Breaking codes was this couple's lifetime career
By James R. Chiles, Smithsonian, June 1987
William and Elizebeth Friedman were roped into an odd profession by
a wealthy eccentric and became America's premier cryptologists
Toward the end of World War I, the British Army began manufacturing thousands of small cipher machines, "Pletts Cryptographs", for use by the Allied forces. The British asked the American forces to use them as well. No one in the French, British, or American military had been able to break the ciphers; the machine had a mechanism that regularly altered the ciphering scheme, so the first a might be turned into an f and the next a into an r.
Just to be sure that it was safe from enemy codebreakers, the American military passed it on to a remarkable husband-and-wife team in Illinois for testing. William and Elizebeth Friedman received a package with five telegram-length messages. It took them all of three hours to break the lot, after which they returned them to London, solved. The first of the five messages read: "This cipher is absolutely indecipherable."
Few ciphers were ever indecipherable to the Friedmans. By the end of his life, William Friedman was recognized as the greatest maker and breaker of secret messages in history--the Harry Houdini of codes and ciphers. Repeatedly, he accepted challenges to solve "unbreakable" ciphers, and succeeded. The papers he wrote brought cryptology, an ancient skill as obscure as witchcraft, into the scientific age. The team he trained and supervised broke into Japan's highest diplomatic cipher just before World War II; not only did the group penetrate the secret, it built a deciphering machine that worked as well as Japan's cipher machine. Elizebeth Friedman provided exceptional assistance to the U.S. Coast Guard and Navy, unraveling secret messages from rumrunners during Prohibition, narcotics traffickers during the 1930s and enemy agents during wartime. True to the shadowy world of intelligence work, the pair shunned publicity and avoided discussing their work -- even with each other.
Neither had planned a career of codebreaking. In keeping with the strangeness of their profession, it all began at a peculiar place called the Riverbank Laboratories. The Riverbank estate, located on the Fox River in a small suburb west of Chicago, was the magnificent hobby of Col. George Fabyan, an eccentric millionaire who had retired from cotton trading early in the century. Here, on 600 well-kept acres, Fabyan's stable of hired scientists pursued whatever he found interesting or worthwhile, from sound waves to cryptology to plant genetics. In the evenings, they enjoyed the genteel life of the "minor idle rich," in Elizebeth's words. Other employees tended his greenhouses, livestock- breeding operation, Dutch windmill, Japanese garden and caged vegetarian bears. A monkey and other pets had free access to Fabyan's house, which was also notable for its furniture hung from the ceiling by chains.
Fabyan, a large and loud man with iron-gray hair who wore frayed formal clothes in the city and a sort of horseman's outfit in the country, "had great natural gifts of energy and dynamism," the Friedmans wrote later. "He also had the trick of parroting other people's jargon; his conversation was usually impressive-superficially, anyway."
In the spring of 1915, Fabyan went hunting for a geneticist to improve crop strains. A professor at Cornell suggested William Friedman, a graduate student in genetics. Friedman, who had been born to a Russian Jewish family in 1891 and brought to America as an infant, was preparing for a future in biology. When Fabyan offered him the job, Friedman asked him what he raised on his estate. "I raise hell," Fabyan replied. Friedman signed on at a hundred dollars a month as head of Riverbank's Department of Genetics and began work the following September, taking up quarters in an upper floor of the windmill. In a short time Fabyan had him planting oats by the light of the moon, to see whether the phases made any difference.
The other half of the Friedman team, Elizebeth Smith, graduated from Hillsdale College in Michigan in 1915 with an English degree, and a year later sought work in Chicago. At the Newberry Library, a staff member mentioned to her that a George Fabyan was looking for someone to study Shakespeare and offered to call him at his Chicago office. Fabyan arrived in his chauffeured limousine and insisted that Elizebeth come to the estate immediately. "He was the kind of man who did not take 'no' for an answer," she would reminisce. That evening she met Elizabeth Wells Gallup, a woman who had convinced Fabyan that Francis Bacon was the real author of Shakespeare's plays and sonnets, and that Bacon had hidden a number of secret messages inside the original printed copies. Elizebeth agreed to assist Mrs. Gallup in this unusual project.
