Historical Document · 1981
Rowlett's own account (from his 1980 memoir material): hiring 1 Apr 1930, Kullback spotting "The Japanese Government" by letter pattern, breaking RED via "oyobi", the six-buster, and Grotjan's 20 Sept 1940 breakthrough at about 2 p.m. (18 pp)
UNCLASSIFIED Frank B. Rowlett 4 UNCLASSIFIED UNCLASSIFIED Theodore ·M. Hannah -A Personal Profile Finally, Genevieve Grotjan finds the evidence we are looking for. Time is about 2 p.m. Ferner, Small, and I are at table in working area discussing prospects and reviewing work. Grotjan enters room, obviously excited, politely interrupts, asks if she can show us what she has found. She takes us to her desk in next room, lays out worksheets, points to one example, then another, then a third. She stands back, with eyes tranced behind her rimless glasses. Al Small dashes around the room, hands clasped above his head like a victorious prizefighter. "Whoopee!" he yells. Ferner, the quiet one, clasps his hands, shouting "Hooray, Hooray." I jump up and down - "That's it! That's it! " The room gets crowded; everyone in section suddenly in room. Friedman comes in and asks, "What's all the noise about?" I settle down and say, "Look what Miss Grotjan has just discovered." Gene wipes her eyes, tries to regain her composure. I point to the worksheets - "Gene's found= what we've been looking for. Look here, and here, and here." Friedman examines each one and understands what he sees; he looks suddenly tired. We take a break and send out for Cokes. The excitement gradually wears off and we look ahead to the neit step. ber 1940 represented the decisive breakthrough into a highly sophisticated machine system that for 18 months had stubbornly resisted the concerted attack by some of the best cryptanalysts in the world. Its solution has been called "the greatest feat of cryptanalysis the world had yet known," one which, in the words of another author, "involved a unique intellectual effort of heroic proportions." With its solution the United States became an unlisted subscriber to the communications network serving the Japanese Foreign Ministry and its major embassies abroad, and thereby gained important insight into Japan's diplomatic and military moves. The decrypts were also a primary source of information on Hitler's intentions in Europe, as reported to Tokyo by the Japanese ambassador in Berlin, Baron Oshima. Present at the breakthrough into Purple, and indeed largely responsible for it because he was the cryptanalyst in charge of the Army's work on the Japanese diplomatic intercepts, was Frank Byron Rowlett, one of the very early members of the Signal Intelligence Service, the Army predecessor of NSA. This is the story of his early career, beginning more than 50 years ago. That next step would soon lead to the solution of Japan's highest level diplomatic cipher system, the so- called Type "B" machine, better known as "Purple." What young, studious Genevieve Grotjan had discovered on that warm Friday afternoon in Septem- It. was a few minutes before eight o'clock in the morning on the first of April-All Fool's Day-1930. I was standing in front of the main entrance to the Munitions Building at 21st and B Streets, N. W., in Washington, D.C. In my hand was a letter telling me UNCLASSIFIED 5 UNCLASSIFIED to report at 8 a. m. to Room 3406 in that building. I was to be employed as a Junior Cryptanalyst in the Signal Service at Large, Office of the Chief Signal Officer, War Department. I was not quite 22 years old. I had passed the Civil Service examination for Junior Mathematician the previous summer and had been hired as a Junior Cryptanalyst at $2,000 per year, which was about $500 more than my wife and I together were making as teachers in southern Virginia. I had no idea what a cryptanalyst did; I didn't even know what the word meant-I couldn't find it in the dictionary-but there was a lot of publicity at the time about bringing back the World War dead from Europe, and I naively thought I'd be doing some kind of statistical work on crypts! It wasn't until my first day at work that I learned, from William F. Friedman, Chief of the Signal Intelligence Service, what a cryptanalyst was. "I suspect," he said, "that there are only a very few people who know what the word 'cryptanalyst' means. In fact, the words 'cryptanalyst' and 'cryptanalysis' were officially adopted by the War Department only a few weeks ago, although I started using them several years ago. The first official use of these words was in the description of the duties of the staff of the Signal Intelligence Service, and you have the honor of being the first individual to be employed by the U.S. Government as a Junior Cryptanalyst." Before my arrival, the Signal Intelligence Service (SIS} consisted of Mr. Friedman; his secretary, Annie Louise Newkirk; and Captain Norman Lee Baldwin, who was setting up the Second Signal Service Company, the collection arm of the SIS. By the end of the month we grew by another 50 percent when two