalcanal to be alert for subs over a three-day period. Several radio direction-finding stations in New Zealand tracked I-boat transmissions showing subs nearing Guadalcanal. Based on this, Cactus naval command alerted all units to a Japanese sub off Kamimbo. On the night of the twenty-ninth when Commander Sakamoto surfaced, lookouts reported torpedo boats in sight. The I-1 dived. But the “torpedo boats” were actually the New Zealand corvettes Moa and Kiwi. The latter detected the sound of Sakamoto’s I-boat. Lieutenant Commander Gordon Bridson’s Kiwi dropped depth charges. The I-1, now with damaged steering, engines, batteries, and leaks, went into an uncontrolled dive far past her design depth. Desperate to save his ship, Sakamoto managed to regain control and blew his ballast tanks to bring the I-1 to the surface, where he rushed toward shore on his one good engine, trying to beach the submarine.
Commander Bridson went to full speed, intending to ram. He rejected the complaints of officers who warned of the danger of collision damage, holding out the possibility of shore leave during repairs. Kiwi’s four-inch gun, illuminated by star shells from Moa, set afire the landing barge on I-1’s deck and killed Sakamoto and most of those on the conning tower. The executive officer, Lieutenant Koreda Sadayoshi, took command and tried to defend the boat, now unable to submerge. Bridson rammed the I-1 three times. During one of these collisions the sub’s navigator tried to board the New Zealand corvette and attack with his sword, but ship movements shook him loose and he fell into the sea, rescued to enter the POW cages. The sub ran aground on a reef 300
yards from the Guadalcanal shore.
Loss of the I-1 set back the Imperial Navy’s effort to develop specialized transport subs—but far worse was the loss of secrets of enormous value. Of the Japanese soldiers, nineteen survived. Of the sailors, forty-seven reached shore, taking with them papers that included the current version of the JN-25 code, alternately reported burned or buried. But later they realized that a case of documents had been left behind. With little time, a working party of Koreda and a couple of crewmen, along with sailors from the Japanese evacuation flotilla, were unable to demolish the sub. The survivors left with the destroyers. Koreda reached Rabaul, where NGS officers questioned him. Now the Japanese were anxious to gut the wreck with its secret trove. By February 8, Ultra had intercepted a message expressing dismay that codes had been compromised. On February 10, a strike by Buin-based Vals of the 582nd Air Group made a bomb hit on the wreck, though most planes failed to find it. The next day Lieutenant Koreda sailed from Shortland aboard Lieutenant Commander Inada Hiroshi’s I-2 in yet another desperate attempt. Ultra revealed that too, and the Allies interfered. The wreck eluded the I-2 twice. On the first try Inada could not find her in the dark bay at night; on the second, during the night of February 15, PT boats depth-charged Inada’s sub and an aircraft finally drove him away.
By then it was too late anyway. The New Zealand corvette Moa had nosed around the wreck the morning after the battle. By some accounts codebooks were taken then. On February 11, Army G-2 officers went to the site aboard PT-65 and discovered much intelligence to exploit. Submarine rescue vessel Ortolan returned on the thirteenth, and divers salvaged documents including five codebooks, which were delivered to the Cactus Crystal Ball. Phil Jacobsen remembers the red-bound books, weighted with lead to be thrown overboard if necessary, and how the signals intelligence people worked to restore the waterlogged pages. Clean sheets of dry paper were placed between each leaf and the codebooks put on a hot radio receiver whose vacuum tubes did the job. On February 15, Cactus naval base reported acquiring seven codebooks, including two in the JN-25 code. CINCPAC immediately instructed the South Pacific command to forward the material by aircraft as quickly as possible.
Some historians have objected to an estimate in a history of Japanese naval communications that the I-1 disaster resulted in the loss of 200,000 codebooks. This is a misinterpretation. The number refers to the copies of various code publications throughout the Imperial Navy that had to be replaced as a result of the compromise, not to the books actually lost in the submarine. Others have minimized the importance of the windfall, arguing that the current version of the code had actually been destroyed. That observation fails to take into account how codebreaking worked. JN-25 was a “book code” of groups of letters that stand in for words—and there were thousands of words in the Imperial Navy’s fleet code. Until the I-1 incident, Allied codebreakers never even knew what all the words were, did not know how the Japanese rendered certain technical terms, and remained free to dispute meanings. Even an old codebook settled those matters. Commander Thomas Dyer, codebreaker extraordinaire at Pearl Harbor, would later comment, “It was very useful to have a complete code, fleet vocabulary. It settled a number of arguments.” Suddenly the codebreakers had a panoramic view of JN-25’s structure.
In addition, the I-1 yielded lists of the Imperial Navy’s geographic designators, radio call signs, short-time and area codes, and a wealth of technical data on Japanese subs. A petty officer rescued from the crew furnished supplementary detail. Equally important, the Japanese considered the loss an intelligence crisis of the first order. The Imperial Navy immediately declared a cryptologic emergency, changed the additive tables used with JN-25, and began compiling a new codebook with different values. But they did not change the basic codebook. After the I-1, Allied codebreakers were two steps ahead on JN-25 throughout the war.
The Guadalcanal intelligence bonanza burst upon an Allied intelligence community morphing into new configurations. In the works for months, these changes created an even more dominant juggernaut. Some changes resulted from sincere efforts to improve the information that underlay operations, some from personality clashes; others could be traced to officers jockeying for position. Several strands of this story began and evolved through the first year of the war, but they came together early in 1943.
