THE CODEBREAKERS

by DAVID KAHN

  This is why Friedman has said, in looking back over his career, that The Index of Coincidence was his greatest single creation. It alone would have won him his reputation. But in fact it was only the beginning.

  He and Mrs. Friedman quit Riverbank near the end of 1920. The situation had become intolerable. Fabyan had lured him back after the war with raises and promises of absolute freedom to prove or disprove the existence of ciphers in Shakespeare. But he had squelched every attempt to do so and had embarrassed Friedman into apparently acquiescent silence at lantern-slide lectures on the subject. On January 1, 1921, Friedman began a six-month contract with the Signal Corps to devise cryptosystems. When it expired, he was taken on the civil-service payroll of the War Department at $4,500 a year.

  One of his first assignments was to teach a course in military codes and ciphers at the Signal School, then at Camp Alfred Vail, New Jersey. For this he wrote a textbook that, for the first time, imposed order upon the chaos of cipher systems and their terminology. These had sprouted in a bewildering variety, and writers treated them individually, with little comprehension of the close connection between, say, the Vigenère and the Gronsfeld. Friedman sorted them out on the basis of structure instead of aspect, and so logical and useful was this classification that it has become standard. He modeled his nomenclature on his categories, so that the names he minted have the great merit of making the relations between the various genera of ciphers evident on sight. An example is the complementary pair “monalphabet” and “polyalphabet”; Givierge was even then calling polyalphabetic systems by the almost obfuscatory “double substitution,” which tells absolutely nothing at all about the system. Friedman’s most important coinage was the word “cryptanalysis,” which he devised in 1920 to clear up a chronic source of confusion in cryptology—the ambiguity of the verb “decipher,” then used to mean both authorized and unauthorized reductions of a cryptogram to plaintext. He titled his book Elements of Cryptanalysis, and the term has so prospered that today it circulates in general conversation and print.

  While the book’s main contribution is its taxonomy, each of its 143 pages of text manifests the author’s concern for always making clear to the reader why things happen as they do. As a result, the student understands principles and phenomena, and the lessons stick. Partly because of this pedagogical effectiveness, partly because of its substantive values, Friedman’s book, issued by the Chief Signal Officer in May of 1923 as Training Pamphlet No. 3, has guided the development of all American cryptology since then.

  At the start of 1922, Friedman became Chief Cryptanalyst of the Signal Corps in charge of the Code and Cipher Compilation Section, Research and Development Division, Office of the Chief Signal Officer. To help him carry on the work of the office he had a single clerk-typist—a cauliflower-eared ex-prizefighter. Because Yardley’s Black Chamber was doing the cryptanalysis for the War Department, Friedman’s functions were nominally cryptographic. He installed the M-94, or Jefferson-Bazeries cylinder, as the Army’s field cipher. Paradoxically, however, his job involved a great deal of cryptanalysis. He was continually testing the new systems of cryptography urged on the Army as “absolutely indecipherable” by zealous amateurs.

  Most difficult of these was the machine with five wired codewheels—rotors—invented by Edward H. Hebern, whose principle is today the most widely used in high-level cryptography. Each of the rotors generates a progressive cipher, and in 1925 Friedman devised the kappa test and extended his Index of Coincidence analyses to determine the order and starting positions of the rotors. The five progressive ciphers intertwine in a cipher of hideous nightmare complexity, but in a later solution Friedman sorted them out and reconstructed the wiring of the rotors. This work was of the utmost importance, for it laid the foundations for the PURPLE machine solution and for today’s many solutions of modern rotor machines. The technique was far in advance of its time. So far as is known, not another cryptanalyst on the globe could duplicate it—and none did, apparently, for more than two decades. With this solution of Friedman’s, world leadership in cryptology passed to America.