more Junior Cryptanalysts, friends · from high school_ and college, Abraham Sinkov and Solomon Kullback, started work. Sinkov, Kull back, and I were hired because we had all studied math and foreign languages-Sinkov had taken French; Kullback, Spanish; and I, German. Friedman's efforts to round out the team with a combination mathematician and Japanese linguist had so far failed: the Civil Service Commission, Military Intelligence, and the State Department had all been unable to produce anyone with that rather unusual combination of talents. Soon, however, through the good offices of Congressman Schaeffer of Virginia, Friedman learned of a young man named John Hurt, the congressman's nephew, who, though lacking an extensive math background, was unusually fluent in Japanese. The Army major who interviewed him told Friedman that he had never met an American so proficient in that difficult language. "He is unbeliev- 6 UNCLASSIFIED able," he said. "He is fluent in conversation, he can read both forms· of the written language, and his vocabulary is fabulous. He knows Japanese much better than I or, for that matter, anyone else in G-2." We learned later that Hurt had mastered the language mainly by rooming with Japanese students at college, where he had majored in Latin and French. While Hurt was processing for employment, the rest of us continued our cryptologic training under Friedman, using mainly his pamphlets and exercise books. As soon as Hurt arrived, Friedman wanted him to begin teaching us the rudiments of Japanese, for Japan was to be our number one priority. Two very important things happened to me toward the end of that first month: I got my first paycheck and I rented an apartment. My pay for the first two weeks of work (we were paid twice a month) was the grand sum of $83.33, only about $8.00 less than I had received for a full month's work as a schoolteacher in Franklin County, Virginia. The apartment was in the old Boulevard Apartment building at 2121 New York Avenue, N. W., within easy walking distance of the Munitions Building; if I hurried, I could have lunch at home instead of having to eat at the Navy cafeteria, in which I was by now beginning to lose interest. The apartment was the barest minimum offered for living purposes in the area, consisting of a single room (with a so-called Murphy bed which folded up into the wall when not in use) and a small kitchenette and bathroom. It rented for $50. 00 per month furnished, with a free month's rent if it was occupied for a full year. I figured that my wife and I could make do with the one-room apartment for at least a couple of months until we could find something a bit larger, but still within our means. It was not a good year for the American economy. The Great Depression was deepening, and military budgets, already small during those peacetime years, were shrinking even more. So austere was the Army budget that we sometimes had to supply our own pencils and paper. We bought penny pencils at the dime store, and we brought in scratch paper and tablets from home. We once had to take a temporary reduction in salary; another time, when funds ran out completely, we all had to take a month's leave without pay. My wife and I just closed up our apartment and took a springtime vacation on the farm in southern Virginia, so for us it was no particular hardship. The Black Chamber Revisited One hot, humid morning in late June 1930, Friedman came into the vault and asked Kully, Abe, and UNCLASSIFIED me to accompany him to the G-2 area of the Munitions Building. He escorted us down the stairs to the second floor corridor without any explanation of the purpose of our trip. I clearly remember that I was wearing, for the first time, a brand-new summer outfit - a blue serge jacket, white pin-striped trousers, and white suede shoes. I recall that as we walked along the second floor corridor behind Friedman I was feeling very proud of myself for being so well dressed for what, judging by Friedman's attitude, seemed to be a very special mission. When we arrived at the cross corridor of the seventh wing, Friedman swung to the left without missing a step. We followed closely behind. The cross corridor was deserted. After a few steps he turned abruptly to his right and stopped at Room 27 42. He took from his inside coat pocket a small card, and started to operate the combination lock on the front of the steel door. We watched with high curiosity. When he had set the final number, he threw the bolt and swung open the door, exposing the second steel door behind it. He again reached into his pocket, extracted a key, and unlocked the inner door, which he opened with a flourish. The area behind the second door was pitch dark. He now produced a small box of matches, lit one, and by its light sought out the pull cord of the switch of the nearest ceiling light. When the light came on, it revealed a room about twenty-five feet square, almost completely filled with filing cabinets, arranged back-to-back