THE CODEBREAKERS
Page 84
The information produced by the cryptanalysts and the direction-finding units with which they were affiliated enabled Japanese Army units to raid guerrilla posts, often with success. But this was defensive; it removed a thorn in the Japanese side, but it did nothing positive for their war effort. And even this declined steeply in the fall of 1943. “Although enemy wireless stations in the Philippines were disturbed by the strength of our punitive activities,” a September report said, “they avoided punishment by skillful concealment, and they are maintaining communication between themselves as well as with America and Australia.” In November, after some empty bragging about “a fatal blow to the guerrillas,” the Japanese admitted that “As always all the systems were very active and communication was carried on extensively within the Philippines and with outside stations.” And at these stations, notably Heindorf House, the Brisbane headquarters of MacArthur’s Allied Intelligence Bureau, where the cryptographic section, under Lieutenant C. B. Ferguson, deciphered the incoming traffic with quiet efficiency, the producers of intelligence for the strategic planners noted with satisfaction that as 1944 arrived the stream of information from the Philippines was assuming impressive proportions. Thus the only Japanese cryptanalyses that might be considered to any degree successful turned out in the end to be short-lived, limited, and utterly inconsequential to the greater course of the war.
Japan’s cryptography was as poor as her cryptanalysis, though it looked good on paper. Codes were numerous, with different users having their own; they were also changed regularly and varied from one geographical area to another. But the system did not work out well. Poor administration, distribution troubles, and lax security vitiated the theory.
Admittedly, the Japanese faced a difficult situation. Just getting codebooks to the thousands of ships and island garrisons scattered over the 20,000,000 square miles that they had conquered posed a staggering physical problem. Indeed, this problem had defeated them twice before Midway. Furthermore, the problem was magnified by the great numbers of codebooks that had to be distributed. Exactly how many codes Japan employed throughout the war may never be known. Probably the figure would reach several hundred, if every edition of the small codes carried by airplanes and auxiliary vessels were counted. An outline of Japanese naval cryptosystems—all of which were codes—will give an impression of the bristling panoply employed by a modern nation, as well as of Japanese distribution difficulties. Probably not all were in use together. In each case, the letter or the kana names the codebook.
Strategic and Administrative
KO—the flag officers’ system, a four-numeral code superenciphered by transposition; called AD by Americans; abandoned in 1942 or 1943 because of excessive garbles
D, later called RO—the fleet cryptographic system, the most widely used; called JN25 by Americans; a superenciphered two-part code
SHIN—a special logistics code, in practice usually replaced by D
Tactical
OTSU—for tactical communications of surface forces
BO—for local actions
F—for air; revised every two months
C—for air and miscellaneous
H—for air in China; a simple, easy-to-revise code
KI—for land combat in China; widely used during the China Incident, but generally neglected after Pearl Harbor in favor of OTSU
Joint Army-Navy codebook—suspended after the Army compromised it
A—combined fleet special code phrases
Attaché and intelligence
J—for attachés in Europe and the Americas
IC series—for intelligence agents
IC-A—in Great Britain, France, Italy, Spain, Portugal, Turkey, Soviet Union
IC-B—in China
IC-C—in Korea and Manchuria
IC-D—in the Americas
IC-E—in Burma
HEI—for intelligence in the China area
“The New Code Book”—for intelligence officers on the western coast of America
Overseas Secret Telegraph Code Book—in reserve for attachés
Extra-naval
HATO—for joint use of the three ministries of Foreign Affairs, Army, and Navy; superenciphered by additive
S—for merchant ships of more than 1,000 tons
[no name given]—for fishing boats
W—for reporting foreign vessels clearing Japanese ports; distributed to custom houses, harbormasters, and resident officers
Many special systems also served. The polyalphabetic date-time cipher that Wright solved before Midway and the CHI-HE map grid code were two examples. There were, in addition, a primary signal book for visual signals, lists of standard and special abbreviations, and call-sign tables for both strategic and tactical communications.
Once the Japanese introduced a special code for extra secrecy. The effects were disastrous. On June 15, 1944, when they set in motion their big A-Go operation to ambush the Allied fleet, the carrier Taiho, flagship of their 1st Mobile Fleet, carried this highly secret code for communications with Combined Fleet headquarters. Four days later, an American submarine torpedoed it, and a delayed gasoline explosion destroyed all communications facilities, including the special code. Urgent messages from headquarters piled up until other ships reported the loss. Among them was one reporting that the fleet was being trailed by an enemy task force, which attacked and sank another carrier. The A-Go operation ended in failure and the Marianas were lost.
The Japanese Army encoded its messages with four-figure codes that were superenciphered with the usual additives. For example, in the code used by the 6th Division around Bougainville late in 1943, 9019 stood for 23rd Infantry, 9015 for division headquarters, 9022 for 6th Cavalry, and so forth. The cipher section of the Kwantung Army in Manchuria compiled a 100-page geographical supplement to Army Codebook No. 3, listing codenumbers for places across the border in Siberia as preparation for a possible attack. The Army convened an annual conference in Tokyo for the chiefs of cipher sections of its various Army headquarters.