And so began the Friedmans' long involvement in the esoteric world of ciphers and codes. For most of the rest of us, here's the difference between the two: in a typical cipher (such as the Pletts Cryptograph solved by the Friedmans in 1918) each letter of the original message is changed into another letter or symbol according to some orderly scheme. To decipher the message, it helps to know that e is the most frequently used letter in English. With additional knowledge of the language and guesses based on the context of the emerging message, the rest of the alphabet follows.
Most ciphers, however, are vastly more complicated than a single substitution: e will be represented by different symbols at different points in the message. A basic way to accomplish this is to assign each letter of the alphabet a number: a is 1, b is 2, and so on. Then we add a key, a regularly repeating group of characters or words superimposed onto the message. Say we want to send the word "trouble," and our key is "fast." It works like this:
20(T) 18(R) 15(O) 21(U) 2(B) 12(L) 5(E)
6(F) 1(A) 19(S) 20(T) 6(F) I(A) 19(S)
26 19 34 41 8 13 24
... and you have a cipher that your nosiest neighbor probably couldn't break. Your cousin on the other end will know the key, subtract it, and then distill the message. But it wouldn't even make lunchtime recreation for a cryptologist, who would aim a barrage of statistics at it and solve it in minutes.
A code is different. A typical code takes ideas in the message--words or even whole phrases--and changes them into something else, usually groups of numbers taken out of a codebook resembling a small dictionary. To compose a coded message, the sender first writes what he intends to say ("Attack is imminent") and then looks up the words in his codebook. "Attack" is 1140, and "imminent" is 4539. And that's his message: 1140 4539. To further confound the enemy, the sender might scramble the signal by using a key to encipher those numbers, by switching around the order, or both. The receiver must have an identical codebook and know anything extra the sender has done to scramble the number groups.
Codes and ciphers go back many centuries. Ancient Greek commanders scrambled messages by wrapping a strip of parchment or leather around a tapered staff in a tight spiral and writing their message down the length of the staff; unwound, the strip displayed a jumble of illegible letter fragments. A few centuries later, Julius Caesar wrote to associates in a simple cipher that took each letter of the message and substituted the letter three places farther down in the alphabet. Islamic people were the first to figure out how to attack ciphers in an organized way. During the Renaissance many European countries began setting up full-time codebreaking bureaus, called black chambers, to try to read each other's messages.
In 1625 Francis Bacon contributed a new kind of cipher to this busy scene, one that used only the letters a and b to represent the entire alphabet. In the table he set out, aaaaa stood for A, aaaab for B, aaaba for C, and so on. He called it the biliteral alphabet, and described how it could be disguised in an ordinary paragraph by using two different typefaces: one typeface representing a, and the other b.
Which brings us to Elizabeth Wells Gallup's Bacon-as-Shakespeare theory. It rested on two piers: Bacon had indeed invented the biliteral cipher, and the original printed folios of Shakespeare's plays employed an odd variety of typefaces.
Mrs. Gallup believed herself able to take printed lines from the plays and identify each character as either an a or a b typeface. Consulting the table that Bacon had published in 1625 she transformed each group of five a's and b's into a letter of the alphabet. The revelations she concocted were voluminous-one-fifth the length of the plays themselves--and sensational. Bacon, she claimed, not only wrote the plays but also was an illegitimate son of Queen Elizabeth I and the rightful heir to the throne. She convinced Fabyan to spend thousands of dollars digging holes around London, hunting for buried manuscripts. (Decades later, the Friedmans presented evidence that demolished the hidden-message theory. The peculiar typefaces were due to the economical habits of English printers, who preferred to repair old type rather than replace it.)