in double rows with barely enough space between them for the drawers to open. Friedman's next move was to switch on the electric fans to their highest speed and to turn on the remaining three ceiling lights. While he was going through this process, the three of us remained just outside the door, waiting for his instructions. He came back to the vault door, peered up and down the corridor, and then waved us inside the vault. When we were all inside, he set the inner steel panels slightly ajar and turned to us, saying in a solemn and very imposing manner: "Welcome, gentlemen, to the secret archives of the American Black Chamber!" I was puzzled by the serious manner he assumed as he made the announcement. I had never heard of The Signal Intelligence Service about 1935. Seated: Mrs. Louise Newkirk Nelson. Standing, left to right: Herrick F. Bearce; Solomon Kullback; Captain Harold G. Miller, USA; William F. Friedman; Abraham Sinkov; Lieutenant L. T. Jones, USCG; Frank B. Rowlett. John B. Hurt was ill when the picture was taken. UNCLASSIFIED 7 UNCLASSIFIED anything called the American Black Chamber and the words meant nothing special to me, but I said nothing and neither did Abe or Kully; we just stood, looking as impressed as we could, surveying the dust-covered contents of the room. Friedman then explained that the room contained the working files of a highly secret cryptanalytic organization that had operated in New York City for several years and which had been closed down during the previous summer. Military Intelligence had had an interest in this operation and had brought its files to the Munitions Building, where they were being held in storage. The files had been turned over to the Chief Signal Officer for the use of the Signal Intelligence Service, and it was our job to organize and catalog them. He suggested that we spend the next couple of hours looking over the files, familiarizing ourselves with their contents. He pointed out that as the first order of business we should try to locate all the information they contained about German Army field ciphers and Japanese diplomatic codes and ciphers. After saying that he would return by lunchtime, he left, carefully setting the steel doors so that the contents of the room were not visible from the outside corridor. After Friedman left the room, Abe drew his index finger across the top of the nearest file cabinet. When he held up his finger and noted the thick smudge of dust on it, he grimaced. Besides the file cabinets, the only pieces of furniture in the room were a single oak chair and a small table on which were arranged some pencils and ruled pads. I took off my new blue jacket and hung it on the chair. After Abe and Kully had disposed of their jackets, Kully turned to Abe and, with an impish grin on his face, pointed to my new and immaculate white trousers. Then they both laughed. We now started to look at the materials in the file cabinets. In a few minutes we forgot all about the thick dust that covered everything in the room, and we forgot about the summer heat and the poor ventilation. We were completely hypnotized by what we were finding. Here were the secret records dealing with the American code-breaking activity sponsored by the State Department and the Director of Military Intelligence; here were copies of the secret codes and ciphers, both purloined and cryptanalytically recovered, of many of the great nations of the world; here were the worksheets used in breaking Japanese diplomatic codes; here were the translations of Japanese messages relating to the negotiations of the Washington Naval Conference, to which were attached letters of appreciation signed by high State Depart- 8 UNCLASSIFIED ment officials; here were the decipherments of the German field ciphers of 1917 and 1918, with descriptions of how they were broken; here were hundreds of copies of unsolved code messages from every important capital of the world; and here was a wealth of other cryptologic material which could be appreciated only after hours and hours of further detailed study. King Solomon's mines could have offered no greater treasures for us. We were so intrigued by the contents of the vault that we lost all track of time. Friedman's return startled us. He found three sweaty and grimy but nonetheless starry-eyed Junior Cryptanalysts. Kully had a dirty streak across his forehead. Abe's shirt was spotted with smudges of dust and soaked with perspiration. My brand-new white pin-striped trousers were no longer white; they were a dirty mess. Friedman looked us over and, after remarking on "what a dirty business cryptanalysis is," suggested that we clean up and go to lunch. He asked us to spend some time at his desk after we had finished lunch, stating that we would be joined by Hurt for an important discussion about our future duties. Our Mission Described Early that afternoon the four of us met with Friedman in his office. He settled us in chairs around his desk and carefully closed the office