The wartime increase in communications naturally accelerated the Japanese Navy’s prewar schedule of changing a code every few years. The new standard called for changing the fleet system every six months to a year, its additives every month to six months, and the tactical code every month. In general this was met, except for the tactical OTSU, which was revised only once or twice a year. JN25, for example, went through about a dozen editions during the war—JN25b, c, d, and so on.
The numerous codes, and the frequent revisions of the most important ones, overburdened the Navy’s distribution system. To ease the strain, the Navy divided its whole theater of operations into eleven “code areas.” Distribution to one area would not depend upon distribution to another, and failure to get a code to an outpost would not compromise the entire system. The plan was for each area to have its own code, called HA-1, HA-2, and so on, according to the area. But the HA codes were not compiled in time, code TEN had to be used instead, and, as it worked out, all areas got the same code anyway with only the additive tables differing from one code area to another. High commands additionally held the RO code (JN25) for communications with other code areas.
This snafuing seems to have resulted from an administrative arrangement that was both unnecessary and unintelligent. Communications security in the Imperial Japanese Navy was basically the job of the 10th Section of its 4th Department (communications, whose Special Section was the Tokumu Han). The 10th Section planned cryptographic procedures, outlined the training of code clerks, compiled codebooks and additive tables, and supervised production. At first the printing of codebooks was done by the Printing Bureau of the Japanese Cabinet, later by the Navy Ministry’s printing office. Volume increases brought the presses of the Naval Torpedo and Communications Schools into service. Still later, most of the cryptographic printing was transferred to the Navy-controlled Bunjudo Printing Office in Yokohama, and in 1944 the 10th Section moved onto the firm’s premises. (This followed the
transferral in September, 1943, of routine encoding and decoding of headquarters messages from the 10th Section to the Tokyo Communications Unit, thereby relieving a staff of a line function. Headquarters now conformed to the Navy-wide arrangement, in which local communications units encoded and decoded for their commands.)
After being printed under the control of the 10th Section, the codebooks were physically taken to, and passed into the custody of, the Navy Library, which distributed all naval publications. This main library in turn sent the codebooks out to the district libraries, under the escort of librarians. District staff officers then carried the codebooks en masse to their area’s commands, where they were handed out to individual ships and units. Minor recipients picked up their codebooks themselves at the district library, or sometimes received them by registered mail. The Japanese sought to disseminate to combat units not only the new code, but a reserve code that was to supplant the new one. In addition, another back-up code was held at the district libraries. Canceled codebooks were at first returned to Japan, but later in the war they were simply burned and notification filed in writing.
The severance of the distributive function from the productive, and its bestowal on personnel less alive than the 10th Section to the imperatives of communications security, endangered Japanese cryptography. The librarians sometimes lost codebooks; sometimes they simply failed to observe the proper precautions. In 1943, a box of cryptographic publications en route from Kure to the Tsingtao Base Force in China was inadvertently opened. This was specifically charged to failure to provide a responsible escort aware of the publications’ importance. When a freight car carrying codebooks from the Yokosuka Naval District Library arrived at the Ominato Ship and Ordnance Department in 1943, its door lock was found to be missing. Investigators could not ascertain whether it had fallen off or been knocked off, but since the packing appeared undisturbed, they assumed that no compromise had occurred. In 1944, a whole load of codebooks disappeared during their rail trip from the Chinhae Communications Unit to one of its detachments. The unit was simply issued replacements.
Part of a page of the encoding section of a 1943 edition of the main Japanese Navy code
The Japanese stiffened this rather lax attitude when it came to losses in the combat zone. A submarine unloading cargo at Salamaua, New Guinea, crash-dived to avoid air attack and some codebooks were washed off deck. “Emergency measures”—whatever they were—were taken. They were also taken in 1943, when a transport plane jettisoned some cargo because of motor trouble between Truk and Rabaul. Included was a tightly packed box of codebooks which might have floated to nearby land. When the Allies assaulted Biak in May, 1944, a radio crew carrying codebooks to a safer location encountered an Allied patrol, and during the skirmish the codebooks were lost. The responsible officer did not report it for three weeks. Prompt investigation showed that only one or two lesser codes were missing, but since the loss took place while the Japanese were planning their A-Go operation, they changed codes wholesale. A new edition of JN25 was issued. Ironically, one of the main effects of the changes was to hamper Japanese communications on the eve of the operation.
But their remedial steps could not always prevent serious breaches of their communications security. One unavoidable case was that of the submarine I-1.
