The Emperor's Codes

by Michael Smith

  The British and the Americans had already confirmed that RZP stood for Port Moresby and the designation Operation MO given to the scheme by the Japanese left little room for doubt. Five days later, Colombo identified the Japanese 1st Air Fleet as being ‘concerned with an operation against RZP’. The following day, the British code-breakers reported that the Japanese aircraft carriers the Shokaku and the Zuikaku had been detached from Admiral Nagumo's 1st Air Fleet and sent to Truk to lead the Operation MO strike force.

  The amount of intelligence flooding in from the previously impenetrable JN25b was now immense. Only around 20 per cent of the messages that were being intercepted could be read and most of those only partially, but the codebreakers were now providing MacArthur and the Commander-in-Chief of the US Pacific Fleet, Admiral Chester W. Nimitz, with extensive intelligence on Japanese operations.

  But this ‘priceless advantage’ was very nearly brought to an end by the first in a series of security lapses. On 27 April a story which appeared in the Washington Post under the dateline of ‘Allied HQ Australia’ spoke of a major concentration of Japanese naval vessels in the Marshall Islands ‘apparently preparing for a new operation’. General George C. Marshall, the US Army Chief of Staff, sent a furious signal to MacArthur pointing out that if Japanese intelligence spotted the story it would be ‘justified in believing their codes broken – which would be disastrous’.

  By now the Japanese had at any event ordered a new codebook to be introduced on 1 May and, for the first time, the additive tables used to encipher the encoded message were to be changed at the same time, promising to hamper the work of the codebreakers severely. But for a few crucial weeks the Japanese naval operators did not receive their copies of the new book and were forced to continue using the JN25b codebook with the old additive cipher tables. ‘The Japanese had the first of their many difficulties with distribution,’ MacInnes recalled. ‘The life of the table was extended and the total number of readable messages rose to over a hundred daily before the book and the table expired on 25 May 1942.’

  Those messages were to be among the most crucial read by the codebreakers throughout the Pacific War. Led by Rochefort's men in Hawaii but with important contributions from both Colombo and Fabian's unit in Melbourne, the codebreakers could trace virtually every major movement made by the Japanese throughout the final days of the JN25b codebook.

  Fully briefed by the codebreakers, who laid bare both the organization and plans of the Japanese forces preparing to occupy Port Moresby, Nimitz ordered Rear Admiral Frank ‘Black Jack’ Fletcher to prepare the aircraft carriers USS Lexington and USS Yorktown for action. The resultant Battle of the Coral Sea was the first naval battle in which surface ships failed to engage each other, relying entirely on carrier-borne aircraft to destroy the enemy.

  It was to be a catalogue of errors by both sides. Large numbers of US and Japanese aircraft were launched against minor targets misidentified as being of much greater importance. The elements of farce were accentuated by the Japanese pilots’ lack of night-flying experience. At one point the Japanese flagship sent out a homing signal, giving the American intercept operators stationed on board the Yorktown and Lexington the location, speed and course of the main Japanese strike force. A further Japanese force of twenty-seven aircraft set off in an abortive mission to find the American aircraft carriers. Having failed to find any sign of their targets, they jettisoned their bombs and headed back to the Japanese strike force, only to blunder upon the American force in the dark. Nine Japanese aircraft were shot down by American fighters sent up to intercept them and the pilots of a number of others became so confused they tried to land on the deck of the Yorktown.

  The Americans did, however, succeed in sinking the Japanese carrier Shoho. The Japanese managed eventually to sink the Lexington. But they also thought they had sunk the Yorktown and a number of other ships, some of which had either existed only in the imagination of Japanese pilots or had not been hit, and as a result claimed a major victory. In terms of vessels and aircraft lost, it was more accurately described as a draw, the Americans losing more ships, the Japanese more aircraft. But overall the victory belonged to the Allied forces. The Japanese were forced to abandon their plans to occupy Port Moresby. The threat to Australia and the supply lines with America was averted. The Japanese had lost their first major warship and for them the aircraft and pilots downed would be far more difficult to replace than any of the losses suffered by the Americans.