doors. For me, it was the most electric moment that I had yet experienced, and all sorts of romantic ideas were racing through my mind. Friedman began by describing the work of what he called "Yardley's bureau," a joint intelligence activity conducted by the State Department and War Department G-2, with the State Department supplying most of the money. Herbert 0. Yardley had conducted the activity in New York City under the guise of a legitimate enterprise known as the "Code Compiling Company." In the summer of 1929, when the new Secretary of State, Henry Stimson, learned of the nature of Yardley's activities, he had summarily abolished "Yardley's bureau." He ordered that the organization be closed out in the shortest possible time and said that no such activity would be tolerated inside the Department while he was Secretary. G-2, viewing Stimson's ·order as a major disaster to the American intelligence effort, had tried to have his order reversed. But Stimson was adamant and G-2 finally had to accept his decision. When the Director of Military Intelligence realized that Stimson meant what he said, he proposed to the Chief Signal Officer that a code-breaking operation be set up in the War Department. He offered to supply UNCLASSIFIED the funds previously contributed to the State Department in support of Yardley's operation. The money would be used to hire a small group of young men who would be trained in cryptology, and who hopefully would become the cadre of an effective cryptanalytic organization. The Chief Signal Officer had enthusiastically received the proposal and had accepted the responsibility for administering the organization. The funds made available by G-2 to the Chief Signal Officer were barely sufficient to provide for the employment of four Junior Cryptanalysts and one Cryptographic Clerk. The four of us represented the first step in the implementation of the long-range plan to develop a new cryptologic capability in the War Department. Friedman then went on to explain that the longterm mission of the Signal Intelligence Service was twofold. The first and more important objective was to insure that U.S. military codes and ciphers were the best and most secure that could be devised. The second was to undertake the peacetime interception and analysis of foreign code and cipher communications for the production of intelligence for the country's top planners and policymakers and, in the event of war, to be prepared to intercept and solve enemy messages at all levels. He also disclosed to us that an intercept service was then in the process of being organized and that it was being staffed by the best radio operators in the Signal Corps. Some had already been assigned to the intercept organization and were being trained for intercept work. He added that the first of a worldwide network of intercept stations was then being established at Fort Monmouth, New Jersey. He impressed on us the need for secrecy, pointing out that both the Director of Military Intelligence and the Chief Signal Officer were afraid that if the State Department became aware that a new cryptanalytic activity was being established, steps might be taken at the highest level of government to have the organization abolished. He explained that the organization had been administratively located in the Office of the Chief Signal Officer rather than G-2, for if its existence did become known to the Secretary of State, its continuation could be justified as being part of the Chief Signal Officer's responsibilities to design, compile, store, and issue all cryptographic materials required by the War Department, and to supervise the use of all cryptographic systems in the Army. Friedman said that for the next several months the four of us would spend part of our time studying cryptology with the aim of mastering all that was currently known about cryptography and cryptanalysis; the remainder of our time would be devoted to producing improved codes and ciphers for the War Department. He emphasized his conviction that all currently used U.S. codes and ciphers were "woefully inadequate, outdated, inefficient and, worst of all, insecure." He added that it was imperative that we get started as soon as possible on improving all cryptographic systems used in the War Department, so that at least the Army would have adequate cryptographic security in the event of another war. Our Codes Are an Open Book A fortuitous combination of circumstances soon allowed me to see firsthand how insecure the War Department's codes really were. The Signal Corps had recently become responsible for all phases of War Department cryptographic security, which until then had been shared with the Adjutant General's Office. One result of the realignment-and this was my first lucky break-was that there developed a temporary shortage of trained code clerks. It was decided that Kullback, Sinkov, and I would be detailed to work in the code room for a few weeks until new clerks could be trained. The second