When Dot was hired, Department K—the unit that decoded 2468 messages—was doubling in size, from 100 workers in July 1943 to 217 in 1944, and it was steadily growing more expert and efficient. “The history of the department during the past year is one of increasing production with a continuous decrease in the time necessary for solution,” noted a 1944 memo. “Where ten days was considered to be good time for the solution of an overlap at the beginning of the year, some are now recovered in three days.” Individual messages could be solved much more quickly.
The 2468 code breakers operated on a twenty-four-hour basis. There were three shifts—day, swing, and graveyard—and women rotated between them. One report noted that the Japanese Army unit “probably handles the most enemy traffic for deciphering of any agency in the world.” The unit had been fashioned along the principles of an American assembly line: routine, simplicity, flexibility. During training, instructors tried to root out people who were prone to hysteria or nervous breakdowns or who did not seem easily adjustable. “This is a business organization,” said one memo, “and not a country club.” Innovations were always being sought. A pneumatic tube was set up between Building A, where a machine unit was located, and Building B, where the code breaking happened, which sped up the delivery of traffic and eliminated a lot of courier work.
The assembly line also was designed by women. Alice Goodson set up a bank listing address preambles in alphabetical order, as well as stereotypes and other devices to help readers like Dot get a start. Helen O’Rourke designed the overlap unit. There was a ten-person unit that coordinated between the women working on 2468 and those assigned to other code systems. A color-coded system—Lavender, Orchid, Lilac—was devised to keep track of the time periods when the 2468 code or cipher books changed.
The women in Department K—Dot’s unit—were a “very fine group,” according to one Lieutenant Bradley, an Army officer who rotated into the unit late in 1943. He was there when a break into a new square led to “great glee over the entire place.” He watched as dozens of new recruits joined and became adept. At first, readers needed overlaps that were ten or fifteen messages deep to recover code groups, but they soon needed fewer and fewer. “The pattern was the main thing,” said Lieutenant Bradley. “There was no information service or cribbing section at that time. Each reader had to depend on what he could remember.”
The women knew they were performing well. “The great value of the intelligence derived from the [2468] messages is a constant incentive to the department,” a report noted, praising Dot’s Department K, which was producing all kinds of shipping intelligence, foretelling what units were about to receive oil or gasoline; what ships were in a given harbor; what convoys were getting ready to sail and where they were headed.
That report also cited one consequence of all this foreknowledge. On May 3, 1944, Department K read a series of messages indicating the noon positions through May 8 of fifteen ships headed for New Guinea. Shortly thereafter, the U.S. Navy sank four of them. Another memo pointed out that the New York Times had reported on one of many successes. In September 1943, a Times piece had noted a Pacific engagement in which “our strongly escorted medium bombers attacked an enemy convoy of five cargo ships and two destroyers which arrived during the night with reinforcements and supplies for the enemy garrison. Coming in at masthead height, our bombers scored direct hits with 1,000 pound bombs on three freight transports, each of 7,000 tons, sinking them.” The internal Arlington Hall memo noted that a Times reader probably thought it was “chance” that the bombing mission found the convoy. Not so. In fact, “a message had been intercepted and read” two weeks earlier.
November 1943, one month after Dot’s arrival at Arlington Hall, marked the war’s most devastating month for Japanese tonnage sunk. U.S. subs sank forty-three ships and damaged twenty-two. American sub captains received intelligence of seventy-six movements of enemy ships. In December, American subs sank or damaged about 350,000 tons, including thirty-two ships sunk and sixteen damaged.
Behind the success of the U.S. Navy were the code breakers. “The success of undersea warfare is to a certain extent due to the success with which Japanese code messages were translated,” noted a naval report. An American naval commander pointed out in a postwar memo that sometimes a convoy might slip through, but only because U.S. submarines were kept so busy by information from decoded messages that they could not handle all the convoys they were alerted to. Over at the Naval Annex, the assembly line of WAVES identified the movements of marus supplying the Japanese Navy. Findings from both operations found their way to the submarine captains, who could hardly keep up with the bounty of intelligence.
After the war, a census would be taken of Japanese marus and their fates. The files take up boxes and boxes. Here is a single page from that immense archive:
On July 2, 1943, the Isuzu Maru was sunk by a submarine.
On December 2, 1944, the Hawaii Maru was sunk by a submarine.
On October 16, 1944, or thereabouts, the #23 Henshuu Maru was sunk by aircraft.
On August 31, 1944, the #20 Hinode Maru was sunk by submarine.
On October 16, 1944, the #16 Hoorai Maru was sunk by aircraft.
On January 27, 1945, the Hisn Yang Maru was sunk by a mine.
On January 2, 1944, the Isshin Maru was sunk by submarine.
On January 20, 1944, the Jintsuu Maru was sunk by aircraft.
On September 12, 1944, the Kachidoki Maru was sunk by submarine.
And so on.