The Secret in Building 26

by Jim DeBrosse

  Single hot points, however, were rare and inefficient. A bad guess at the stecker often proved more useful to codebreakers. It led the Bombe to conduct more tests of inconsistencies and allowed an instantaneous check for solutions. *7

  If the guess at the stecker was incorrect but the wheel positions were correct, the current would race through all but one of the twenty-six connections, creating what codebreakers called a “cold point.” The register’s relays would sense the all-but-one condition and signal the Bombe to stop. *8

  The Bombe’s primary task was to eliminate vast numbers of impossible settings rather than to pinpoint a correct solution. To do that well, it needed menus with enough letters, loops, and special relationships to focus its search on worthwhile solutions. The British soon calculated that a menu with eleven or twelve letters and two or three loops would prevent excessive false stops. Keen thus increased the number of Enigmas in each Bombe bank to twelve, so that the Bombe could test each menu letter and stecker guess simultaneously and speed the way to a solution. Keen tried to make his machines even more efficient by having three sets of banks (twelve Enigmas) in each of his Bombes so they could run more than one wheel order at a time. But that wasn’t enough to make the Bombe a full-power elimination machine.

  For that, Keen needed another important component, outlined by Turing’s colleague Gordon Welchman, called the Diagonal Board. Its power came from the reciprocal nature of the Enigma’s plugboard—that is, if A is steckered to B, then B must be steckered to A. The board was so named because its wiring made a diagonal connection between a column of letters, A to Z, and an identical row of letters. The columns and rows were all connected reciprocally to one another, in the same way as the plugboard on an Enigma machine. During a Bombe run, if current was sent to column B row A on the Diagonal Board, the board automatically shunted it back through row A column B and zapped it again through the Bombe’s circuitry for more tests. Welchman’s revolutionary idea increased the power of the Bombe’s loop and stecker-contradiction tests by 60 percent.

  Bletchley’s 1941 breakthroughs did not create a secure future against the Enigma. In early 1942, when the U-boats began using M4 Shark, Bletchley was locked out of the system until the three-wheel Bombes at last broke into Shark in December, and then only after codebreakers got some unintended help from the Germans, who began using some M4s in the old three-wheel Enigma mode.

  Even at that, the powerful cribs needed for the turnaround were secured at the cost of the lives of two brave British sailors who captured the weather-signal codebooks from U-559 in October 1942. Lieutenant Anthony Fasson and Able Seaman Colin Grazier of the HMS Petard were trying to retrieve the last of the sub’s secret materials when the damaged U-boat, without warning, plunged beneath the waves and pulled them to their deaths.

  IN THE SUMMER of 1942, the U.S. Navy was just beginning to digest all the information that Bletchley Park had given them on the Enigma and the Bombe, including Welchman’s ingenious Diagonal Board. In light of the new information, the theoreticians at OP20G redrew the logic of their machine to draw on the power of the Diagonal Board, Turing’s cold-point test, and Keen’s two-way flow of current. They then set off to Dayton in August 1942 to have Desch turn their ideas into hardware.

  But the Navy researchers did not bring with them to Dayton a finished plan. Desch had much to figure out on his own and many adjustments to make in the fall of 1942 as further information from the British about their Bombes and their methods trickled down from OP20G.

  Desch’s struggle might have been less hurried, less intense, and less frustrating if he and his Navy superiors had known more about the workings and the successes of the British Bombes prior to the summer of 1942. Whether the British withheld crucial information from their closest ally was, and still is, being debated. But all the arguing would have been moot—and the U.S. Bombe on schedule to prevent the U-boat slaughter in March 1943—if the year before, another Ohioan had not spurned an offer of help from Bletchley Park.

  3

  Miss Aggie’s Big Blunder

  August 18, 1941—Washington, D.C.