lucky break was that I could type, and since messages had to be typed before they cleared the code room, I was assigned there more often than ' the others. What I learned during those long night shifts was both rewarding and discouragingrewarding because I gained a better understanding of military communications and procedures; discouraging because I learned how woefully weak was the cryptographic security of those communications. Most messages, regardless of their classification (Restricted, Confidential, or Secret), were encoded from the same codebook, the old War Department Telegraph Code of about 50,000 groups, and this of course was an inherently insecure procedure. The military attache network used a smaller, more sophisticated code of about 10,000 groups; messages sent in this system were always enciphered. Unclassified messages went out in the clear when transmitted over the War Department radio network. When sent over commercial circuits-RCA or Western Union, for example-they were encoded, not for security but for economy, since the government had to pay by the group, and encoded messages were always shorter, hence cheaper, than the plain-language originals. Friedman assured me that as soon as appropriations came through, he intended to produce new codes and eventually to develop automatic cipher machines for use in Washington and the overseas departments. "In the next few years, Mr. Rowlett," he said, "we are going to revolutionize the security of governmental UNCLASSIFIED 9 UNCLASSIFIED Frank Rowlett on . .. What's Important in Cryptology In my experience most people have assumed that the production of intelligence is the primary purpose of the cryptologic effort. I disagree. It's important but it's not paramount. The most important goal of the cryptanalyst is to use his knowledge of foreign cryptography to improve our own cryptographic security. The Effect of The American Black Chamber I think Yardley's book was a good thing for U.S. cryptology because it forced the Japanese to replace their cryptosystems with better, more secure systems and that in turn forced us to improve our cryptanalytic capability. Moreover, were it not for Yardley's book, the Japanese might not have introduced their new systems in time for us to master them before communications, and you 4nd the other young men you are associated with are going to be deeply involved in these developments." Growth and Training By the end of 1930, the staff of the Signal Intelligence Service had grown to eight. The two newcomers were Lieutenant Mark Rhoads of the Signal Corps, the first Regular Army officer to be assigned full time to the SIS, and Larry Clark, a chemistry major attending evening classes at George Washington University, who was hired, as an Assistant Cryptographic Clerk, to work on secret inks, one of our several responsibilities at the time. Meanwhile, Sinkov, Kullback, Hurt, and I continued our cryptologic training under Friedman. Early in 1931 we began studying machine cipher systems, starting with a modified version of the Wheatstone cipher device. That led us to the Bazeries multiplex alphabet system, which we solved with relative ease. Next was the German Kryha machine, said to be indecipherable because of the enormous number of possibilities17x26! x26!, as I recall-that would have to be tested before text could be recovered. According to the manufacturer's brochure, it was humanly impossible to make all these tests, thus the cipher could not be 10 UNCLASSIFIED war came. I don't condone disclosures of sensitive information, but in this case it had a beneficial effect. Personnel Security The rules on security. clearances were pretty loose in the early days. A~ long as you were a secondgeneration American you were assumed to be loyal and you were clear_ed. If you were first-generation, you might be cleared. If you weren't born in America, you couldn't be cleared. In my own case, I guess it was assumed that a mountain boy from the heart of southwest Virginia would be unlikely to have any allegiances except to the United States. I didn't get formally investigated and cleared until the war was half over, or some 12 or 13 years after coming to work. broken. Nothing daunted, we pressed on and solved the system in fairly short order. Next on our list was the B-211, an electromechanical device invented by the Swedish engineer Arvid Gerhard Damm. It was a beautifully made machine, about the size of a portable typewriter. Cryptographically, however, it wasn't particularly formidable, being nothing more than a fractionating system easily duplicated on paper. To satisfy ourselves that we fully understood the functioning of the device, we enciphered a short message using both the machine and our paper analog. We were delighted to find that the cipher texts were identical. Our training in machine cipher systems continued with the study of the Vemam device, a sophisticated machine developed by AT&T. Probably the first fully automatic cryptographic device