  Friedman’s horizons were continually expanding. In 1922, he had filed applications for his first two patents—improvements on a device recently invented by Gilbert S. Vernam. In 1924, he testified before a Congressional committee to his reading of some messages in the Teapot Dome scandal. When Mars made an extremely close approach to Earth a few months later, he joined in the Roaring Twenties wackiness by standing by to translate any revelations the Martians may have condescended to pass along. He had returned to mundane problems by 1927, when he wrote a history and theory of commercial codes for the American delegations to international communications conferences, which were then heatedly discussing the pronounceability of codewords as a basis for cable toll rates. The following year, he served as secretary and technical advisor to the American delegation to the International Telegraph Conference of Brussels. In 1929, he became widely known as one of the world’s leading authorities on cryptology when the Encyclopaedia Britannica published his article on “Codes and Ciphers.”

  Meanwhile, the Army had been studying its divided cryptologic operation and, shortly before the State Department withdrew support from Yardley’s bureau, had decided to integrate both cryptographic and cryptanalytic functions in the Signal Corps. The closing of the Black Chamber eased the transition, and on May 10,1929, cryptologic responsibility devolved upon the Chief Signal Officer. To better meet these new responsibilities, the Signal Corps established a Signal Intelligence Service in its War Plans and Training Division, with Friedman as director. Its officially stated mission was to prepare the Army’s codes and ciphers, to intercept and solve enemy communications in war, and in peace to do the training and research—a vague enough term—necessary to become immediately operational at the outbreak of war. To carry out these duties, Friedman hired three junior cryptanalysts, all in their early twenties, at $2,000 a year—the first of the second generation of American cryptologists. They were Frank Rowlett, a Virginian, and Solomon Kullback and Abraham Sinkov, close college friends who had taught together in New York City high schools before coming to Washington and who both received their Ph.D.’s in mathematics a few years later. It was the beginning of an expansion that led to the massive cryptologic organization of today.

  By this time the Navy, too, had its cryptologic section. Like the Army’s, it had evolved gradually.

  When the Navy was reorganized during World War I in its present form, with a Chief of Naval Operations, responsibility for cryptography was transferred in October 1917 from the Bureau of Navigation, which had long held it, to the new Office of Naval Communications. The four young assistant communication officers—who were burdened by the unfortunate Navy jargon-abbreviation “ASSCOMS”—coded and decoded messages in their office in the old State-War-Navy departments building, where Yardley also worked. They were doing this work even before the transfer. When an inquiry was held on the loss of a battle signal book, letters flooded in from amateurs all over the country who thought they had the answer to the Navy’s code security. It fell to the senior assistant communication officer, Lieutenant (j.g.) W. W. (Poco) Smith, to reply. He picked up the gauntlet thrown down in the form of challenge messages, and solved them, learning a great deal about cryptology in the process.

  In 1916, the Navy had three main codes: the old, ponderous, seldom-used Secret Code of 1887, the five-letter SIGCODE, which could be used only by officers and which had a variety of ciphers, some for flag officers, some for all Navy ships, and the four-letter radio code that was only confidential and that could be worked by enlisted men. But at the time of the Marine landing in Haiti, a State Department message was transmitted there in the SIGCODE; the plaintext was published; the code was assumed compromised, and Smith was designated to prepare a new one.

  “It was,” he recalled, “a colossal job. First, I simplified but expanded the context…. Now, a more difficult problem: five-letter code groups and the t
ext were both arranged alphabetically. This would not do. Exhausting the possibilities of arranging the letters of the alphabet into mixed groups of five letters each,* I typed these in columns, scissored them, and dropped them into a bucket. After mixing, I drew the groups one at a time and typed them in double-spaced columns to be placed opposite the text words or phrases to be encoded. Tedious work.” Also crude and time-consuming. Naval Code A-1 was not completed and printed by the Government Printing Office until after the United States had entered World War I.

  While Smith was making up the code, the Navy set up a Code and Signal Section in naval communications to handle the cryptographic duties. In charge was Lieutenant Russell Willson, who devised a strip form of the Jefferson cylinder with fixed indices as a superencipherment system. The metal frame and strips on which the mixed alphabets were stamped were manufactured at the Naval Gun Factory in southeast Washington, and the device, called the “NCB” (for “Navy Code Box”), was used from 1917 on. (Congress awarded Willson $15,000 in 1935 for the Navy’s use of the device, which was then still in service.) Meanwhile, Smith, who wanted to remain a line officer for future command and not become a deskbound specialist, sailed off to war in January, 1918, vowing never to return to communications duty—which he never did. An experience like his in code-construction would have conditioned anybody against cryptology. (He did, however, write one of the classic expositions of the solution of the Playfair cipher, which appeared in J. C. H. Macbeth’s translation of André Langie’s De la cryptographie.)

