THE CODEBREAKERS

by DAVID KAHN

  Arthur Evans’ decipherment of the message of 9:30 a.m., August 2, 1943, that reported the sinking of John F. Kennedy’s PT 109

  This was just what Kennedy and his crew had done. They had swum to Plum Pudding Island, one of a group that hangs from the southeastern tip of Gizo Island. This group was behind enemy lines, and Gizo itself, only three or four miles away, was garrisoned by Japanese troops. Though messages about the missing crew continued to stream for the rest of the week between PWD, KEN, and GSE, as Evans called his station (after his wife, Gertrude Slaney Evans), the Japanese made no attempt to capture them. Yet the importance of the crew should have been obvious to the Japanese from the many messages concerning it and from the search mission flown by P-40s, and a capture could not have caused them too much trouble, since on one occasion a Japanese barge chugged right past the island hideout of Kennedy and his crew. Even if they had been intercepting and reading the cryptograms, however, the Japanese may not have wanted to waste time looking for the Americans, since none of the messages specified their location.

  This excuse vanished at 9:20 a.m. Saturday morning, August 7. Two natives had found the sailors, who had moved to Gross Island, and had reported the find to Evans. He wrote a brief message: Eleven survivors PT boat on Gross Is X Have sent food and letter advising senior come here without delay X Warn aviation of canoes crossing Ferguson RE. He drew up a square based on the current key of PHYSICAL EXAMINATION,

  P H Y S I

  C A L E X

  M N T O B

  D F G K Q

  R U V W Z

  and enciphered the message, departing from traditional Playfair only by leaving doubled letters unenciphered, as the S’s in Gross and crossing: XELWA OHWUW YZMWI HOMNE OBTFW MSSPI AJLUO EAONG OOFCM FEXTT CWCFZ YIPTF EOBHM WEMOC SAWCZ SNYNW MGXEL HEZCU FNZYL NSBTB DANFK OPEWM SSHBK GCWFV EKMUE. A message of this length would alone suffice for the solution of a Playfair, and there were four others in the same key, including one of 335 letters, beginning XYAWO GAOOA GPEMO HPQCW IPNLG RPIXL TXLOA NNYCS YXBOY MNBIN YOBTY QYNAI …, for Lieut Kennedy considers it advisable that he pilot PT boat tonight X ….

  These five messages detailed the rescue arrangements, which offered the Japanese a chance to get not only the shipwrecked crew but the force coming out to save it. All of them could have been solved within an hour by even a moderately experienced cryptanalyst. Yet at 10 p.m. the operation went off without the least hint of enemy interference. It seems likely that had the Japanese solved these elementary enciphered messages, they would have taken some action against the rescuers or the rescued or both. They did nothing. If their communications intelligence had been better, how might contemporary history have been changed!

  Their failure sharpens the contrast with Allied successes. For Allied cryptanalysts—which in the Pacific meant mostly Americans—galloped like Tartars through the phalanxed ranks of a legion of Japanese cryptosystems. They ravaged and plundered with a prodigality that did not trifle with petty matters. One system, when solved, proved to be used by direction-finding teams; though this might have afforded some indirect clues to Japanese attacks, it was cast aside for richer treasure. Commander Dyer estimated that American cryptanalysts demolished 75 Japanese naval codes during the war.

  Among them was the four-digit code used by the marus, or Japanese merchant vessels—the s code. Presumably this was attacked after the more important combat codes had been resolved. From about 1943, it yielded information of the greatest value: the routes, timetables, and destinations of Japanese convoys. Japan’s conquests consisted almost entirely of islands which could be supplied and reinforced only by sea, and Nippon itself was an island empire. American submarines therefore undertook in the Pacific what U-boats were attempting in the Atlantic, and, as with the U-boats, cryptanalysis helped them achieve their greatest successes.

  A direct line led from FRUPAC to the office of Captain R. G. Voge, operations officer of the Commander, Submarines Pacific Fleet. The Japanese convoys radioed the positions where they estimated they would be as of noon on the next few days. This was to inform their own forces of their locations, but FRUPAC solved the messages, and Jasper Holmes, an ex-submariner himself, relayed them to Voge, who broadcast them to the American submarines. This fattened their kill. Vice Admiral Charles A. Lockwood, Jr., who was COMSUBPAC during most of the war, estimated that cryptanalytic information stepped up American sinkings by about one third on the trade routes to the Philippines and the Marianas. Eventually the submarine commanders received it so regularly that they complained if a convoy reached its noon position half an hour late!