actually constructed, it had been successfully tested by the Signal Corps on wireline circuits. But it had one great weakness, which Frie~man had earlier exploited and which he now wanted us to discover. As I recall, once that weaknes8 was found and the underlying principle understood, solving the machine was much like completing a word puzzle, and the solution came fairly easily. The most difficult machine we studied was the Hebem cipher machine, developed in the early twenties for the Navy. Friedman had found it to be cryptographically weak, and it was redesigned. Our UNCLASSIFIED job was to duplicate his original ~alysis, using the same test messages provided him by the Navy. Although we mastered the cryptographic principles involved, we were unable to solve the system until Friedman explained the underlying statistical concepts, whereupon we recovered the message texts and the wiring of the electrical commutating elementsloosely called cipher wheels-used in the encipherment. Open for Business In 1932, Friedman announced that we were ready to begin our intelligence mission, with Japanese diplomatic communications being our first priority. But it was not to be a full-time effort, for we were still responsible for producing codes for the War Department, and this took precedence over signals intelligence. We began by reviewing the Japanese material inherited from Yardley's Black Chamber operation in New York. Considerable amounts of live traffic soon began to arrive from the Army's nascent intercept service. This was augmented by the do-it-yourself intercept activity of Colonel Joseph Mauborgne, Signal Officer for the Ninth Corps area. Mauborgne (later General Mauborgne, head of the Army Signal Corps) set up an unofficial intercept station in the basement of his quarters in the Presidio of San Francisco, from which he copied Japanese diplomatic traffic on the San Francisco-Tokyo circuit. He mailed us his undulator tapes and Kullback and I transcribed them. Since I knew Morse code, I read the characters to Kully, who, in order to save paper, wrote out the messages on the backs of old Signal Corps weather reports. Our approach was to sort incoming intercepts into homogeneous groups-an easy matter since the Japanese used a unique identifier on each message, evidently for the convenience of the code clerk-and then test each code against the Black Chamber's solutions. With the help of John Hurt's linguistic talents, we soon produced our first current translation of a Japanese diplomatic message, this one in a modified form of the so-called LA code. While at the time not holding much significance for us, the achievement did cause the Chief Signal Officer and the Director of Military Intelligence to take renewed interest in our work. It also lent new stimulus to the effort to enlarge the staff of the Signal Intelligence Service, since Friedman very convincingly made the point that we could either compile codes or solve Japanese diplomatic systems, but we could not be expected to do both. Frank Rowlett at Arlington Hall Station in 1948. UNCLASSIFIED 11 UNCLASSIFIED In September 1932, Sinkov, Kullback, and I began spending a few hours each week studying Japanese under John Hurt. We used the Rose-Innes text for our study of conversational Japanese and the standard national Japanese elementary language textbooks (Kokugo Tokuhon) for the written language. The latter were in written Japanese (kana and Chinese characters), and so far as I was concerned were much more useful for our cryptanalytic purposes than was the Rose-Innes text. Our lessons continued until the following May, when Hurt was striken with tuberculosis and had to be hospitalized. His replacement was a Mr. Ayvozoglou, a former colonel in the Imperial Russian Army who bad studied Japanese at the Oriental Institute in Moscow. Unfortunately for us, his knowledge of spoken English was so poor and his pronunciation so atrocious that he sometimes had to resort to using .the French or German equivalent of the English term to enable us to understand him. He had to leave after three months because of a lack of funds to pay him, but by that time we were learning more from code recovery than from the language classes, for the messages gave us the vocabulary actually used by Japanese diplomats. When Hurt returned to work, Friedman urged that he concentrate on Japanese, leaving the strictly cryptologic work to the rest of us; this pleased Hurt, for he much preferred spending his time on language. Kully's Discovery One afternoon the three of us, Kully, Hurt, and I, were quietly working at our places in the vault when, with no warning whatsoever, Kully started to beat the top of the table at which he was working and loudly exclaimed, "That's it! That's what they have done! It's an English monoalphabet!" . I had no idea what he was eo excited about. I hurriedly got out of my chair and went to look over his shoulder. He started to wave a worksheet wildly through the air. "That's what they are doing,'' he exclaimed. "They're enciphering English texts with a monoalphabetic substitution. And all along we were looking for something else." "Show me,'' I said, as Hurt joined us. Kully slammed the worksheet down on the table. "Look at that," he said, pointing to three of the cipher groups he had underlined in red. "Look at the letter patterns for those three words. Can they be anything but a monoalphabetic substitution of the English words 'The Japanese Government'?" 