THE CODEBREAKERS

by DAVID KAHN

  All these developments, however, resulted essentially from the inter-reaction between cryptology and the outside world; they were externally oriented. World War I originated no developments that were internally oriented, as, for example, was the emergence of the field cipher. On the contrary, two of the most central activities—the actual cryptographic operations, which were performed by hand, and the techniques of solution, which were brute frequency analysis—had exhausted their usefulness.

  Manual systems sagged under message loads for which they were never designed. Not a few cryptographic clerks dreamed of machines that would lift the onerous burden from their shoulders. In a sense, the codes that became so popular might be regarded as a rudimentary form of mechanical device that does the work for the encoder: the phrases are prepared and equated with their code equivalents in advance, and the encoder has but to pick out the ones he wants. But the trench codes were to the printing cipher machines of later years as the taxis of the Marne were to the armored troop-carriers of Panzer columns.

  At the same time, the classic principles of frequency analysis had been stretched to their utmost. They were applied with great subtlety, as in Painvin’s matching of frequency distributions to determine the odd and even columns of the ADFGVX transposition block. But no new principles had been evolved, and the old ones had barely coped with such concepts as fractionation.

  In these two internal matters, which lie at the core of cryptology, World War I marked not a beginning but an end, had reaped not fulfillment but barrenness. So viable had the science become, however, that this very vacuum, this want, held promise.

  * The only previous works on cryptology to appear in the United States were magazine or encyclopedia articles and two pamphlets—Mauborgne’s and a totally obscure work of 31 pages by one Harvey Gray, entitled Cryptography and published at Boston in 1874.

  * These six were apparently chosen because their International Morse symbols were sufficiently distinct to minimize garbles: A ·—D—·· F ··—· G——·V · · ·—X—· ·—

  12

  TWO AMERICANS

  THE MOST FAMOUS CRYPTOLOGIST in history owes his fame less to what he did than to what he said—and to the sensational way in which he said it. And this was most perfectly in character, for Herbert Osborne Yardley was perhaps the most engaging, articulate, and technicolored personality in the business.

  He was born April 13, 1889, in Worthington, Indiana, and grew up in that little Midwestern town during the tranquil, sunlit years that preceded the First World War. A popular youngster, he was president of his high-school class, editor of the school paper, and captain of the football team, and though only an average student, he had a flair for mathematics. From 16 on he frequented the poker tables of the local saloons, learning the game that was to be a passion of his life. He had wanted to become a criminal lawyer, but instead landed at 23 as a $900-a-year code clerk in the State Department.

  It was a case of purest serendipity, for the man and the subject were ideally matched. His romantic mind thrilled to the stream of history that daily poured through his hands in the form of ambassadorial dispatches, and cryptology fired his imagination. He had heard vague tales of cryptanalysts who could pry into secrets of state, and when a 500-word message from Colonel House passed over the wires to President Wilson one night, Yardley, with characteristic audacity, determined to see whether he could solve what must be the most difficult of American codes. He astonished himself by solving it in a few hours.* His success cemented his attachment to cryptanalysis, and he followed this demonstration of the low estate of high-level cryptography with a 100-page memorandum on the solution of American diplomatic codes. While absorbed in possible solutions for a proposed new coding method, he diagnosed what has ever since been known among cryptologists as the “Yardley symptom”: “It was the first thing I thought of when I awakened, the last when I fell asleep.”

  Soon after the American declaration of war in April of 1917, he sold the idea of a cryptologic service to the War Department. He succeeded partly because the need was genuine, partly because he himself was an exceedingly convincing young man. Yardley had proven his cryptanalytic ability, and moreover had done well enough in his regular duties to have won raises to $1,400 in 43 months. Major Ralph H. Van Deman, later to be known as the Father of American Intelligence, commissioned the thin, balding 27-year-old as a lieutenant and set him up as the head of the newly created cryptologic section of the Military Intelligence Division, MI-8.