  The pigboats accounted for nearly two thirds of Japanese merchant tonnage sunk during the war. Their torpedoing of 110 tankers from the East Indies resulted in oil shortages in the homeland that prevented the training of badly needed pilots and forced a split-up of Japan’s Navy, with serious tactical results. Starvation at home caused Japan to make surrender overtures even before the islands were invaded, before the atom bombs exploded. After the war, Tojo said that the destruction of the merchant marine was one of the three factors that defeated Japan, the others being leapfrog strategy and fast carrier operations. This is why Dyer, looking back, regarded FRUPAC’s solution of the maru code as one of its primary contributions to victory.

  American cryptanalysts scored some long-range combat triumphs as well. Shortly after MacArthur invaded Leyte, they discovered from their reading of coded enemy messages that 40,000 soldiers were on their way to reinforce Japanese troops in the Philippines. American air and sea power met and destroyed this force, and not a man reached Leyte. During the Okinawa campaign, the sharp ears of the cryptanalysts overheard the orders that directed the superbattleship Yamato, a 72,000-ton monster with 18-inch guns that could hurl a projectile 22 miles, to sortie in a last-ditch defense effort. They passed this news to the American commanders on the spot. Thus alerted, the commanders prepared to attack her, and after a picket submarine reported her position, flung wave after wave of carrier-based planes at her. They struck her at 12:32 p.m. April 7, 1945, and after less than two hours of repeated bomb-hits and torpedoings, the world’s largest battleship slid to the bottom, rumbling and exploding, and taking with her 2,488 officers and men of her complement of 2,767.

  FRUPAC also engendered what is probably the most spectacular single incident ever to result from cryptanalysis.

  In the spring of 1943, Admiral Isoroku Yamamoto came down to Rabaul to take personal charge of the deteriorating situation in the Solomon Islands. Japan had just been pushed off Guadalcanal and her supply lines were being snarled by Allied air attacks. Yamamoto welded together the biggest Japanese air armada of the war and sent it against the Allies, achieving some tactical successes. In preparation for further aerial offensives, the stocky, black-browed seaman decided to make a one-day morale and inspection tour of bases in the upper Solomons. Those bases would have to be alerted, together with several other units, so that they could make the many preparations needed for an inspection by the Commander in Chief Combined Fleet. At 5:55 p.m. on April 13, 1943, the commander of the 8th Fleet broadcast Yamamoto’s itinerary of five days hence to the 1st Base Force, the 26th Air Flotilla, all commanding officers of the 11th Air Flotilla, the commander of the 958th Air Unit, and the chief of the Ballale Defense Unit. The great variety of addressees, plus the need to safeguard the person of the head of the Navy, makes it almost certain that the Japanese communicator selected the current edition of JN25—the most widely distributed high-security code—in which to armor this information.

  Unfortunately for the Japanese, this armor plating had been dissolved in the acid of Allied cryptanalysis. As with the pre-Midway solution, the scattered codebreaking units had exchanged their results—possibly augmenting them this time with documents salvaged a few weeks previously from the grounded submarine I-1. Though the additive had been changed only two weeks before, on April 1, large portions of it had been recovered. At FRUPAC, these results had been punched onto cards for the I.
B.M. machines. FRUPAC’s monitors had intercepted the message that the 8th Fleet commander had spread on the airwaves, and when this was fed to the robot cryptanalyst in a form palatable to it, it swallowed it, digested it to the accompaniment of horrendous clickings and rattlings, and disgorged the Japanese plaintext.

  Because of the many addressees, the “scanners,” or traffic analysts, had probably flagged the message as one of more than ordinary importance. Hence the plaintext went to a translator of more than ordinary competence, a 38-year-old Marine Corps lieutenant colonel, Alva Bryan Lasswell. He had studied Japanese as a language officer in Tokyo from 1935 to 1938 and had helped with communications-intelligence activities in Hawaii since May, 1941. The message was essentially complete, but he helped fill in some holes, while Dyer recovered some additives and Wright determined the meaning of internal geographical codegroups: RR for Rabaul’; RXZ for Ballale, a small island in the Solomons group, just south of Bougainville; RXE for Shortland, another of the Solomons, also south of Bougainville and west of Ballale; and RXP for Buin, a base on the southern tip of Bougainville. When this work was completed, Lasswell translated the message.