  WALKING WITH A cane and a severe limp from her buckled right leg, Agnes May Driscoll—at age fifty-three one of the Navy’s top codebreakers—entered the office of her boss, OP20G director Laurance F. Safford. Waiting there to meet her was the head of Bletchley Park, Commander Alastair G. Denniston. From the moment of their introduction, Driscoll and Denniston must each have been fighting a feeling of distrust—which is why Denniston had traveled to Washington in the first place. After months of misunderstanding and tensions between the British and American codebreaking operations, Denniston was prepared to be more open about what the GCCS had learned in tackling the Enigma—in part because the British feared that the Americans, if left too far out of Ultra’s tight circle, might try to build an anti-Enigma device of their own. Denniston wanted the U.S. Navy to leave the Enigma message cracking to the British, who (unlike the Americans, in his view) could be counted on to keep a secret.

  The two veteran codebreakers presented a striking contrast as they sat across from each other in Safford’s office. Denniston, sixty-three, a diminutive and dapper Scot who often wore a flower in his lapel, was ten years Driscoll’s senior but a robust, outgoing man known for his relaxed, conciliatory ways. In fact, his easygoing nature was to later lead to charges that he was a poor manager of people and was a factor in his eventual removal as head of Bletchley Park.

  On the other hand, Driscoll, who was crippled and increasingly frail since a car accident four years before, was a conservative dresser, a workaholic, and, although pleasant to others, largely a social recluse. She was also deeply religious, a convert to Christian Science who had declined medical care after the car accident that had broken her jaw and right leg below the knee. As a result, the leg had never healed properly, and it bowed outward from her slight frame at about twenty degrees.

  Despite their differences in temperament and appearance, the two were remarkably similar in their backgrounds. Both were children of small-town professionals—Denniston’s father had been a doctor in rural Scotland and Driscoll’s had been a music professor at a college in Westerville, Ohio. Denniston had taught foreign languages before he was recruited for naval-intelligence work as a German linguist; Driscoll had been a math teacher in Texas before signing up with the women’s naval auxiliary during World War I. Both were dedicated to their craft and their countries.

  But Denniston, unlike Driscoll, had seen that codebreaking was now moving into a new era, a mathematical one. Even before the Nazi invasion of Poland, he had begun recruiting a new class of codebreaker—young mathematicians such as Alan Turing and Gordon Welchman—who would lead the way toward automation and state-of-the-art electronics to help turn the new methods into practical attacks.

  Driscoll, however, still put her faith first in the codebreaker’s intuition and the old-fashioned, hard-earned methods of pencil and paper. To Denniston’s surprise, she was more than uninterested in Britain’s automated attacks against the Enigma. In that critical August 1941 meeting, she declared she did not want or need a Bombe, nor did she ask any questions about the machine or its supporting technique of Banburismus. And, further, she told Denniston she had already devised an Enigma method far better than those in use at GCCS. She then refused his invitation to travel to England, where she would be shown all GCCS had achieved and could tell all about her own attack. Her brush-off of Denniston was to cost the United States dearly, and its effects were compounded when the British commander was soon “promoted” out of GCCS and replaced by leaders less sympathetic to the Americans.

  Neither Driscoll nor Safford informed their Navy superiors of Denniston’s offer—an oversight that hampered British-American relations for the next four years. Driscoll’s blunder never found its way into U.S. Navy records.

  Why did Driscoll spurn Denniston’s overture? Perhaps it was professional hubris. Perhaps it was the Navy’s continuing distrust of En-gl
and, whose military leaders had so dominated the American venture into World War I. Or perhaps it was a product of too many years of isolation at OP20G as the Navy, understaffed and overwhelmed, tried to hold on to its precious few tricks and secrets. Driscoll and Safford seemed to have been cozily optimistic and dangerously ignorant about the ability of their small and inexperienced crew to conquer the intricacies of the Enigma.

  Driscoll’s age may have been a factor, too. At fifty-three, she was beyond her peak creativity and increasingly prone, like many of us, to inflexibility. Codebreakers, like mathematicians, tend to produce their most innovative work early in their careers. At the same time, OP20G’s restricted budgets prior to the war had made it difficult to bring young, bright, and well-trained people into its fold—people who could stimulate new ideas and new methods, the way Agnes Driscoll had been able to do decades before.