On the technology side there were two key developments. One resided in the increasing use of punch cards and mechanical sorting devices. The basic work of penetrating JN-25 and the other Japanese codes was accomplished by hand. Progress remained dependent on the ability to cross-reference possible discoveries, and there the card sorters proved critically important in accelerating the pace. The machines were not new, but the war brought great numbers of the sorters into service, and that injected fresh vigor and speed. “Radio Fingerprinting” (RFP) was a new development. It had long been recognized that experienced radiomen became so familiar with the circuits they monitored that, much like handwriting, over time they could recognize the “hand” of a given Japanese operator. The RFP technique capitalized on this phenomenon and regularized the process. An oscilloscope would be connected to the radio receiver, and whenever monitors overheard a new enemy operator, oscilloscope photos would be taken of his transmission, identifying his method and “hand.” Over time intelligence accumulated an extensive file of these “fingerprints,” which helped radio traffic analysts follow changes in the Japanese system. RFP helped identify the key enemy messages—which would be given to the most trusted operators for transmission—and it could reveal the character of a given ship or unit or even its identity when call signs were not available.
The most important organizational change concerned merging different pillars of intelligence to produce “all source” data. When the Central Intelligence Agency in the 1990s trumpeted so-called “fusion centers,” it was really reviving this concept from World War II in the Pacific. Not long after Pearl Harbor, Washington began considering fusion centers, and soon after that the Marine Corps commandant suggested these be created in the field, starting with Pearl Harbor. A Washington delegation went to Hawaii to talk up the fusion concept, its Navy representative Commander Arthur H. McCallum, longtime chief of the Japan desk at the Office of Naval Intelligence. McCallum knew everybody in intel, but not Admiral Nimitz, and he was also afraid to broach the subject with Captain Edwin Layton, CINCPAC’s fleet intelligence officer, who might feel threatened by a joint center. McCallum was right about Layton, but that turned out not to matter. Shortly before Midway, Chester Nimitz put his name to a paper accepting the concept. A week after that battle, Admiral Ernest J. King approved the creation of an intelligence center. Layton, meanwhile, was wrong about the bureaucratic threat—his job with CINCPAC was secure so long as he wanted it.
Admiral Nimitz was notorious on the subject of staffs—none could be small enough for him. When Commander McCallum first broached the intelligence center with CINCPAC, Nimitz laughed as the emissary suggested a unit of 120 people. He couldn’t fit that many on his flagship, Nimitz countered. But the realities of command across the Pacific soon forced the admiral to move CINCPAC headquarters ashore, and size became less problematic. In July 1942, seventy-six officers and men were ordered to Pearl Harbor to create the Intelligence Center Pacific Ocean Area (ICPOA), which grew like Topsy. The center absorbed Commander Jasper Holmes’s combat intelligence unit, then an Army map unit, then newly trained language officers to interrogate prisoners, and later Pearl’s photographic interpretation group. Station Hypo, reconstituted as FRUPAC, worked alongside ICPOA. At a December 1942 conference with Nimitz, despite his opposition Admiral King affirmed the decision to expand the center and have the radio spies report to it. In September 1943 the fusion took final shape, becoming the Joint Intelligence Center Pacific Ocean Area under an Army brigadier general. By war’s end 1,700 people worked there, and Admiral Nimitz appreciated every single one of them.
The codebreaking group mushroomed too, even with the transfer to ICPOA o
f Commander Holmes’s unit, by then a virtual FRUPAC ancillary. At the time of Midway there had been 168 persons—including radio operators—in the Station Hypo codebreaking unit, with just two cryptanalysts and three traffic analysis experts. By Eastern Solomons, FRUPAC had grown to 283 people, among them fifty-four experts in the code or traffic work or in the Japanese language, plus forty-six additional radio intercept operators in training. The Navy soon ordered FRUPAC personnel increased to 500, with seventeen more code experts and twenty-four extra language officers. Having outgrown its quarters, in early 1943 FRUPAC moved to a huge new wood frame building near CINCPAC headquarters on Makalapa Hill. The Intelligence Center moved with it. Construction mirrored the burgeoning expansion: Before the end of the war FRUPAC had taken over the entire building, and an identical one had been built next door for the Joint Intelligence Center, which would also have an advanced echelon on a captured Central Pacific island. By then the U.S. Navy alone was operating 775 receivers across the Pacific entirely devoted to intercepting Japanese message traffic.
One man who did not make the move was Commander Joseph J. Rochefort, FRUPAC’s eccentric chief. Until the fall of 1942, Rochefort had personally handled every Ultra decrypt at Pearl Harbor. Smoking his pipe, sitting at his desk in a bathrobe, Rochefort had penciled code solutions on pads used for that purpose. In a way, what happened was a harbinger of codebreaking come of age: The exponential growth of the effort, the increasing numbers of messages in the system—tokens of Ultra success—forced a mass-production approach. Joe Rochefort was an icon of a past age, when the gentleman codebreaker held sway. But he was also a victim of envy and ambition.
The codebreakers’ head office in Washington—Station Negat in the cable addresses and Op-20-G by its Navy Department title—was headed by Captain John R. Redman and seconded by Commander Joseph N. Wenger. Redman, a communications expert, went to CINCPAC in the fall of 1942, leaving Wenger, promoted to captain, in charge. They denied Rochefort the award of the Distinguished Service Medal after Midway. Captain Wenger disliked the Hypo chief and hated the way Rochefort ignored Op-20-G directives. But Hypo had been right about Midway and Op-20-G wrong, and Rochefort saw no reason why FRUPAC should abandon work on key codes of which it had the deepest knowledge just because Wenger wanted the glory for Negat. Rochefort got Admiral Nimitz to approve a cable stating that FRUPAC worked only for CINCPAC, and only through him for Washington. Captain Layton also believed that Op-20-G harbored unduly alarmist fears of Japanese masterstrokes and counseled Nimitz to listen to Rochefort. Thus Nimitz protected Rochefort, untouchable after Midway.