William Friedman spent time on the Bacon project as a photographer, taking close-up pictures of typefaces. Then, over the winter of 1916-17, with the approach of war, he turned over his genetics work to an assistant and joined Elizebeth in studying everything they could find on secret writings. Not much was available: the American literature, for example, consisted of a couple of short works by military men, and two articles and a story by Edgar Allan Poe. "We had a lot of pioneering to do." Elizebeth wrote later. "Literary ciphers may give you the swing of the thing, but they are in no sense scientific. There were no precedents for us to follow." In May 1917 the two married.
Meanwhile, Fabyan was bringing in coded diplomatic messages from unfriendly powers; he had offered his services free to the government, which had accepted. For almost a year, until the creation of the Army's Cipher Bureau, Riverbank was the only organization in the country capable of working out secret messages.
Among the most interesting problems to arrive at the large cottage where the Riverbank codebreakers worked were the cases involving the Hindu plotters. The first case was delivered personally by an official of Scotland Yard, who carried an attaché case packed with dozens of intercepted letters. For several years the Germans had been encouraging the aspirations of independence held by Indian citizens, on the theory that any trouble Indian radicals could stir up would siphon British strength from the war. Some of these radicals lived in the United States, and they had been passing cryptic messages around regarding arms shipments and internal politics.
One of the Hindu systems the couple broke into was a code that used numbers to indicate words and letters from a book that both sender and receiver had available. The cryptic number groups came in clusters of three, and the middle section always contained a 1 or a 2; that indicated a two-column page, probably a dictionary. The frequent use of some number groups indicated certain common words like "the" and "of." The first break was 99-2-14 in the code --page 99, the second column, the 14th word-- which they concluded from context was "you." That led to 99-2-17, which the couple guessed was near enough in the dictionary to be "your." Eventually they managed to make out nearly every word without having the actual dictionary at hand. Still, it was important to somehow obtain the dictionary for the upcoming trial of 155 Hindus in San Francisco. The Riverbank group polled large numbers of booksellers without results. Finally, having arrived in San Francisco to testify at the trial; William went into a university bookstore in Berkeley. He described what he was seeking (which must have been quite interesting in itself), and an employee dug around and produced the second volume of a German-English dictionary published in 1880. A quick check of the pages showed that this was, indeed, the book.
By late 1917 the Army had created its own Cipher Bureau, and the flow of intercepts to Riverbank ended. Fabyan determined to stay in the center of the action, arranged for the Friedmans to conduct classes in cryptography for Army officers.
Soon after the classes ended, William accepted a first lieutenant's commission in the Army and left for France. He spent the last five months of the war with General Pershing's staff, concentrating on breaking into the German codebooks. It was enough time to learn some important things about how armies employed codes and ciphers in wartime, and what could go wrong. For example, he observed that the German messages often began and ended with ritualized phrases, and that when a new code or cipher was introduced, the enemy's clerks were most likely to make a serious error (called a "bust" in the trade), exposing the system.
The following spring, William and Elizebeth returned to Riverbank. There William returned to the studies he had started before he left for Europe, which Fabyan issued as individual volumes in his Riverbank Publications series. In 1920 Friedman finished No. 22, titled The Index of Coincidence and its Applications in Cryptography. This booklet, says David Kahn, author of The Codebreakers, "must be regarded as the most important single publication in cryptology. It took the science into a new world." Friedman went beyond simple counts of letter frequency to discover techniques for applying statistical methods to cryptanalysis.