  Routes followed by American airplanes (solid line) and by Japanese (dotted line) on mission to shoot down Admiral Isoroku Yamamoto

  The Commander in Chief Combined Fleet will inspect Ballale, Shortland, and Buin in accordance with the following:

  1. 0600 depart Rabaul on board medium attack plane (escorted by 6 fighters); 0800 arrive Ballale. Immediately depart for Shortland on board subchaser (1st Base Force to ready one boat), arriving at 0840. Depart Shortland 0945 aboard said subchaser, arriving Ballale at 1030. (For transportation purposes, have ready an assault boat at Shortland and a motor launch at Ballale.) 1100 depart Ballale on board medium attack plane, arriving Buin at 1110. Lunch at 1st Base Force Headquarters (Senior Staff Officer of Air Flotilla 26 to be present). 1400 depart Buin aboard medium attack plane; arrive Rabaul at 1540.

  2. Inspection Procedures: After being briefed on present status, the troops (patients at 1st Base Force Hospital) will be visited. However, there will be no interruptions in the routine duties of the day.

  3. Uniforms will be the uniform for the day except that the commanding officers of the various units will be in combat attire with decorations.

  4. In the event of inclement weather, the tour will be postponed one day.

  Yamamoto was known to be almost compulsively punctual. He adhered to his schedules virtually to the split second. And Lasswell was now reading almost a minute-by-minute listing of his activities on a day during which the admiral would come closer to the combat zone than he had probably ever done before! The cryptanalyzed intercept amounted to a death warrant for the highest enemy commander.

  The question was: Should it be executed? It was not an easy one to answer. Nimitz wrestled with the pros and cons. If Yamamoto were shot down, would a better man be appointed to succeed him? Commander Layton, the fleet intelligence officer, set out the arguments, most of which Nimitz well knew.

  Yamamoto, 59, was the dominant figure of the Japanese Navy. A prophet of air power, aggressive and determined, he devised bold, imaginative plans and executed them under strong leadership. He was the Shogi (Japanese chess) champion of his navy, and in the 1920s had enjoyed matching wits with Americans at poker, which he played very well indeed. He had lost two fingers of his right hand in battle, and he manipulated the cards with the remaining three in so wizardly a manner that he distracted his opponents. American intelligence rated him as “Exceptionally able, forceful, and quick-thinking.” His men idolized him. “If, at the start of the Pacific War,” wrote Commander Fuchida, leader of the Pearl Harbor attack, “a poll had been taken among Japanese naval officers to determine their choice of the man to lead them as Commander in Chief Combined Fleet, there is little doubt that Admiral Yamamoto would have been selected by an overwhelming majority.”

  Layton summed up with the observation that Yamamoto was preeminent in all categories, that any successor would be personally and professionally inferior, and, finally, that the death of the Commander in Chief would demoralize the Japanese, who venerate their captains much more than Occidentals do. Nimitz concurred. He realized that the shock of such a leader’s death, combined with the elimination of the finest strategist of the enemy war machine, would equal a major American battle victory. He was furthermore probably influenced by the general American hatred of Yamamoto. Naval officers knew that he had conceived the treacherous strike at Pearl Harbor that had slaughtered their shipmates and wrecked their ships. He had, they thought, arrogantly boasted that he would dictate peace in the White House.* This was why Admiral William F. (Bull) Halsey made him “No. 3 on my private list of public enemies, closely trailing Hirohito and Tojo.”

  By chance, the Ballale-Shortland-Buin area was in Halsey’s theater of operations. Consequently Nimitz sent him a top-secret command-level communication referring to the Yamamoto itinerary and authorizing him to shoot down the Japanese planes if his forces had the capability of doing so. Halsey was in Australia; his deputy, Vice Admiral Theodore S. Wilkinson, reported that he could do it, but invited Nimitz’ attention to the danger of making the Japanese suspicious that the Allies were reading their codes. If they changed them, might not this deprive the Allies of possibly even more valuable intelligence in the future?

