THE CODEBREAKERS

by DAVID KAHN

  As the pressure of his duties declined, Friedman and his wife returned to the cryptologic field that had gotten them started—the Baconian ciphers. They summed up the experience of a lifetime in a long and exhaustive report that won them the Folger Shakespeare Library literary prize in 1955. After his retirement, they collaborated in preparing this work for 1957 publication by the Cambridge University Press as The Shakespearean Ciphers Examined. While, as The New York Times Book Review accurately said, they buried “these pseudocryptograms beneath a mass of evidence as crushing as an avalanche,” they also introduced their readers to a rogues’ gallery of pseudocryptologists not to be met elsewhere in literature. The Friedmans here display a rather surprising—surprising to one who has perused only his technical writing—wit and talent for personality sketches.

  In 1956, the 84th Congress voted to pay Friedman $100,000 in compensation for profits he had been unable to realize because security prevented him from marketing cipher machines that he had invented for the government. It marked the successful end of a battle that had begun six years earlier when his lawyers decided that under existing law he could not sue to recover his losses and that he must seek legislative relief. “The immeasurable stress of his work,” they stated in a memorandum asking the Defense Department not to oppose the measure, “and the burden of responsibility imposed by the necessity for constant secrecy ever since 1921 were major factors in the impairment of Mr. Friedman’s health which now makes his livelihood increasingly precarious. It is this last consideration which finally induced Mr. Friedman to permit us to bring the matter to the attention of the Department of Defense.”

  Involved were nine inventions made from 1933 to 1944, two with Rowlett’s aid, though the bill was not limited to them. Two were so secret that no patent applications had ever been filed. Four are held in secrecy in the Patent Office: three of these pertained to the Converter M-134-C, a rotor machine, and one to the Converter M-228. Three have issued as patents: a strip form of the Jefferson cylinder; the Converter M-325, another rotor machine; and a facsimile enciphering system.

  “Procurement by the United States of devices constructed in accordance with the principles of Mr. Friedman’s inventions has approximated $10 million,” the Secretary of the Army wrote to Congress in support of Friedman’s case in 1953, “most of which occurred during the active phase of World War II, and has involved the use of substantially all his inventions…. Under the circumstances of his employment, it appears that the Government has at least a nonexclusive license in Mr. Friedman’s inventions, Mr. Friedman retaining the right to otherwise exploit them. Because of security considerations, however, Mr. Friedman has been prevented from attempting to derive any gain from his inventions commercially or from foreign governments.”

  The legal question was fearfully confused, but the Secretary, Robert T. Stevens of Army-McCarthy-hearings fame, felt that Friedman deserved equitable redress—in the sum, however, of only $25,000. The following year, he changed his mind and agreed that $100,000 “would not constitute more than adequate compensation.” This followed a reappraisal by N.S.A. Director General Canine, who observed that a large market existed among foreign governments for cipher machines and that the excellence of Friedman’s inventions would have given him an important competitive advantage. The Bureau of the Budget questioned the award on the ground that it was inconsistent with government policy on secret inventions made by federal employees. It also put its finger on what appeared to be one of the chief motives for the award: Friedman’s outstanding achievements. His lawyers were always careful to found their claims on the alleged financial loss—but they never failed to cite Friedman’s record.

  Two bills for Friedman’s relief had died in committee during these prolonged negotiations. Finally, a hearing was held on the third bill before the Senate subcommittee on patents, trademarks, and copyrights. It was brief, mainly because Senator Joseph C. O’Mahoney, who was presiding, was in a great hurry to get to the Senate floor. The witnesses—notably Friedman’s attorney and the lawyer for Swedish cipher-machine manufacturer Boris C. W. Hagelin, who had become a millionaire when Friedman had ordered his machines for the U.S. Army in World War II—discoursed eloquently on the glowing opportunities of commercial cryptography. The subcommittee approved the bill; Congress passed it; President Eisenhower signed it on May 10, 1956, and Friedman got his $100,000.

  It must be stated that justice was not served thereby. In the case of the seven inventions that were filed in the patent office, at least five were derivative—mere improvements upon the basic creations of others. Such were the rotor machines, compensation for which should have gone to the estate of Edward Hebern, and the strip device, recompense for which should have been paid either to Jefferson’s estate or to Parker Hitt, who first conceived the principle in strip form. (The other two were quite probably derivative as well.) Hebern’s lawyer, in fact, tried to make this point at the hearing, but O’Mahoney cut him short. The Friedman award more rightfully belonged to others; it went to him because of well-situated friends, picayune mechanical differences, and a great but totally irrelevant record.*

  This blot dims but little the luster of Friedman’s escutcheon. He has ranged over more cryptologic territory than anyone else, and has mined it more deeply. Some of this was due to the accident of time and circumstance, which were more propitious for cryptology then than before or since. The era of radio had opened; mechanization had begun to transform cryptography; armies were becoming more mobile and larger and increasingly dependent upon control by communications; the United States had emerged as a major power, and politics were global. These currents gathered momentum, culminating in the Second World War; after a brief respite, the cold war renewed them. Friedman was lucky enough to come to maturity as this surge was swelling, and smart enough to see and seize the opportunity it presented. Yet environment alone does not explain the magnitude of his achievements; none of his contemporaries approached them.

