With our country at war, the often-heard quip around my fraternity house ran something like, “This college deferment thing is going to last about as long as Max Schmeling did against Joe Louis.” As we found out, it wasn’t a joke. In 1942 most of us were told to pack up and get ready for some real soldiering. In no time we found ourselves dropped into a sea of khaki uniforms at Camp Maxey, Texas ready to start basic training.
Having endured the careful attention of our drill instructors and the 110-degree Texas heat, at the end of the course those of us who had survived were assembled to learn our fate. The Army Specialized Training Program (ASTP) would take charge of us next, and we would be taught the intricacies of becoming Army engineers. If we failed the course, we would be assigned to the infantry, which was devastating news for someone as mathematically challenged as I was.
My salvation came when the major who addressed us mentioned that one or two lucky individuals from the regiment would be sent to study Japanese. My heart leapt. On a lark I had taken a noncredit course in Japanese the previous summer.
With this meager résumé, I inveigled my way into the Army language program and wound up at the University of Michigan. While it was refreshing to be back on a college campus, it was not the carefree sort of life I had enjoyed before induction. Study was constant, and there was little or no time to enjoy the nightlife of nearby Detroit. For the most part, no English was allowed. You quickly learned to say, “Onaka ga suita” (“My honorable stomach has become empty”) or you didn’t eat.
As tough as it was, I had no desire to end up in a rifle squad, so I applied myself to my books. Motivation, not ability, put me near the top of the class at graduation. That placed me in the Signal Corps’ 2nd Signal Service Battalion (2d Sig), the Army’s code-breaking outfit, part of the Signal Security Agency (SSA).
Now officially an Army linguist, my next stop was Vint Hill Farm, Va., which was, as its name implied, a farm. The scattered silos and barns were meant to give casual observers the impression of an ordinary dairy farm. Those with more discerning eyes might have wondered what the 200-foot-high antennae on the hill behind the farm were for.
For the next six weeks, I spent my time delving into the mysteries of cryptography and advanced military Japanese. In addition to conducting the school, the base received radio retransmissions from monitoring stations in the Pacific, India and other locations around the world. Both male and female soldiers worked there, translating the dots and dashes of the Japanese Kana-code into numbers.
During the classes, our instructors often filled us in on some of the SSA’s spotted history. America’s prewar security personnel studied a phenomenon known as the “character of the fist” (individual telegraph operator’s unique style, making them identifiable). This allowed them to associate specific Japanese telegraph operators with their battleships and carriers. In late 1941, Admiral Isoruku Yamamoto began to suspect that his transmissions were being monitored, and he removed the regular telegraph operators from his ships. Instead, he instructed them to send dummy messages from the Japanese naval base at Yokohama, while he took the real fleet to sea to attack Pearl Harbor. That left our signal intelligence units assuming the fleet was still in home waters.
We were also told the saga of the Russian transport Uritsky. Loaded with Sherman tanks, it left Portland, Ore., on December 4, 1941, and headed across the North Pacific. Moscow, knowing the Japanese navy was maneuvering in the area, contacted Tokyo. “If you sink our ship,” Russian officials told the Japanese, “we’ll have to declare war on you. Leave it alone, and it won’t say a word about what it may see.”
In the wee hours of December 7, Uritsky found itself in the middle of the Pearl Harbor attack task force as it was preparing for its historic bombing run against America. The Japanese warships trained their guns and antennae on the Soviet ship, but true to Moscow’s word, the Russian captain steamed on without a peep.
The sharpest mathematical minds in the country became part of SSA. In addition, IBM set up a large array of state-ofthe-art computers (known then as punch card tabulators). Despite having an enormous supply of coded message traffic on which to work, SSA was unable to decrypt a single word through 1943.
We almost got a break in late 1943. Fighting in the Marshall Islands had all but wiped out a Japanese headquarters unit. In a desperate effort to hide the code books, the remaining enemy soldiers buried the material. Later, a group of American sailors came ashore for repairs after their ship rammed a Japanese submarine. While wandering around, one bored sailor stumbled on the cache of code material. Thinking it some sort of training manual, he smuggled the code books aboard his ship as a souvenir. Only months later, after the book had become obsolete, did officers find it in the sailor’s footlocker in Hawaii.