  If the battle had been dominated by blunders, it had been won by superior intelligence, produced in Hawaii, Colombo and Melbourne. After the failure to predict the Pearl Harbor attack, Signals Intelligence had proved its worth. Nevertheless, there had been drawbacks for the Allied codebreakers. Fletcher had shown himself prone to ignoring intelligence provided by the Radio Intelligence Unit based on board the Yorktown and appeared to have totally failed to understand the potential and capabilities of tactical signals intelligence.

  The codebreakers also made their own mistakes, in large part, the British believed, because the Americans were too quick to jump to conclusions. ‘During the whole of the JN25b period it was apparent that the US Navy units had not yet learned the art of book-building, being too ready to adopt doubtful recoveries as confirmed,’ MacInnes said. ‘Two discrepancies whose origins can be traced directly back to JN25b were the wrong identification of the new Japanese aircraft carrier Shoho sunk at the Battle of the Coral Sea as the Ryukaku and the confusion as to whether prior to the outbreak of war the main Japanese fleet had assembled at Truk or at a port in Kyushu. The latter, however, was righted in good time for the Battle of Midway.’

  The Battle of Midway, widely regarded as the turning point in the Pacific War, was the result of a calculated strategy by Yamamoto to draw the US Pacific Fleet en masse into an ambush. Despite the loss of the Shoho, the Japanese had a massive naval superiority over the Americans. But it could not last. The US Navy had huge numbers of ships under construction. If it was not beaten into submission now, it would eventually win control of the Pacific by sheer weight of numbers.

  Yamamoto decided to pre-empt this by seizing Midway, just over 1,000 miles west of Pearl Harbor, using an attack on the western Aleutian Islands, off the Alaskan coast, as a diversion. Japanese occupation of Midway, with its potential as a base for an attack on Hawaii, would leave the US Navy with little choice but to commit the bulk of the Pacific Fleet to a concerted counter-attack. The Japanese Combined Fleet would be waiting and its superior forces would prevail. Hawaii could then be taken with ease and with luck the Americans might even sue for peace.

  The preparations for the Midway attack were being picked up by the codebreakers even before the Battle of the Coral Sea. There was frequent mention of a ‘forthcoming campaign’ in association with the cover designator AF, which Corregidor and Colombo realized was Midway as early as 7 March. Rochefort agreed but Washington did not accept this deduction, believing that the thrust of the Japanese attack might be Hawaii, Alaska or even the West Coast of America. More proof was needed if the codebreakers were to win the day and Hypo began taking wireless operators off the training courses and putting them straight to work, Phillip Jacobsen recalled.

  The need for operators on watch was dire so as they met the qualifications they were assigned to operations watch sections in the same building in Wahiawa. I think I was about the seventh or eighth operator to go on watch, to the best of my memory, just before the Battle of Midway. I do recall seeing those in charge poring over navigational charts in their office. They talked in hushed tones, but I could see the charts with two tracks from the Marianas Islands and the main Japanese islands eastward to a point somewhere around Wake Island. Some general knowledge filtered down to us operators that a large Japanese operation was pending and Midway was mentioned as the possible target. This probably came from Rochefort to Intercept Watch Supervisors to be vigilant on certain circuits to ensure their coverage and for immediate transmission of vital information
to Rochefort's group by teletype.

  By early May both traffic analysis and deciphered JN25 messages were indicating the massing of Japanese naval vessels, in preparation for the impending operation, in Saipan, the most logical base for an invasion of Midway or Hawaii. A message from Nagumo intercepted on 16 May revealed that the Japanese proposed to stage their air attacks from a point fifty miles north-west of ‘MI’ two days before the invasion. Hawaii, Melbourne and Colombo were all agreed. But still OP-20-G refused to accept that Midway was the target. So Rochefort and one of his senior analysts, Jasper Holmes, hit upon a plan to prove it beyond doubt.

