Details like that may sound dull and ordinary, but they provide a lot of eye-opening intelligence about the makeup and location of enemy forces. The address was in its own code system, used only by the radiomen and distinct from that of the message itself. If Arlington Hall could break the address code, it would tell them who in the Japanese Army was where, but it also might provide clues that could help get into the messages themselves.
Wilma Berryman was assigned to the address problem in April 1942, and in June, as they were plunging into work in earnest, Ann Caracristi joined the group. As they labored and strategized, Ann and Wilma formed an immediate bond. It was an unlikely pairing: Wilma was a down-home, dyed-in-the-wool West Virginian, a shrewd, world-wise, tall, big-boned, genial southerner who liked to call people “honey” and to use adjectives like “stinkin’.” Her husband had died soon after they moved to Washington—she would marry three times—and her colleagues got her through her grief, gathering “round like family and it made life really worth living and going on,” as she later said. Ann was ten years younger, a northerner, new, well-bred, uncertain yet of her own abilities, inexperienced in the workplace, barely initiated into the tight fraternity of Arlington Hall cryptanalysts. But both had an appreciation for humor: Wilma was “a fun person,” and Ann liked working for her. The two women shared high spirits, humor, brains, imagination, and a relentless determination to prevail.
Both enjoyed the work. Ann would later say that she felt her time at Arlington Hall was “sport,” rather than labor. The women tried a number of approaches. At the suggestion of one military officer working with their unit, Ann was assigned to do something called “chaining” differences, which is a long, agonizing process that involves subtracting one code group from another with the hope that two code groups might have been enciphered with the same additive. Chaining differences was a task of the most routine and time-consuming order. It is code breaking the hard way, a brute-force method used when there are no other clues or ways to get a start.
They had caught a small break, though. When a Japanese plane crashed in Burma, the British captured some message templates—a template is a blank form with some code groups filled in to speed the process—and sent them to Arlington Hall, which had them in its possession by late January 1943. Wilma and Ann began having quiet conversations about how best to use this material. They agreed that chaining differences was “silly” and that it made more sense to use the templates and the bits of information they contained.
They had a couple of other tools as well. They had received a report from Australia with some names and ID numbers for Japanese Army units in the southwest Pacific. In addition, the U.S. Navy had sent over some cribs. At times, the Japanese Army was obliged to send messages over Navy radio circuits. In that event, a naval address code was appended. The naval address was in a simple code that had been solved, and it provided the basic pattern of Japanese military addresses; the unit, followed by two numbers, then an honorific such as butaicho, or “squad leader.”
The Arlington Hall address team had well-kept records, and Wilma Berryman remembered some Japanese Army messages with addresses that might be the same as these Navy ones. She found them and began fiddling around, writing the plain words from the Navy address codes below the Army ciphers. “I sort of remembered having seen something in that file and I went back to the file and found it,” was how she put it later. “I found what I thought looked like it ought to be that, the same thing. I had it on my desk and I just wasn’t positive.”
Suddenly she noticed one of the men who worked with their unit, Al Small, standing behind her. He stood there for a while, watching.
“Wilma, what are you doing?” Small asked. She showed him how she was lining up the Army messages against the Navy cribs. It seemed to be working. The captured message blanks gave her the underlying code group, and the Navy crib provided its likely meaning. If she saw, say, the enciphered group 8970, and knew from the captured blank that the basic code group was 1720, she could figure out that the additive was 7250. The Navy crib told her this was the syllable mo. She could hardly believe it. Was she imagining things? Al Small stood and looked at it for a long time. “You’ve got it,” he said finally. “That’s it! That’s it!”
“I’m afraid I’m forcing it,” replied Wilma, fearful that the juxtaposition was an illusion. “I’m pushing too hard. I want it to work.”
“No—that’s it!” Small repeated. “You’ve got it!”
Her solution was Arlington Hall’s first real break into any Japanese Army system. In early February 1943, a memo reported that “with the aid of the captured messages it has been possible to read the first encoded and enciphered addresses.” It would be followed by months of what amounted to the most laborious kind of coal-mining extraction. The problem with enciphered codes is that they do not yield to immediate solution like machine ciphers do. Even after a basic understanding is achieved, there remains a great deal of work. The code breakers had to build a bank of additives and figure out what each code group stood for. It had taken Agnes Driscoll and her team years to master the Imperial Japanese Navy code systems, but the Arlington Hall team did not have years. Deriving the meaning of each code group is like extracting ore, chunk by hard-won chunk. Ann Caracristi dove in, blissfully at home in an environment where there was very little oversight and “you assumed that you were going to have to figure your way out of most problems.”
Arlington Hall began producing weekly memos to keep the Pentagon and others apprised of their progress. On March 15, 1943, a memo reported that more address code values were emerging. Ann, Wilma, and their few colleagues had ascertained that 6972 meant i, 6163 meant aka, 4262 meant tuki, 3801 meant si, 0088 meant u, and 9009 meant dan. They were beginning to recover additives and to understand nuances such as the system’s unique sum check. Like the Japanese Navy, the Japanese Army cryptographers were fond of sum checks as a guard against garble, but they had devised a method different from the “divisible by three” sum check of the Japanese Navy. In this system, the code breakers discovered, the four-digit code groups were actually three digits, plus a fourth digit that served as a sum check. If a code group was 0987, for example, 098 was the actual code group, and 7 (using false math) was the sum check: 0 + 9 + 8 = 7.