12 UNCLASSIFIED "Now look at the next words," he continued. "Decipher them by using the equivalents from the other three words, and see what you get." "I'll tell you what it is," he said. "The rest of the line of text will decipher to give 'requests that the Government of the United States.' " "Why don't you write in the equivalents on the worksheet?" I asked. "Then you can check out the remainder of the message." Kully rapidly filled in the assumed plain text while Hurt and I looked on. Every letter he entered fitted accurately. It was indeed a monoalphabetic substitution, with vowels replacing vowels and consonants replacing consonants. He continued his decipherment of the remaining groups and in a short time had finished the entire message. There was no doubt about it-the Japanese were using the most insecure type of cipher possible for enciphering this message. "I can't understand it,'' Kully remarked. "Why would they do anything so stupid?" "Why don't you teat another message?" Hurt asked. Kully soon located another intercept with the same indicator. As Hurt and I looked on, he deciphered a couple of groups using the substitution alphabet recovered from the message he had just deciphered. "That's not English," I observed. "But it's perfect Japanese!" Hurt exclaimed. "Decipher some more groups so that I can translate it." Kully continued his decipherment of the message until Hurt interrupted him. He took the worksheet from Kully and studied it for a few seconds. "Except for a couple of garbled letters, this is perfect Japanese. It's apparently a news release which the Japanese Foreign Office is transmitting to its installations abroad for the information of Japanese nationals in foreign countries. I've seen similar examples in the plain-language messages we've been intercepting." Kully sought out some more messages with the same indicator. All could be deciphered and all were in Japanese. It was really a stroke of luck that Kully had spotted the only intercept with English plain text. Otherwise, it might have been wee~ before we realized that this particular category of traffic was indeed a monoalphabetic substitution, for we were looking for something much more sophisticated. "Let's see what Mr. Friedman has to say," suggested Hurt. I stepped to the vault door and opened it. "Excuse me,'' I said. "Kully has discovered something that you will find most interesting. Would you mind stepping into the vault to look at it?" UNCLASSIFIED Arlington Hall Station, 1942 or 1943. Seated: A. J. McGrail, W. Preston Corderman, William F. Friedman. Standing: Mark W. Rhoads, Solomon Kullback, John B. Hurt, Edward J. Vogel, Frank B. Rowlett, Abraham Sinkov. Friedman pushed his work aside and followed me into the vault. "What have you found?" he asked Kully. "The Japanese are using a monoalphabet for enciphering some of their diplomatic messages," Kully answered. Friedman looked doubtful. "Why do you think that?" he asked. "I just solved one," Kully answered gleefully. "Well, that's a rather convincing argument," Friedman replied. "But there must be something wrong. No government with any competence whatever in cryptography would intentionally use such a patently insecure system. A code clerk must have made a mistake, or they are using a cipher machine in which a mechanical failure occurred, thereby producing a monoalphabetic substitution encipherment which was transmitted by mistake." "But there are several messages on different dates which use the same substitution," Kully answered. He then spread out his worksheets and demonstrated the decipherments. Friedman picked up the worksheet of the message in English and read it carefully. "Thie message seems to be a note which the Japanese ambassador has been instructed to deliver to the State Department," he said. "It is possible that they used a simple encipherment merely to disguise the contents of the message from the communications personnel who would be handling it. If the Japanese were planning to provide the exact text of one of their messages to another government, they certainly wouldn't want to encipher it in one of their secret UNCLASSIFIED 13 UNCLASSIFIED codes. To do so might result in their code being compromised, for it would take an exceptionally secure cryptographic system to resist analysis when such a long message together with 'its exact decipherment were available for study." Friedman was obviously pleased with Kully's discovery. "Well done, Mr. Kullback," he said as he left the vault. The Red Machine The mid-1930s saw