  Like Topsy, MI-8 just grew. First to arrive, to take charge of the instruction subsection for training A.E.F. cryptanalysts, was Dr. John M. Manly, a 52-year-old philologist who headed the Department of English at the University of Chicago and was later president of the Modern Language Association; a longtime hobbyist in cryptology, he was to become Yardley’s chief assistant and one of his best cryptanalysts. Manly brought with him a bevy of Ph.D.’s clanking with Phi Beta Kappa keys, mostly from the University of Chicago: David H. Stevens, 32, an instructor in English, later director of the division for the humanities of the Rockefeller Foundation; Thomas A. Knott, 37, associate professor of English and later general editor of Webster’s Dictionaries, including the colossal 1934 Second New International Unabridged; Charles H. Beeson, 47, associate professor of Latin, later president of the Mediaeval Academy of America, who had gotten his doctorate at Munich and knew German well enough to write scholarly works in it; and Frederick Bliss Luquiens, 41, professor of Spanish at Yale University, general editor of the Macmillan Spanish Series, and author of An Introduction to Old French Phonology and Morphology.

  The instruction subsection did its teaching at the Army War College. It advanced far enough to offer as Problem 20 “General Principles of attack on enciphered code when the book is known but the system of encipherment unknown.” Another subsection popped into being for code and cipher compilation; it produced a military intelligence code, two geographical codes for combat information from France, and a casualty code, which was never used. Soon a communications subsection was handling close to 50,000 words a week. As the organization expanded, it shifted to ever-larger quarters. Beginning in the balcony overhanging the library of the War College, MI-8 moved to the Colonial, an apartment house at 15th and M Streets barely ready for occupancy, and then to a building on the site of what is now the Capitol Theatre on F Street, all in Washington. For security, its offices were always on the top floor.

  Growth continued apace. An intercepted letter in a German shorthand instigated a shorthand subsection that soon could read missives in more than 30 systems, most commonly Gabelsberger, Schrey, Stolze-Schrey, Marti, Brockaway, Duploye, Sloan-Duployan, and Orillana. A blank piece of paper discovered in the shoe heel of a woman suspected of working with German espionage in Mexico turned out to bear a message in invisible ink. Fortunately, it proved one of the simpler kinds, which can be developed by heat. But it sparked the establishment of a secret-ink subsection whose expert chemists could detect writing in an invisible ink disguised as a perfume with an actual odor and with only one part in 10,000 of solid matter.

  The Germans later replaced inks in so bulky and conspicuous a form as liquids with chemicals that were impregnated into scarves, socks, and other garments. They had only to be dipped in water to create the writing fluid. These miracles of the test tube, called F and P inks by the British chemists who taught the Americans much of what they knew, were so precisely formulated that they would react with only one other chemical to form a visible compound.

  Eventually, the Allied chemists discovered a reagent that brought out secret writing in any kind of ink, even clear water. Crystals of iodine, heated gently, sublimated into fumes of a beautiful violet hue that settled more densely in those fibers of paper that had been disturbed by any kind of wetting action, thus tracing the pen’s course. The Germans replied by writing in a sympathetic ink and then moistening the entire sheet. The Allies struck back with a chemical streak test that would show whether the paper surface
had been dampened. This was almost as incriminating as actual development of a secret-ink letter, for who but a spy would wet a letter? The seesaw battle between the chemists of Germany, traditionally world leaders in that science, and those of the Allies reached a stalemate when both sides discovered the general reagent—one that would develop any secret ink at any time, even on moistened paper. Formulas differ slightly, but all use a mixture of iodine, potassium iodide, glycerine, and water, dabbed on with cotton. The liquid concentrates in the more disturbed fibers and reveals the writing. By the time this general reagent appeared, MI-8’s secret-ink subsection was testing 2,000 letters a week for invisible writing and had discovered 50 of major importance. Among them were letters that led to the capture of Maria de Victorica, a beautiful German spy who was planning to import high explosives for sabotage inside the hollow figures of saints and the Virgin Mary!