  Driscoll may not have been on the cutting edge in 1941, but she had reason to trust her own professional judgments. Prior to the war, she had often been the first to break into the latest and toughest codes and cipher systems confronting the Navy. In the 1920s and 1930s, she had trained the first generation of Navy officers and civilian cryptanalysts inside the sprawling Main Navy Building on Constitution Avenue in Washington, including Safford, who became her commanding officer and biggest advocate. Safford had been not only one of Driscoll’s star students but the first regular Navy officer to commit his career to cryptology. Both he and Driscoll were from the old school of codebreaking, and neither had taken the lead in OP20G’s sometimes disappointing ventures into statistical, automated cryptanalysis. Together, however, they shared an unwavering commitment to the Navy’s, and the nation’s, vital stake in cryptanalysis and were quick to defend its boundaries and its secrets, even against the British.

  HISTORIANS KNOW MUCH about Driscoll’s many career accomplishments but frustratingly little about her personal background. For professional reasons, Driscoll jealously guarded her private life, shunning the spotlight and even refusing, except on rare occasions, to be photographed. From the record alone, however, it is clear that she was as pioneering in her career path as she was remarkable in her abilities.

  She was born Agnes May Meyer in 1889 in Geneseo, Illinois, the second daughter and third child of a music-professor father and a mother who reared eight children. Despite the large family to support, her parents made sure that Agnes received a thorough education. She graduated from high school and went on to college, an unusual privilege at the time for men or women.

  From 1907 to 1909, she attended Otterbein College—a small Methodist institution in Westerville, Ohio, where her father then taught music theory. Founded by the United Brethren in Christ, Otterbein was among the first colleges to welcome students from nonelite backgrounds and among the first in America to encourage female students to take all types of courses, including those in the sciences.

  Meyer finished her undergraduate degree in 1911 at Ohio State University, not far from Westerville, in Columbus. There, her main subjects of study were mathematics, physics, music, and languages. After college, she wasted no time striking out on her own, venturing west to Amarillo, Texas, where she served first as a music director of a small military academy and later as head of the mathematics department at Amarillo High School, a position of considerable status at the time.

  America’s entry into world war in 1917 changed forever the safe course of Meyer’s professional life. No doubt driven by both her adventurousness and her patriotism, she was one of the first to answer the call when a shortage of labor, and some intense political prodding, induced the Navy to open its doors to women. She and her sister enlisted in the brand-new women’s Naval Reserve, known as the Yeomanettes, in June 1918. Meyer was given the highest ranking attainable for a woman at the time, chief yeoman (F), partly because she had a working knowledge of stenography but as much because of her education and her age, which was nearly thirty. Even the Navy at the time couldn’t ignore her many other skills—including proficiency in German, French, Latin, and Japanese, as well as in statistics, math, physics, and engineering.

  After the war, she was discharged from the reserve and rehired as a steno clerk for the Office of the Director of Naval Communications. But given her skills for codebreaking, the Navy soon sent her off for special training in 1920 with George Fabyan, the wealthy and eccentric textile magnate who had founded his own school of codebreaking at Riverbank Laboratories, near Chicago.

  Meyer so impressed Fabyan with her abilities that he offered to pay her expenses and match her Navy salary in order to keep her working at Riverbank until the Navy wanted her back. Nevertheless, after her training ended Meyer returned to the Navy, where she helped develop an enciphering machine called the CM. Congress later awarded her a large cash sum in lieu of patent rights for her invention.

  In 1921, Meyer ran across a challenge she could not resist. Edward Hugh Hebern, the inventor of one of the first cipher machines to use rotors, placed an ad in a maritime magazine touting his device as “unbreakable.” Meyer solved the sample message in the ad and sent it off to Hebern, who must have been both humbled and dazzled to have a woman prove him so wrong. In February 1923, Hebern did the sensible and politically astute thing by hiring her as his technical adviser and liaison with the Navy. Even though she remained a Navy employee, Meyer worked for a company that was trying to sell its products to the Navy—a touchy ethical issue today but not so uncommon then.