  Naval participation in the war was too limited for much cryptanalytic development, but interest was stimulated. Accordingly, in January of 1924, Lieutenant Laurance F. Safford was ordered to set up a radio intelligence organization in the Code and Signal Section. When he left for sea duty two years later, a small, highly secret organization was functioning in Room 2646 of the “temporary” Navy Department building on Constitution Avenue. Lieutenant Ellis M. Zacharias, who trained seven months in 1926 with the cryptanalytic organization, told what it was like:

  My days were spent in study and work among people with whom security had become second nature. Hours went by without any of us saying a word, just sitting in front of piles of indexed sheets on which a mumbo jumbo of figures or letters was displayed in chaotic disorder, trying to solve the puzzle bit by bit like fitting together the pieces of a jigsaw puzzle. We were just a few then in Room 2646, young people who gave ourselves to cryptography with the same ascetic devotion with which young men enter a monastery. It was known to everyone that the secrecy of our work would prevent the ordinary recognition accorded to other accomplishments. It was then that I first learned that intelligence work, like virtue, is its own reward.

  On completion of his apprenticeship, Zacharias took charge of an intercept post on the fourth floor of the American consulate in Shanghai to learn as much as he could from Japanese naval messages. Safford returned to cryptology in June, 1929, and, except for a four-year tour at sea from 1932 to 1936, stayed with the science from then on. He built up the communications intelligence organization into what later became OP-20-G and, by adding improvements of his own to Edward Hebern’s rotor mechanisms, gradually developed cipher machines suitable for the Navy’s requirements of speed, reliability, and security. His contributions to cryptanalytics were minor, since his talents lay more in the administrative and mechanical fields. But he is the father of the Navy’s present cryptologic organization.

  In the Munitions Building next door to the Navy Department, Friedman had begun tutoring his junior cryptanalysts, who had not the feeblest knowledge of codes and ciphers, in these arcane mysteries. They discovered an aptitude for them. In November, 1931, they and Friedman solved in a few hours a teletypewriter cipher machine offered for sale to the State Department by its inventor, Parker Hitt, then of International Telephone and Telegraph. In 1934 they prepared a paper on a general solution for the ADFGVX, and in 1935 Kullback devised the phi and chi tests, publishing them in an important monograph entitled Statistical Methods in Cryptanalysis. Friedman wrote Elementary Military Cryptography, Advanced Military Cryptography, and Military Cryptanalysis, the latter an expansion of his Elements of Cryptanalysis, as texts for Army extension courses. Military Cryptanalysis, which appeared in four parts, comprises the finest, most lucid exposition of the solution of basic ciphers that has ever been published.

  Gradually, despite depression and isolationism, the Signal Intelligence Service expanded. In July of 1934, First Lieutenant W. Preston (Red) Corderman, who had studied in what was rather grandly known as the S.I.S. School (the faculty consisted of Friedman and his assistants), became an instructor in that school when it was formally constituted as a separate section. In August of 1935, Major Haskell Allison replaced Friedman as administrative head of S.I.S., though Friedman continued to direct the cryptologic activities. Its first sizable expansion came in the fiscal year 1938, when the number of civilian employees (clerks included) was raised from six to eleven on a personnel budget of $24,360.

  During these years, Friedman further expanded his interests. He discussed the cryptologic abilities of Edgar Allan Poe and Jules Verne in scholarly articles, solved ciphers posed as challenges by earlier writers on cryptology, investigated historical problems such as the Zimmermann telegram and the field codes of the A.E.F. He got important works translated, and annotated them with his ubiquitous “W.F.F.” He continued to patent inventions. Unfortunately, caught between the need for secrecy and a desire for fame, he tended to play the dog in the cryptologic manger—if he couldn’t have the glory, no one else would. His usual tactic was to blacken amateur contributions, often quite worthwhile, as “unprofessional.” His wife, who had solved the codes of rumrunners during Prohibition, continued her cryptanalytical activities for the Treasury Department. They even managed to raise two children, Barbara and John Ramsay.