  His theoretical studies, which revolutionized the science, were matched by his actual solutions, which astounded it. Both are complemented by his peripheral contributions. He straightened out the tangled web of cipher systems and introduced a clarifying terminology for his arrangement. Words he coined gleam upon more than one page of today’s dictionaries. His textbooks have trained thousands. His historical articles have shed light in little-known corners of the study, and the Shakespeare book has done much to quash one major area of a perennial literary nuisance. Singlehandedly, he made his country preeminent in his field. And finally, the vast American cryptologic establishment of today, with its thousands of employees, its far-flung stations, its sprawling headquarters—this gigantic enterprise (except for the Navy branch started by Safford) is a direct lineal descendant of the little office in the War Department that Friedman started, all by himself.

  This life’s work, as extensive as it is intensive, confers upon William Frederick Friedman the mantle of the greatest cryptologist.

  * The President and his advisor were then using two main systems. One was external—a superencipherment applied to the five-digit numerical groups of what probably was a State Department code. The first digit was enciphered by one of two alternate letters; the two pairs by a vowel-consonant combination. Thus, in one edition of the superencipherment, 40606 became FEDED, 40699, KEDIR, and so on. The other was internal—a jargon code of such less-than-Stygian incognitos as MARS for the Secretary of War, NEPTUNE for the Secretary of the Navy, BLUEFIELDS for William C. Redfield, Secretary of Commerce, ALLEY for Franklin K. Lane, Secretary of the Interior, and MANSION for David F. Houston, Secretary of Agriculture. Yardley does not specify which he solved.

  * In addition to this and the Zimmermann telegram, two messages to the diplomat from his home office, encoded in the English-French half of Clifton’s Nouveau Dictionnaire Français, which had replaced the betrayed Cipher 13040, were solved by MI-8. They disclosed Germany trying to bribe Mexico to remain neutral.

  * In 1940,
as Secretary of War, he had to reverse himself and accept the cryptanalyses of MAGIC. But the international situation then was totally different. “In 1929,” he himself has written, in the third person, “the world was striving with good will for lasting peace, and in this effort all the nations were parties. Stimson, as Secretary of State, was dealing as a gentleman with the gentlemen sent as ambassadors and ministers from friendly nations….” In 1940, Europe was at war, and the United States was on the verge.

  * At different times.

  * The Greek letter kappa is frequently used in mathematics to designate a constant.

  * Poetic license. There are actually 11,881,376 permutations of the 26 letters in groups of five, far more than any code has ever used.

  * The same remarks apply, though in a more attenuated degree in all respects, to the Congressional awards of an identical $100,000 on essentially the same basis of equity to Safford in 1958 and to Rowlett in 1964.

  13

  SECRECY FOR SALE

  ON A MORNING in December of 1917, a rather handsome young man of 27 hurried through the colonnaded lobby of the American Telephone & Telegraph Company at 195 Broadway in downtown Manhattan. He rode the elevator up to the 17th floor, where he worked in the telegraph section of the company’s development and research department. This section, composed of some of the brightest engineers in the company, was concentrating on the newest development in telegraphy, the printing telegraph or teletypewriter.

  Gilbert S. Vernam was—if things were as usual—a little late that morning. He nearly always was, and, his boss said, “It used to burn me up to see him come sneaking in and, slink into his seat.” The yearbook of his alma mater, Worcester Polytechnic Institute, had wondered “what would happen to Tech if ‘Tau’ should accidently get to class on time in the morning.”

  A native of Brooklyn, Vernam was graduated from the Massachusetts college, where he had been president of the Wireless Association and had been elected to Tau Beta Pi, the engineering honorary society, in 1914, after having spent a year working. He immediately joined A. T. & T. and, a year later, married a Brooklyn girl, Alline L. Eno. They had one child. Vernam was a clever young man—one of the stories about him has him stretched on his couch each evening wondering aloud, “What can I invent now?” He had the rare type of mind that can visualize an electrical circuit and put it down on paper without having to try it out with wires. He did so well in the telegraph section that its head, Ralzemond D. Parker, assigned him to a special secrecy project. And late though he may have been that winter morning, Vernam had brought a bright idea to work with him. Quiet and unassuming, though with a droll sense of humor, he probably put forth his suggestion with diffidence, but his co-workers on the secrecy project saw at once that he had something.

  The project had begun during the summer, a few months after war had been declared, when Parker directed some of the telegraph section members to investigate the security of the printing telegraph. Would its very newness, the fact that the enemy might not have developed such means, guard its messages? The secrecy group soon found that it did not. The fluctuations of the current could be recorded by an oscillograph and the messages read with ease. Even multiplexing—sending several messages simultaneously in both directions over a single wire—offered no real security. The engineers resolved the oscillograph undulation into its constituent curves and read the eight individual messages. The group discussed altering connections inside the printing telegraph mechanism. This would have the effect of enciphering one letter into another in a monalphabetic substitution. The engineers realized that this offered no real secrecy but, stymied, did not pursue the matter until Vernam bounded in with his idea.