  Midway was ordered, via the secure submarine cable from Hawaii, to report in a low-level code that its desalination plant had broken down and that it was short of fresh water, information that would of course be of vital interest to a potential occupation force. On 22 May Melbourne intercepted a message from Japanese naval intelligence reporting that AF had radioed Hawaii to the effect that it only had enough water to last two weeks. On the same day, Colombo reported that it had intercepted a message indicating that the Japanese were planning to invade Midway. Rochefort had proved Washington wrong, but at what would later prove to be an extraordinarily high price.

  The change to the new cipher tables and codebook (JN25c) was only days away but the crucial message was already in the bag. On 20 May the Wahiawa station had intercepted a long message from Yamamoto to his commanders. It was his final operations order for the Midway and Aleutians attacks.

  The Wahiawa operators had a teleprinter link with Rochefort's codebreakers but most messages had to be sent by courier to Pearl Harbor, recalled Phillip Jacobsen.

  Our normal intercept positions had two high-frequency receivers elevated on a standard US Navy metal operating table with a typewriter well in the centre for the operator's chatter and watch log plus a side table for the typewriter used to copy messages. Each watch had a courier who made two trips per watch to Pearl Harbor with accumulated messages. The windows had blackout shutters on them and although the temperature at Wahiawa was rather temperate, the heat from the radios and lack of air-conditioning made the operating spaces rather warm and uncomfortable at times. We stood three rotating eight-hour watches starting off with the evening watch, day watch and finally the mid-watch. The latter watch was a killer until you got used to getting some sleep in the short time between watches. In addition, the junior operators had to do clean-up work, burn classified paper, etc. and relieve the day watch for chow in the morning of their evening watch. One of the permitted harassments was to give a ‘hot foot’ to any operator who went to sleep on the mid-watch by putting paper matches in between the sole and top of his shoe and lighting the matches. Somewhere along the line someone complained and the practice was discontinued.

  When the Yamamoto message arrived at Hypo, the codebreakers struggled to decode it. The attack was led by cryptanalyst Wesley ‘Ham’ Wright and Lieutenant-Commander Joe Finnegan, a Japanese interpreter who had seen service as a language officer at the US Embassy in Tokyo, Jasper Holmes recalled.

  Finnegan barricaded himself behind a desk with two flanking tables, all piled high with IBM printouts, newspapers, messages, crumpled cigarette packs, coffee cups, apple cores, and sundry material, through which he searched intently, usually with success, for some stray bit of corroborative evidence he remembered having seen days or weeks before. He paid little attention to the hours of the day, or the days of the week, and not infrequently he worked himself into such a state of exhaustion that his head dropped into the rat's nest on his desk and he reluctantly fell asleep.

  Although Finnegan and Wright led the attack, all the senior officers at Hypo were involved. They were able fairly quickly to decode the parts of the message in which Yamamoto described what he wanted his forces to do, Rochefort recalled.

  We could tell them what was going to happen, such things as where the Japanese aircraft carriers would be when they launched their planes, degrees and distance from Midway. Then, of course, the rest of the dispatch would be the strength of the attack and the composition of the attack forces and so on. The only two things we lacked were ‘where and when’ and these were especially enciphered within the basic operation order by the Japanese in a separate little cipher system. Now, the ‘where’ was easily solved. There was no problem there at all. This was the AF. The date, time and hours – ‘when’ – we were unable to get this.

  The separate internal cipher used to disguise the date and time of the attack had only been seen in use on three occasions, Rochefort said.

  This was the third time it had been used and we, of course, had the other two instances available to us which we kept studying. But unfortunately one of them was garbled, so we had nothing in which to prove or disprove our assumptions as to 4 June, or 8 June or 10 June or whether it was 19 July. We had no way of knowing. But by concerted attack by everybody concerned we were finally able to restore or to rebuild the little system just based on these three particular little incidents and, admittedly, it was rather shaky, but it was the best we could do under the circumstances.

  On the morning of 25 May Rochefort was summoned to report on Hypo's latest findings.