  MI-8 also solved cryptograms. It read diplomatic telegrams of Argentina, Brazil, Chile, Costa Rica, Cuba, Germany, Mexico, Spain, and Panama. The Spanish-language texts constituted the bulk of its cryptanalytic work. The censorship office sent over intercepted cipher letters; most of these turned out to be merely personal notes in very simple systems, though some of the love letters were so torrid that Yardley said, “It rather worried me to see husbands and wives trust their illicit correspondence to such unsafe methods.”

  Perhaps the most important of the MI-8 solutions was the one that largely resulted in the conviction of the only German spy condemned to death in the United States during World War I. This was Lothar Witzke, alias Pablo Waberski, who was suspected of setting off the Black Tom explosion. He was captured in January, 1918, by an American agent, who found in his baggage in the Central Hotel in Nogales, Mexico, a cipher letter dated January 15. It did not reach MI-8 until spring, and then it kicked about for a few more months while several men there tried and failed to solve it. Finally Manly took it up.

  This quiet scholar, who never married and whose quiet, simple manner contrasted so sharply with his chief’s, was to become one of the world’s leading authorities on Chaucer. He and his collaborator, Edith Rickert, labored for 14 years to produce their monumental eight-volume work, The Text of the Canterbury Tales, in which, by a tedious collation of scribal errors and variant readings in more than 80 manuscripts of the medieval masterpiece, they reconstructed a text that is as close to the poet’s own original as the extant evidence allows. The cast of mind that can thus sort out, retain, and then organize innumerable details into a cohesive whole was just what was needed for the Gothic complexity of the 424-letter Witzke cryptogram. In a three-day marathon of cryptanalysis, Manly, aided by Miss Rickert, perceived the pattern of this 12-step official transposition cipher, with its multiple horizontal shiftings of three- and four-letter plaintext groups ripped apart by a final vertical transcription. He drew forth a message from Heinrich von Eckardt, the luckless German minister in Mexico whose very involvement with a cryptogram seemed to mean its cryptanalysis,* to the German consular authorities:

  “The bearer of this is a subject of the Empire who travels as a Russian under the name of Pablo Waberski. He is a German secret agent. Please furnish him on request protection and assistance; also advance him on demand up to 1,000 pesos of Mexican gold and send his code telegrams to this embassy as official consular dispatches.” When Manly read this to a military commission of colonels and generals who were trying Witzke on spy charges in a hushed courtroom at Fort Sam Houston, San Antonio, the effect was condemnatory. The handsome young spy was sentenced to death. Wilson later commuted it to life imprisonment, however, and Witzke was released in 1923.

  In August of 1918, Yardley sailed for Europe to learn as much as he could from America’s allies. He obtained entrance to M.I. 1(b) after demonstrating his abilities to Brooke-Hunt and there studied British methods for the solution of different codes and ciphers. The doors of Room 40 remained resolutely locked against him as against everyone else, though Hall did give him a German naval code and a neutral nation’s diplomatic codes. In Paris that fall, Yardley met Painvin, who gave him a desk in his office and invited him to his home many evenings. But he never gained access to the French Foreign Ministry cryptanalytic bureau.

  He remained in Paris after the Armistice to head the cryptologic bureau of the American delegation to the Peace Conference. At first there was a tremendous rush to get organized, but then the pressure eased, and Yardley, Childs, who was assigned to assist him, and Lieutenant Frederick Livesey, who had been sent over from MI-8, enjoyed the life of playboy cryptologists. A practical soul, Yardley saw no need for the three officers assigned to the bureau to be present at once, and so a rotation of duties was arranged that permitted them to spend most of their time at the international cocktail parties and dancings that were then the rage of Paris.

  When it ended, as it had to, Yardley, viewing with distaste a return to the State Department code room, and burning with evangelical fervor over America’s need for cryptanalysis, exercised his potent salesmanship on the State and War departments. He won the concurrence of Frank L. Polk, the acting Secretary of State; then, on May 16, 1919, he submitted to the Chief of Staff a plan for a “permanent organization for code and cipher investigation and attack.” Three days later the Chief of Staff approved it, and Polk brown-penciled an “O.K.” and his initials on it. The plan envisioned joint financial support by the two departments at about $100,000 a year, but actual expenditures never reached that sum. The State Department’s contribution of $40,000, which began on July 15, 1919, could not be legally expended within the District of Columbia, and so Yardley soon found himself moving the nucleus of a staff (largely recruited from MI-8) and the necessary paraphernalia—language statistics, maps, newspaper clippings, dictionaries—to New York City.