It is well known that the Navy and diplomatic code-breakers had highly rewarding successes. The capture of the Enigma machine allowed compromise of the German ciphers, and the U.S. Navy’s code-breaking turned the tide at Midway, but the Army’s effort at breaking Japanese army codes foundered.
In early 1944 I was transferred to Virginia’s Arlington Hall Station, which served as headquarters for the Army’s code-breaking efforts. Two satellite posts, Vint Hill Farms and Two Rock Ranch near Petaluma, Calif., served as gathering stations for enemy radio traffic.
Women’s Army Corps personnel generally handled operations at these outlying bases. The WACs studied Kana-code, the Japanese equivalent of Morse code. The women sat for eight-hour tours, transcribing around the clock what must have seemed like meaningless dots and dashes. These transcriptions were then sent to us to figure out.
By the time my unit reached Arlington Hall, personnel there were reading the current 1944 Japanese army code with considerable clarity—not the result of mathematical genius or IBM technology, but through a piece of luck in the field.
At Sio, New Guinea, the American 126th Regimental Combat Team attacked a Japanese division from the east, while the Australian 9th Division closed the pincers from the west. Outnumbered, the enemy scattered into the jungle.
The enemy division commander, knowing the Americans would soon overrun his headquarters, assigned a lieutenant and a sergeant to burn the new 1944 code books. Just as the two finished pouring oil on an ammo box containing the material and setting it afire, they saw Allied soldiers rushing toward them.
In a panic, the two Japanese soldiers pushed the material into a foxhole and covered it with dirt, then ran after their departing comrades. An Australian patrol discovered the code material, which quickly found its way to Arlington Hall.
It might not be a big advantage to possess the enemy’s code book if he suspects you have it, however. So, code books in hand, the staff at Arlington Hall immediately began looking for traffic from that retreating Japanese unit. Indeed, they found a message stating that the lieutenant attested to the fact he “watched the code book burn to ashes and personally scattered those ashes in the sea.” How we loved that lying Japanese lieutenant! Next, we found out that his noncommissioned accomplice had been killed in action. We were home free.
Careful study of our new treasure revealed a truly ingenious encryption system. The code book consisted of 10,000 four-digit code groups. A code group might stand for a single letter, a word or a phrase. It might be a dummy, with no meaning at all, used to change the appearance of a message. Entirely new code books were published in January of each year.
Raw code groups would never be transmitted. Rather, the Japanese applied additives or ciphers to the groups. Additive pages, consisting of 20 rows and seven columns of four-digit random numbers, were added to the raw groups of the code book. A message might commence, “Start at page 21, row 3, column 7.” This told the recipient that the sender had started at that location in the additive book. By starting at different points in the book, the same message could be sent 10,000 times a month and never look the same. With dummy groups inserted, it wouldn’t even have the same length.
Our biggest challenge was to keep current with the additive pages, which changed monthly. The solution lay in our own progress in fighting our way across the Pacific. Of necessity, during the island-hopping campaign several remote dots of land had been bypassed, their Japanese garrisons left to fend for themselves. Imperial headquarters could no longer hand-deliver the updates, so the new information had to be sent using the older additives, which we could already read. Comparing the two sets of data allowed us to translate the new codes within days.
Our IBM machines were soon programmed to spit out fanfold printouts of each message. Many of the groups in the Japanese code book consisted of kanji— Chinese picture characters—instead of alphabetical letters. The machines were even able to convey this concept to us translators. Although we encountered a few glitches and garbles, we became adept at guessing what the Japanese code clerk intended.
By mid-1944, our machines were running full tilt, and 2d Sig was able to intercept, decrypt and translate messages and get this information on its way to commanders in the field before the clerk who sent it had a chance to adjust his glasses.
My specialty became strength reports. At the end of each month, every company-size unit in the Japanese army had to report its personnel strength and the status of its equipment. Often a plaintive P.S. followed: “We desperately need replacements.” Combining this information with analysis of other message traffic we received allowed us to tell our field commanders what unit they might be facing and its strength.
Our progress remained good until January 1945, when the basic Japanese code book changed again. We knew the enemy used many of the same words and phrases in each edition of the book; only the four-digit code groups changed. What Japanese code-developers tried to do was assign the groups in a random order, which is not as easy as it sounds. Even modern computers may have difficulty creating a truly random arrangement.