us continuing our efforts to solve Japanese diplomatic cryptosystems, particularly any that might be carrying traffic relating to a suspected secret codicil to an Axis pact then under discussion. Our search led us to a system used on two important networks-the Far Eastern diplomatic network and a net linking Tokyo with major world capitals, including Berlin and Rome. The system was characterized by five-letter text preceded by a five-digit group. Friedman, who by now was taking a strong interest in our examination of the unsolved intercepts, thought the Japanese might be using a cipher machine, and suggested that the fivedigit group served not only to identify the system but also to indicate the specific keys employed for enciphering the message. With the help of our IBM accounting machines we tested the texts for repetitions. We found none we considered significant, but in the process we discovered that six of the cipher letters displayed a significantly higher or lower frequency count than the other 20, that the frequency pattern changed at 10-day intervals, and that some messages used the same initial five-digit group. Our statistical tests also showed that when two messages in the same 10-day period used the same five-digit groups, the messages were enciphered with the same keys. The job of breaking the system fell to Kully and me, for in 1936 Sinkov, the only bachelor among us, was assigned to the Panama Canal Zone, the site of the first Army intercept station outside the United States. While Kully and I studied the traffic from the various 10-day periods, Friedman told his opposite number in the Navy, Commander Joseph N. Wenger, about our work on the system and the progress we had made so far. Wenger said that the Navy had recently broken into a Japanese naval system which used an electromechanical cipher machine and the kana alphabet, but beyond that he would not go. I should explain that, while there was a complete exchange of intercepted Japanese diplomatic traffic between ourselves and the Navy, there was at that time no direct contact between us except for an 14 UNCLASSIFIED occasional meeting between Friedman and Wenger. Though he didn't say so, I sensed that Friedman felt he was being unfairly treated by the Navy by being denied technical information, whereas the Army, through Friedman, had always shared information fully with the Navy. He made an appointment with Wenger for the next day with the intention of requesting complete details about the machine the Navy had solved. Despite Friedman's argument that the cryptographic principles employed in the Japanese naval machine could well be similar to those used in the diplomatic machine, Wenger still declined to divulge any more information, or even to confirm or deny that there were similarities. He would say only that the kana sequence had been broken into two subsets, but we already knew that. When pressed, he suggested that we search for a basic cycle of "more than 40 but less than 50'' elements, and that was all Friedman could get out of him. Our test for a cycle of the length suggested by Wenger yielded nothing. Somewhat dejected, we admitted that for the time being we had run out of useful ideas, and we decided to consolidate our materials and organize the files while considering how best to attack the system. Later, after an exhaustive examination of the best intercepts, Kully and I succeeded in deciphering certain parts of an exceptionally long message in which the vowel and consonant patterns corresponded to those expected in romanized Japanese plain text. When we found a recurrence of Analog or the Japanese Red Machine, made by Frank Rowlett sometime around 1936. UNCLASSIFIED the plaintext word oyo- i in two of the passages we had tentatively deciphered, we knew we were into the system, for oyobi is the Japanese word for "and." That led us to review what we knew about the system. There were two sequences, one for vowels, the other for consonants. These sequences apparently provided for independent encipherment of the vowels and consonants: a vowel would always be enciphered by a vowel and a consonant by a consonant. Evidently the two sequences operated synchronously, both advancing with the encipherment of each letter of the message, whether consonant or vowel. There was a motor key involved, which controlled the advancement of the two sequences, introducing additional steps at certain points in conformity with an overall pattern. Armed with that knowledge and some educated assumptions of the underlying plain language, we soon recovered enough text to know that we were well on our way to breaking the "Code Machine, Type A," the most important Japanese cryptographic system then in use. Friedman, of course, was elated, particularly so because we had succeeded in recovering an unknown cipher machine by reconstructing the principles solely from intercepted messages, without help from any outside source. As he congratulated us on our success, he added, "You know, g