  It was based upon the Baudot code, the Morse code of the teletypewriter. In this code, named for its French inventor, J. M. E. Baudot, each character is allotted five units, or pulses. Each unit consists of either an electrical current or its absence in a given time. There are, consequently, 32 different combinations of marks and spaces, and a combination is assigned to each character—26 for the letters and one each for the six “stunts” (space between words, shift up to numbers and punctuation marks, shift back down to letters, return type-carriage to left side of paper, feed paper up a line, and idle). Through an electrical arrangement involving rotating commutators, the proper sequence of pulses is sent out when a character’s key is struck on the keyboard. For example, a is mark mark space space space, i is space mark mark space space and the figure shift is mark mark space mark mark. At the receiving end, the incoming pulses energize electromagnets that, in combination, select the proper character and print it. In the punched paper tape which is frequently used to run teletypewriters, marks are represented by holes and spaces by leaving the tape intact. To read the tape, metal fingers push through the holes to make contact and thereby send pulses; where there is a space, the paper keeps the fingers from completing the circuit.

  Vernam suggested punching a tape of key characters and electromechanically adding its pulses to those of the plaintext characters, the “sum” to constitute the ciphertext. The addition would have to be reversible so that the receiver could subtract the key pulses from the cipher pulses and get the plaintext. Vernam decided upon this rule: If the key and the plaintext pulses are both marks or both spaces, the ciphertext pulse will be a space. If the key pulse is a space, and the plaintext a mark, or vice versa—if, in other words, the two are different—the ciphertext pulse will be a mark. The four possibilities are these:

  plaintext key ciphertext

  mark + mark = space

  mark + space = mark

  space + mark = mark

  space + space = space

  Decipherment is unambiguous. For example, with ciphertext mark and key space only mark is possible for the plaintext. The whole system may be set out in a single, compact table. Using the convenient notation of 1 for mark and 0 for space, the rule would be tabulated as follows:

  In accordance with this rule, Vernam combined the five pulses of the plaintext character with the five of the key character to obtain the five pulses of the ciphertext character. Thus, if the plaintext is a, or 11000, and the key is 10011, which happens to be B, the encipherment is this:

  At the receiving end, the key pulses are applied one by one to the successive ciphertext pulses; the rule determines the plaintext pulses. With cipher pulses 10100, and the key pulses 00110, the plaintext would be:

  To combine the pulses electrically Vernam devised an arrangement of magnets, relays, and bus-bars. Since encipherment and decipherment were reciprocal, the same arrangement served for both. He fed the pulses into this device from two tape readers—one for a keytape, the other for the plaintext tape. The mechanism closed a circuit, resulting in a mark, when the two incoming pulses were different, and opened a circuit, resulting in a space, when they were the same. This output of marks and spaces could be transmitted just like an ordinary teletypewriter message to the receiver. Here the Vernam apparatus subtracted out the key pulses, which were supplied by an identical keytape, and recreated the original plaintext pulses. These it would channel into a teletypewriter receiver, which would print out the plaintext, just like a news ticker in a city room.

  That was the beauty of it. No longer did men have to encipher or decipher a message in a separate step (though they still had to prepare keytapes, insert them in the apparatus, etc., since doing away with these would dispense with secrecy altogether). Plaintext went in and plaintext came out, while anyone intercepting the message between the two endpoints would pick up nothing but a meaningless sequence of marks and spaces. Messages were enciphered, transmitted, received, and deciphered in a single operation—exactly as fast as a message in plain English. The advantage was not the mechanical enciphering and printing of the message. That had been accomplished as far back as the early 1870s by two Frenchmen, Émile Vinay and Joseph Gaussin—though not with the speed and ease of a typewriter keyboard. Rather it was the assimilation of encipherment into the overall com
munication process. Vernam created what came to be called “on-line encipherment” (because it was done directly on the open telegraph circuit) to distinguish it from the old, separate, off-line encipherment. He freed a fundamental process in cryptography from the shackles of time and error. He eliminated a human being—the cipher clerk—from the chain of communication. His great contribution was to bring to cryptography the automation that had benefited mankind so much in so many fields of endeavor.

  These values were immediately recognized, and Vernam’s idea quickly kicked up a flurry of activity. He put it down on paper in a sketch dated December 17. A. T. & T. notified the Navy, with which it had worked closely in a communications demonstration the previous year, and on February 18, 1918, Vernam, Parker, Lyman F. Morehouse, equipment engineer of the telephone company, and Edward Watson explained the Vernam system, together with some other possibilities, to a Lieutenant Griffiths. On March 27, the engineers conferred with colleagues of the Western Electric Company, A. T. & T.’s manufacturing subsidiary, and began constructing a couple of Vernam devices, using as many standard parts as possible. They hooked them up to two teletypewriters and, in the Western Electric laboratory, ran the first tests of what the engineers called “automatic cryptography.” The devices worked like a charm. A. T. & T. reported this to the Army. Major Joseph O. Mauborgne, then head of the Signal Corp’s research and engineering division, came, saw and was conquered. Except for the problem of the keys.