By observing the order in which the various groups were assigned in the 1944 book, our experts were able to detect a pattern that they then applied to the new edition. We knew for example that wa, designating the subject of a sentence, and no, making a noun possessive, were the words used most in the Japanese language. We therefore searched for the most frequent code groups and tried to link them to these two words.
Other clues were found in our enemy’s word play. Japanese commanders loved to make up new words, hoping history would credit them with the neologism. However, for words not in the code book, they had to resort to “Chinese Tel,” an open-source Asian code available to us.
Clues like this were a godsend to our translators, IBM programmers and mathematicians who were working tirelessly to get a handle on the 1945 code. Despite their best efforts, results came slowly. Until they did, fresh material that we could translate dried up. To keep us occupied, we spent our time reviewing unread, low-priority messages from the previous year.
That did not seem like such a good idea to someone at the Hall. While I had been in Virginia, I received a direct commission, as had several other translators. Someone must have asked whether our efforts wouldn’t be better spent in the field looking for code material, and we soon had orders to pack our duffle bags and prepare to ship out for warmer climes.
A cross-country trip ended at Hamilton Field, Calif., and we were soon ensconced in a Douglas C-47 winging our way to Saipan. Upon arrival, we were assigned to Quonset huts. Nearby, Seabees worked to improve the runways now being used by the massive Boeing B-29 bombers to blast Japan into submission. Various other Navy units were also busy staging for the invasion of Iwo Jima, which began shortly after we arrived.
We assumed that it was this invasion that had been the rationale for our trip, but we were not sure. To kill time, I made a couple of forays into the jungle in a jeep rigged with a loudspeaker, announcing to any lurking Japanese soldiers that their war was over and that they would be treated fairly if they surrendered. Not surprisingly, silence greeted my nervous pronouncements.
My language skills were also put to use in a compound called Camp Susupe, where I questioned a motley assortment of Japanese civilians, Korean laborers, Chamorro natives and a small contingent of Japanese POWs.
Having accomplished little more than getting a suntan, we soon were ordered back to Arlington Hall. While we had been away, our compatriots had made progress deciphering the new code books, and we would again have work to do.
By the time I reached my old stomping grounds, 2d Sig had deciphered about 60 percent of the 1945 code book. I now spent much of my time with cryptographers playing charades. I remember one dialog. “This decrypt reads, ‘Moving troops to rin_ _n.’ What could go with rin?” the cryptographer asked.
I suggested trying “lin” for “rin.” The Japanese can’t write “L.” We eventually came up with Lingayen in the Philippines, which gave us a couple of better-known code groups. It was not until June that an actual copy of the code book fell into our hands.
After some of the bloodiest fighting of the war, Lt. Gen. Mitsuru Ushijima pulled out of his Okinawa headquarters at Shuri Castle. The code room was nestled in the catacombs beneath the castle. The withdrawing army set the secret material afire before retreating. The concussion from our battleships’ shelling had put out the fires, and Marines moving into the smoldering ruins came across a Japanese private with a partially burned copy of the book. At that point, however, most of the Japanese forces were consolidated on the Home Islands, so there was no longer a need to broadcast information across vast expanses of the Pacific.
The greatest battle of the war was soon to begin—Operation Olympic, the invasion of Japan itself. In planning for the landings at Kyushu, the American generals expected the landing force to encounter three enemy divisions, which they felt could be overcome.
Working away at Arlington Hall, however, we translated intercepts that clearly indicated Japan was ordering in all its troops from China and Indochina, which would have brought the total number of units along the intended invasion beaches up to 13.
While those higher up worked at a solution to this problem, we continued to translate any decryption that fell into our hands. In early August, a fellow translator brought a decrypt to me and asked, “What the hell is a genshi bakudan?” Despite the years I had been working at this, I had no idea. All I could stammer was: “I guess you’d have to translate it as…atomized bomb. It’s probably just a garble.” I was completely unaware that I was likely one of the first Americans to learn of the atomic attack at Hiroshima. It was not until later that, along with millions of other Americans, I shared in the news that the war would soon be over.
Originally published in the October 2006 issue of World War II. To subscribe, click here.