  In the spring of 1924, with Hebern’s company in financial straits, Meyer decided to return to the Navy and continue her work in cryptanalysis. There were no doubt personal reasons as well. In August of that year, at age thirty-five, she wed Washington lawyer and fellow Illinois native Michael Bernard Driscoll.

  Driscoll began her codebreaking duties for the Navy in earnest in the early 1920s, not long after the Office of Naval Intelligence had secretly financed a series of break-ins at the Japanese consulate in New York City whose scope and daring make the Nixon-era burglary at the Watergate Hotel look like child’s play. The entire Japa-nese fleet codebook was photographed, page by page, during repeated undercover operations never detected by the Japanese, and then translated over the next four years. The result was called the Red Book—two thick volumes of encoded word groups bound in red material.

  Codes and ciphers are two different methods of disguising messages. Codes substitute words, letter groups, or numbers for other words or letter groups, usually copied from a list or book. Ciphers are generally harder to crack because they are not simple substitutions: they replace plain-text letters with other letters or numbers according to a specific procedure, known as a “key.” Ciphers were increasingly generated by machines as World War II approached, but codes continued to be dispensed in books.

  A good example of a code system, the Red Book contained a numerical value for every word or syllable likely to be used in a Japanese naval message. But those simple substitutions, easy prey for a crack team of codebreakers, were further disguised by a cipher. Before sending a message over the radio, a Japanese code clerk needed a second book filled with page after page of random numbers. Starting from the top of the page and working down, the clerk added (in some systems subtracted) a different random number, or “additive,” to each of the code groups. The addition was done digit by digit, without carrying the sum, to keep each code group to five letters. Buried somewhere inside the message was an indicator telling the receiving clerk what page in the additive book had been used to encipher the basic code. The clerk turned to that page and stripped off the additives before looking up the meaning of each code group. *9

  In 1925, the immense job of turning the Red Book into a useful codebusting tool fell to Agnes Driscoll. She devised craftsmanlike ways of stripping off the additives and of identifying the meaning of underlying code groups that had not been easily translated. Even though most of the code-group meanings were known by then, this was no small feat: the two volumes contained some one hundred thousand code gro
ups, with as many as three groups assigned to each word.

  Driscoll got a raise and a promotion for her breakthrough, which at last gave the Navy a chance to tap into the secret messages of its chief antagonist in the Pacific. Although handicapped by a lack of people and resources during the 1920s and 1930s, the Navy’s codebreakers were able to process some important Japanese messages, revealing details about their naval maneuvers, fuel supplies, and advances in aviation. But OP20G’s weak hold was always under threat: the Japanese had the smart habit of changing their code systems just before any major shift in their strategy.

  By the spring of 1930, when the Japanese Navy began conducting fleet maneuvers in the Pacific aimed against its envisioned enemy, the United States, they introduced a new cipher system. But they still turned to their Red Book for the underlying code groups. Driscoll showed her stuff again and got the first break into the new system. The results shocked the Navy: they discovered that Japanese naval intelligence had been able to obtain a very accurate picture of the Navy’s own latest war plan, dubbed Orange. The revelation convinced the Navy’s codebreakers that they had to persuade the government to invest in radio intelligence.

  Driscoll’s reward this time was something she had not asked for or even felt she needed: an IBM tabulating machine to help relieve some of the drudgery of her clerical labors. “Miss Aggie,” as she was called by colleagues, was none too excited about mechanized approaches to her labors, having relied for more than ten years on little more than pencil and cross-section paper. But Driscoll soon found the unwanted tabulator a necessity when Japan dramatically altered its main naval code. She had been the first to notice in the fall of 1931 that the Japanese had at last dumped the Red Book for a new code and cipher system. This time, without the benefit of burglary and secretly photographed pages of codes, Driscoll and her team would have to crack both code and cipher. It was tedious, mind-numbing work, but the new machine helped keep track of the nearly eighty-five thousand code groups and changing cipher keys as each Japanese message was intercepted and processed.