  By October 1 the organization that was to become known as the American Black Chamber was ensconced in the former town house of T. Suffern Tailer, a New York society man and political leader, at 3 East 38th Street. It stayed there little more than a year, however, before moving to new quarters in a four-story brownstone at 141 East 37th Street, just east of Lexington Avenue. It occupied half of the ornate, divided structure, whose high ceilings did little to relieve the claustrophobic construction of its twelve-foot-wide rooms. Yardley’s apartment was on the top floor. All external connection with the government was cut. Rent, heat, office supplies, light, Yardley’s salary of $7,500 a year, and the salaries of his staff were paid from secret funds. Though the office was a branch of the Military Intelligence Division, War Department payments did not begin until June 30, 1921.

  Among the twenty people who started with Yardley or joined him soon thereafter were Dr. Charles J. Mendelsohn from MI-8, a philologist who taught history at City College mornings and worked in the Black Chamber afternoons; Victor Weiskopf, also from MI-8, a former agent of and cryptanalyst for the Justice Department, which allowed him to join Yardley’s organization in New York but paid him $200 a month to solve ciphers for it on the side; Livesey, who had been with Yardley in Paris, a Harvard graduate and businessman who later became a State Department economic advisor; Ruth Willson and Edna Ramsaier (who was to become Yardley’s second wife), both specialists in Japanese ciphers; and John Meeth, Yardley’s chief clerk. Livesey, who became Yardley’s prime assistant, was paid $3,000, or about $60 a week.

  One of the organization’s first assignments was to solve the codes of Japan, with whom friction had been growing. Yardley, in an access of enthusiasm, promised the solution or his resignation within the year. He regretted his impetuousness as soon as he plunged into the task, for he almost foundered in the Oriental intricacies of Japanese plaintext, to say nothing of codetext. After some preliminary study, assisted by Livesey, who had a great aptitude for languages, he ascertained that the Japanese employed a watered-down form of their ideographic writing called “kata kana” for telegraphic and—presumably—cryptologic communication, which was transmitted in Latin letters. Kata kana consists of about 73 syllables,
each with a sign of its own which had been given a roman equivalent, and when Yardley had his typists compile frequency tables for the twenty-five plain-language kata kana telegrams he had, he discovered that this script obeyed rules of frequency just like any other. Specifically, the kana n, the only nonsyllabic kana, was most common, appearing often at the end of words, followed by i, no, o, ni, shi, wa, ru, and to, in that order. The list of most common syllables and words began with ari and continued with aritashi, daijin, denpoo, gai, gyoo, and so on. At the end of about four months, the typists had prepared elaborate statistics of frequency and contact for about 10,000 kana.

  He then set them to work dividing the ten-letter groups of the Japanese code telegrams into pairs of letters and drawing up similar frequency and contact data for these pairs. He himself went through the approximately 100 code telegrams underlining with colored pencils all repetitions of four letters or more. But despite the most intensive scrutiny and study, no solution was forthcoming. Livesey’s linguistic abilities had meanwhile brought him a fair acquaintance with Japanese. He found in a bilingual dictionary that he had bought for 75 cents that the word owari meant “conclusion.” Could it be the plaintext of certain codegroups found frequently at the end of telegrams? The hypothesis, involving only three kana, proved barren. He examined the plain-language telegrams and pointed out probable words with conspicuous patterns to Yardley. Two of these, which played a vital role in the solution, were “Airurando dokuritsu” (“Ireland independence”), with the repeated do, and “Doitsu” (“Germany”), which used three of the same kana in a different order. This was a good clue, but it alone was not the answer. Night after night Yardley would climb the stairs to his apartment, weary, hopeless, discouraged, and fall into bed, only to wake up excitedly a few hours later with a brilliant idea—which invariably turned out to be just another blind alley.