  In the late 1930s, as the crisis of war drew near, the Army accelerated its plans for mobilization. Of the entire War Department establishment, the S.I.S. was the first to be augmented in personnel, space, and facilities. On November 2, 1939, authorization was obtained for 26 more civilian employees. Selected civilians, enlisted men, and Navy reserve officers were allowed to take the extension courses previously given only to Army reserve officers; by June 30, 1939, a total of 283 students were enrolled. A few members of the American Cryptogram Association were recruited. The six Signal Service Companies in the field that had supplied intercepts to S.I.S. were centralized on January 1, 1939, in a 2nd Signal Service Company, with an authorized strength of 101 enlisted men.

  The driving force behind this expansion was the Chief Signal Officer, that one-time cryptologist, Joseph O. Mauborgne, now a two-star general. As one of the first steps in the upgrading of S.I.S., he had established it on April 23, 1938, as an independent section in his office. It was he who directed it to bend its energies to the solution of the Japanese PURPLE system and, as an old and close friend of Friedman, urged him to lead the assault. Friedman did—and with that bright genius of a dark science blazing the way, the S.I.S. team struggled upward in one of the most arduous, grinding, extended, and ultimately triumphant cryptanalyses in history. The date was August, 1940; Friedman was 48. With the conquest of this Everest, the greatest career in cryptology reached its climax. A few months later, the captain of the team succumbed to the strain of the solution. He was admitted to Walter Reed General Hospital on January 4, 1941, for a nervous breakdown and was discharged March 24. He had to retire, with a permanent disability, from his lieutenant colonelcy in the Signal Corps reserve.

  Afterward, his superiors refused to allow him to work more than a few hours a day, and then only in the less taxing area of communications security. Though he was still Chief Cryptanalyst of the War Department, he served as Director of Communications Research for the S.I.S. (under its various names, chiefly Signal Security Agency) throughout World War II. The post was a high one; military reviews were held for him when he visited intercept stations or ot
her cryptologic posts. He spent most of the war at the Arlington Hall Station, located in suburban Virginia in what had been a girls’ school. Not a few of the thousands who worked there remember with gratitude the natty, mustachioed man with the bow tie who picked them up as they stood shivering at the bus stop and gave them a lift into Washington.

  Friedman retained his directorship when the agency was divorced from the Signal Corps on September 15, 1945, and placed under G-2 as the Army Security Agency. Upon the creation of the Armed Forces Security Agency in 1949, he became chief of the technical division. When this agency was supplanted in 1952 by the National Security Agency, which handles most of the cryptologic activities of the United States, Friedman became chief technical consultant, and two years later, special assistant to the director. He had also been, since 1947, the Cryptologist of the Department of Defense.

  He retired in 1955, relinquishing all these posts but remaining as a consultant. In 1944, he received the Commendation for Exceptional Civilian Service, the War Department’s highest civilian decoration, and, in 1946, President Truman conferred upon him the Medal for Merit, the highest award for civilian service that the United States government can give. The citations were necessarily vague, referring only (in the case of the Medal for Merit) to “exceptionally meritorious conduct in the performance of outstanding service, conspicuously above the usual.” On October 12, 1955, at a ceremony before 500 people honoring his retirement, Allen W. Dulles, the Director of Central Intelligence, unexpectedly pinned the National Security Medal on Friedman’s breast. A picture snapped just after the presentation shows Friedman standing overwhelmed with surprise, apparently fighting back tears, as Dulles, Sinkov, Kullback, and Major General Ralph J. Canine, director of N.S.A., applaud. The medal, the highest decoration for distinguished achievement relating to the national intelligence effort, was the sixth to be awarded since its creation in 1953.