  Nimitz felt that this bird in the hand was well worth any two in the bush. Nevertheless, he sought to minimize the danger by following Layton’s suggestion of a cover story. This was to the effect that Australian coastwatchers had radioed in the Yamamoto flight information, probably getting it from friendly natives around Rabaul. The coastwatchers enjoyed a superexcellent reputation among airmen and so the story would ring true. If it got back to the Japanese, they might never even think about codes. Even if they did realize that the Allies were reading their codes, either by capture or by crypt-analysis, they could probably do no more than issue a new edition of JN25 and perhaps tighten cryptographic security. But this had happened before, and Allied cryptanalysts had broken the new codes. The most realistic assessment predicted that the Yamamoto mission might temporarily dim Allied communications intelligence while cryptanalysts sought entry into the new code.

  Such a loss of information is never good, but it would be less unfortunate now, when the Allies were resting and consolidating their positions, than during a major operation. No such advance was planned for two and a half months. Hence if the Japanese changed their code immediately after Yama-moto’s death, the cryptanalysts would have ten weeks of relative quiet to break back in. In his reply to Wilkinson, therefore, Nimitz ordered him to brief all personnel on the cover story, iterated his authorization, and added a personal “good luck and good hunting” to the message.

  The death warrant was now signed, sealed, and delivered.

  On the afternoon of April 17, Major John W. Mitchell and Captain Thomas G. Lanphier, Jr., both of the Army Air Corps, walked into a dank and musty Marine dugout on Henderson Field, Guadalcanal. An operations officer handed them a cablegram on blue tissue—the kind used for top-secret dispatches. It detailed Yamamoto’s itinerary, including times of arrival and departure from each place. The airmen vetoed a suggestion to strafe him while crossing from Ballale to Shortland in the subchaser because of the difficulty of identifying the right craft. Instead they decided to intercept him in the air.

  Their plan depended upon Yamamoto’s punctuality and required careful timing of its own: Ballale was near the limit of range of the twin-engined P-38 Lightnings that the pilots flew, so there would be little fuel for waiting. Though the Japanese message specified arrival at Ballale at 8 a.m. after a two-hour flight from Rabaul, calculations showed that the two-motored Mitsubishi (Betty) attack bombers would reach Ballale in an hour and 45 minutes; this was partially confirmed by the estimated hour-and-40-minute return time from the slightly closer Buin. This meant that Yamamoto would arrive at Ballale about 7:45 a.m. Though he would be escorted by six
fighters, Mitchell and Lanphier decided to attack him about 35 miles up the Bougainville coast to avoid the planes that buzzed around Kahili airstrip not far from Buin. This pushed the time of interception back ten minutes to 7:35 a.m.—or 9:35 a.m. American time.

  Next morning, 18 P-38s of the 12th, 339th, and 70th Fighter Squadrons lifted off the Henderson runway at 7:25 (American time). Thirty-five minutes later and 700-odd miles away, Yamamoto’s flight took off right on schedule. Radios silent, the Americans flew a semicircle of 435 miles around Munda, Rendova, and Shortland at wave-top height to avoid radar detection. Mitchell navigated by compass and airspeed indicator, and two hours and nine minutes after take-off was skimming the waves toward the Bougainville coast. He had timed the flight to the split second, and suddenly, as if the entire affair had been rehearsed to perfection, the black specks of Yamamoto’s squadron appeared five miles away.

  “Bogey. Ten o’clock high,” called out Lieutenant Doug Canning, breaking radio silence. Mitchell led 14 fighters up to 20,000 feet as cover and to engage the fighters. Lanphier dropped his belly tanks, and, with his wing man, Lieutenant Rex T. Barber, climbed to within two miles of Yamamoto’s right and a mile in front of him before his escorting Zeros saw them and turned to attack. Lanphier disintegrated one of them, then kicked his ship on its back and looked down for the lead bomber. He spotted it dodging away at tree-top level. As he spun toward it, two Zeros dived at him. But, he said, “I remember suddenly getting very stubborn about making the most of the one good shot I had coming up. I fired a long steady burst across the bomber’s course of flight, from approximately right angles. The bomber’s right engine, then its right wing, burst into flame…. Just as I moved into range of Yamamoto’s bomber and its [tail] cannon, the bomber’s wing tore off. The bomber plunged into the jungle.” The Zeros screamed helplessly overhead. Barber, meanwhile, exploded the other Mitsubishi. Lanphier shook his pursuers in a speedy climb to 20,000 feet, and he and all other members of the mission except one returned safely to Henderson.