On the night of January 29, 1943, the cargo- and troop-carrying I-1 had the misfortune to surface near the New Zealand corvette Kiwi, commanded by Lieutenant Commander G. Bridson. He, his chief engineer, and his medical officer were famed throughout the South Seas for their mastodonic bulk and their practice of parading through Nouméa playing on a dented trombone, a jazz whistle, and a concertina. Upon spotting I-1, Bridson put his helm over and rang up full speed to ram. When the chief questioned this, he was told: “Shut up! There’s a weekend’s leave in Auckland dead ahead of us.” Ram they did, though I-1 was half again as big as Kiwi and had twice the firepower. With all guns from 20-millimeter to 4-inch blasting away at a range never exceeding 150 yards, Kiwi backed off and charged again—this time for “a week’s leave.” A third time she rammed, “for a fortnight”; this time she climbed right onto I-1’s deck. At 11:20 p.m. the submarine grounded on a reef at the northwest tip of Guadalcanal, ending the battle.
Among other things, she was carrying 200,000 codebooks. The crew buried some of these on the enemy-held shores, and when word of the action got back to Japanese headquarters, aerial bombardment and submarine torpedoing was ordered in an unsuccessful attempt to destroy the documents still aboard. But the Allies had already recovered the codebooks, which included both current and reserve codes. The Japanese ordered some new codebooks and additive tables to be used, but JN25 remained unchanged, and the documents as a whole were of great value. Bridson and his chief were awarded the Navy Cross.
So even the best of Japanese intentions proved to be too little and too late. Their communications security was as bad as their communications intelligence. Sometimes it seemed as if they didn’t care. The Navy attempted to find a water-soluble ink for codebooks to dissolve the printing when the books were jettisoned or the ship sunk, but when the Technical Research Laboratory reported that it could not find one that would fully obliterate the writing when immersed and yet would not run when splashed with rain, sea spray, or sweat, this worthy effort simply petered out. Certainly a lackadaisical attitude blighted their communications security. Instructions for a new Army code complained that “in certain situations, the use of the National Army Code was terrifying.” Though these instructions pointed out that sometimes code messages were sent to units which do not have the codebook for them, and urged that “The nature of the B Supplement [of Army B Code, No. 3] will be studied carefully and precautions taken so that such things will absolutely never happen in the future,” they also admitted with refreshing candor that “the time for the compilation being short, printing presses being busy and lacking the materials for one edition of the codebook, it may not be possible to fulfill the needs of each unit.” For communications security, they seemed to depend, not on training or on adequate cryptosystems, but on patriotic exhortation: “Even if there are any blunders, it is necessary that an endeavor be made to decipher the [garbled] message … even though it is just a trivial matter in the use of the general code of the National Army … so that flaws will not be exposed and earnest prayers will be offered for the glorious progress of one phase in the fulfillment of the sacred duties in the Great East Asia conflict.”
In part, the Japanese trusted too much to the reconditeness of their language for communications security, clinging to the myth that no foreigner could ever learn its multiple meanings well enough to understand it properly. In part, they would not envision the possibility that their codes might be read; the success of the Kiska withdrawal—in which they sneaked 5,000 troops off the Aleutian island in mid-1943, leaving only three yellow dogs to defend it against a powerful American force—“proved” to them that their secrets were still intact. Perhaps the cryptographers simply grew tired of printing the 2,000,000 codebooks needed to replace those jeopardized in the course of war. Perhaps their own failures with American ciphers convinced them that cryptanalysis was a practical impossibility. In any event, they hypnotized themselves into the delusion that their codes were never seriously compromised.
An incident of 1943 epitomizes Japanese incompetence in this whole field. It involved a future President of the United States, who, with his crew, formed the subject of a series of dispatches the Japanese apparently never solved.
These messages were transmitted by three brave Australian coastwatchers, part of a widespread network whose members observed enemy activity from the peaks and cliffs of enemy-held islands, collected tidbits from native allies, and radioed their information to Allied military commands. They frequently gave valuable early warning of Japanese bombing raids and ship movements, and they assisted in the rescue of downed Allied airmen.
In the early morning hours of August 2, 1943, coastwatcher Lieutenant Ar
thur Reginald Evans of the Royal Australian Naval Volunteer Reserve saw a pinpoint of flame on the dark waters of Blackett Strait from his jungle ridge on Kolombangara Island, one of the Solomons. He did not know then that the Japanese destroyer Amagiri had rammed and sliced in half an American patrol torpedo boat, PT 109, Lieutenant John F. Kennedy, United States Naval Reserve, commanding. But at 9:30 that morning he received a 20-group message enciphered in Playfair, the coastwatchers’ cipher system. He deciphered it with key ROYAL NEW ZEALAND NAVY and learned, PT boat one owe nine lost in action in Blackett Strait two miles SW Meresu Cove X Crew of twelve X Request any information X. He reported back to the coastwatcher near Munda, whose call-sign was PWD, that Object still floating between Meresu and Gizo, and at 1:12 p.m. he was told by the coastwatcher station KEN on Guadalcanal that there was a possibility of survivors landing either Vangavanga or islands.