This was the dynamic at work in the careerist Navy facility: enmities, lies, thefts, insecurities, power plays. In so many ways, the young women who joined it had no idea what they were getting into. But they were delighted to be there. The number of workers was so small—and the task so large—that the women from the Seven Sisters instantly took on real responsibility. As the women dove in, cryptanalysis was being done by a group called OP-20-GY, which soon was divided into GY-P, for Pacific, and GY-A, for Atlantic, and then subdivided. Room assignments were made and remade as GY overflowed into three wings of the old Navy building, an arrangement that was “neither convenient nor economical,” as one memo put it. The size of a shift might be determined by how many chairs were available. There was nowhere to stow workbooks, no secure telephones between rooms; papers had to be shoved aside to make a little training area; and with so many people stomping in and out of the main naval headquarters, secrecy became a constant headache. The group working on JN-25 soon occupied two wings. Vi Moore found herself doing additive recovery in Room 1515, while Anne Barus, Louise Wilde, Ann White, Fran Steen, and others were doing the same thing in Room 3636, an awkward distance away.
Most of the women were assigned to JN-25, at a time when the complex supercipher had become more complex. The Americans weren’t the only ones galvanized by Midway. In the wake of Japan’s shocking naval defeat, the Japanese decided to divide the fleet code into five “channels,” so that certain regions or kinds of communications—Singapore, the Philippines, operational, administrative—had their own code and additive books. The volume of messages grew and grew. The naval code breakers received 18,000 JN-25 intercepts per month in the first six months of 1942, and more than double that, 37,000, in the second half of the year. By the fourth quarter of 1943 they would be getting 126,000 messages per month.
The women rose to the challenge. Anne Barus, the Smith history major, was assigned to recover additives, a task her collegiate training course had not covered, and one that involved ceaseless mental math performed day after day, week after week, for more than three years. The women in her unit were given big sheets of paper, about a yard long and two feet wide. Each sheet was filled with rows of five-digit numbers—14579 35981 56921 78632 90214, say—that also lined up in vertical columns. The messages were placed so that each code group was directly above a code group enciphered by the same additive, a symmetry determined by women trained to evaluate a key group at the beginning. It was Anne’s job to figure out what the additives were, so that the Japanese additive book might be reconstructed.
To do this, Anne had to master the same “false math” the Japanese used—only in reverse. She and her colleagues had to start with the enciphered numbers and work backward to find the underlying code group. And they had to do it fast. Looking down a vertical column, Anne was tasked with finding the lone additive used to encipher all the code groups in that column. Aside from her own wits, she had one thing to help her: a quirky feature designed to cope with radio garble. Garbling was a huge problem in radio transmissions, so the Japanese developed clever “garble checks” so the person at the receiving end could do a bit of math to be sure the message had transmitted correctly. It was a sensible enough tactic, except that it also was an insecure one: Many of these checks—and the ghostly patterns they left—helped with breaking the messages.
One such JN-25 garble check was the rule that a valid code group was always divisible by three. Looking at her work sheet, Anne would conjecture a possible additive, then go down the vertical column, quickly, in her head, stripping the hypothesized additive out of each group she saw, and looking to see if the remainder was divisible by three. If she conjectured an additive, stripped it, looked at the row of code groups, and saw that all were divisible by three—17436, say, or 23823—then she knew she had gotten down to valid code groups and had therefore conjectured a viable additive. It took all this work to get one single additive, which would be recorded in the book they were building. Whenever the Japanese changed the JN-25 cipher books, the unit would start all over again. It was truly like sweeping the sand from a beach.
Anne, like the other women in her room, learned to look for common enemy mistakes. More than seventy years later she would remember them clearly. In such a massive fleet system, it would sometimes happen that an oblivious radioman would send a message in the clear—plain Japanese—that others were sending in code. The women could use the plain Japanese as a crib. The Japanese, like the Germans, also tended to send out fleet messages that were formulaic and patterned. Japanese merchant ship captains often sent a shoo-goichi message, stating what their exact position would be at noon. Anne learned the code groups for “noon position,” and she learned where the phrase was likely to appear. When she saw an enciphered group in that place, she could subtract the code group and obtain the additive.
Many things the Japanese did with the intention of making the code harder to break made it easier. Sometimes, enemy cryptographers liked to begin a message in the middle. When they did this, they would include a code group that stood for “begin message here” to show where the message started. The women learned the code groups for “begin message here”—there were several—and gained another point of entry. There also was something called “tailing.” Japanese encoders were told not to end a message at a certain point in an additive book and then begin the next message at the next additive number—they were supposed to choose another random starting point—but often, they were lazy or harried, and they did. The women mastered these ins and outs and quirks. Whenever they saw a mistake, they pounced. The shoo-goichi messages did more than help recover additives; the noon position would be swiftly radioed to an American submarine captain, who would be waiting for the Japanese ship when it appeared on the horizon.