  Admiral Nimitz had sent for me to arrive at a certain time at his headquarters and I was late. The reason I was late was that we were still working on the final aspects of this dispatch and when I say that we were working on it, I mean that this would involve an agreement among the senior people at station Hypo and would probably have included Dyer [Hypo's chief cryptanalyst], Wright, Finnegan, the translator, Huckins, the radio intelligence officer and possibly Jack Williams, also radio intelligence officer, and this message that went out then was our consensus of what the dispatch meant and what the dispatch said and our reasons for it. Like everything else in station Hypo, any major decision of this nature would be the result of [what you might call a] staff conference. We never did call it things like that. We just said that we all agree with this. When I say ‘we’ I am always referring to the people who were most experienced and the most knowledgeable.

  The threatened Japanese codebook change occurred as the Hawaii codebreakers were finishing off their report on Yamamoto's final operations order. But Rochefort was still able to brief Nimitz in full on the Japanese plans, giving details of the diversionary attack on the Aleutians on 3 June to be followed the next day by an aerial bombardment of Midway and two days later by a full-scale invasion.

  This was an extraordinary achievement in its own right. For the vast bulk of the war, the only JN25 messages the codebreakers could break were those sent in stereotyped formats: sighting reports; convoy schedules; ships’ movements; routine air moves; antisubmarine attacks; and reports of air and sea bombardments. Yet so detailed was Rochefort's version of Yamamoto's operational orders that Nimitz's staff officers could not believe it was genuine. They argued that it must be part of a deliberate Japanese deception operation. But Nimitz was convinced and positioned his forces for a surprise attack.

  US Navy dive-bombers from the USS Enterprise and the USS Yorktown, the carrier the Japanese believed they had sunk in the Battle of the Coral Sea, were unleashed against the main Japanese carrier force at 10.25 on 4 June, scoring immediate direct hits on the three Japanese aircraft carriers the Akagi, the Kaga and the Soryu. All three were sunk within five minutes. The fourth, and sole remaining Japanese aircraft carrier involved in the battle, the Hiryu, launched its aircraft in two waves against the Yorktown, finally killing her off, but was itself left crippled and ablaze by dive-bombers from the Enterprise. The Japanese were eventually forced to scuttle it, leaving Yamamoto with no carriers or air support and forcing him to retire, conceding defeat.

  Within the space of five minutes, the tide had been turned in the Pacific. The unhindered Japanese progress had been brought to a juddering halt. The psychological effect on the hitherto victorious Japanese forces was immense and the initiative was handed to the Americans. It was Nimitz who gave the credit for his victory to
the codebreakers who had provided him with ‘a priceless advantage’, ensuring that even the newest radio intercept operator was congratulated for his part in the battle.

  ‘As soon as the attack was repulsed, the word was spread in our group of our successes in providing the early warning intelligence,’ Jacobsen said. ‘A commendation came down from Admiral Nimitz and we all were quite proud of our work and organization but realized that the details were top secret and we would not get any public recognition.’

  Unfortunately, the codebreakers’ contribution would not remain secret. Three days later, on 7 June, a dispatch by the Chicago Tribune reporter Stanley Johnston revealed that the ‘US Navy knew in advance all about Jap Fleet’. Johnston's report was syndicated to a number of newspapers, most notably the Washington Times Herald.

  Admiral Ernest J. King, the Chief of Naval Operations, apparently ‘in a white fury while his staff frantically tried to discover the source of the leak’, called an immediate press conference to deflect suggestions that the US Navy had known the detailed dispositions of the Japanese invasion force heading for Midway. But Johnston had published them in his article. The Japanese only had to read it to know that the information was right. It seemed the cat was well and truly out of the bag.

  The British, whose worst fears about American security had been realized, protested to Washington and were told that Nimitz's operations order for Midway had been intercepted by wireless operators from the Lexington who had been on their way back to America after the sinking of their carrier during the Battle of the Coral Sea. Johnston, who was himself a British subject, was on the same ship and had managed to see the Nimitz order. What they did not tell the British was that the reporter had been shown it by Commander Morton T. Seligman, the Lexington's Executive Officer.

  The damaging situation was not improved by the subsequent Grand Jury investigation, which, while it failed to lead to a prosecution, managed to generate yet further publicity about a leak that in the words of the veteran columnist Walter Winchell in the New York Daily Mirror on 7 July, a full month after the original report, ‘tossed security out of the window’.