If you take any two lines of English text and place one above the other, so that each letter lines up with the one above or below it, there will occasionally be places where the same two letters appear in a vertical column. Friedman found that, for English, this coincidence consistently occurs in 6.67 columns out of every 100, or about seven percent of the time. He also found that this rate, the Index of Coincidence, is unique for each language. Such information can be very useful to a cryptanalyst with several enciphered messages from the same source. Usually these ciphers would be created using the same key, but the sender will have enciphered each message by starting at a different place in the key, according to some rule previously arranged between him and the receiver. Using the Index of Coincidence, the cryptanalyst can place the messages one above the other and, by sliding them back and forth, find the right "fit"-an indication that the key in the top message is vertically aligned with the same key in all the messages below it. Once the messages are aligned according to their key, he can work from message to message knowing that each letter in a vertical column has been enciphered with the same key letter as all the other letters in that column.
Shortly after this breakthrough the Army offered to employ the Friedmans as civilian "code experts" for a six-month trial period. With a few wary glances backward at Fabyan, wondering whether his tenacious grasp would reach to Washington, they left Riverbank for good and started military contract work in Washington, D.C., early in 1921.
In 1922, the War Department hired William permanently as chief codebreaker. Over the next few years he solved several "unbeatable" ciphers produced by machines. One of these, developed by Edward Hebern, foretold the hellishly complicated cipher machines that would dominate the next two decades. It used five movable rotors to scramble electrical signals between a keyboard and a set of glow lamps showing the enciphered letters. William had to determine the ciphering scheme, and then reconstruct much of the internal wiring. "He was discouraged to the point of blackout," Elizebeth said later. "Probably he sat for six weeks before he thought of a way to attack it.... It was all resolved in the end by the Index of Coincidence."
The Hebern solution was particularly important, says Louis Kruh, co-author of Machine Cryptography and Modern Cryptanalysis, because that machine "introduced a whole new concept of encipherment. It was the forerunner of any decent cipher machine between then and beyond World War II. Some of the report that he wrote on the solution is still classified today."
The latter part of that decade was exciting for Elizebeth. Her codebreaking began in earnest in 1927 as a "special agent" on loan from the Department of Justice to the Coast Guard, which was struggling to enforce the Volstead Act against a flood of smuggled shipborne liquor from the Bahamas, Canada and elsewhere.
The rumrunners' system required radio communications for rendezvous points, warnings and prices. It was this encoded and enciphered radio traffic that Elizebeth attacked. The early messages were elementary and quick to break. But as time went on and profits climbed, larger and more sophisticated syndicates took over the distribution network, and coding systems became more and more elaborate. In the first three years she and her assistants solved 12,000 messages using dozens of different schemes. In 1954 her decoding, and the detective work it inspired, cleared up a major diplomatic problem between the United States and Canada dating to 1929. She proved from old telegrams that the I'm Alone, a Canadian-flag schooner shelled and sunk by the Coast Guard in the Gulf of Mexico, had been secretly owned by an American smuggling ring.
Elizebeth, now working for the Treasury Department, was a star witness in a large and expensive trial in New Orleans of top officials of a Canadian outfit that controlled most of the liquor smuggling via the Gulf of Mexico and the Pacific. She deciphered and decoded cryptic radio messages that made prosecution possible.
The Friedman family, which by now included a son and a daughter, was aware that their mother's contact with organized crime had its dangers. Recalls daughter Barbara Atchison, "I remember Dad jesting once, when Mother was late getting home, that she might have been taken for a ride." And, in fact, while the I'm Alone case was under way, agents of the Treasury Department kept her under constant protection. Elizebeth broke open the code-ciphers used by narcotics smugglers. Her contributions to several sensational trials, and the associated convictions, earned her much unwanted attention from the press: "U.S. Woman Helps Smash Drug Ring," and "Key Woman of the T-Men," read two headlines.
War came and her work changed again. She devised a code system for William (Wild Bill) Donovan as he was organizing the Office of Strategic Services, and deciphered messages from German spies in Allied lands.
All this time William had been equally busy, but his work was well hidden from public view. In 1950 his responsibilities changed abruptly from the bookish work of composing codes. It was prompted by the closing of the Black Chamber, a cooperative codebreaking effort of the State Department and the Army that had evolved from the Army's old Cipher Bureau. The Black Chamber had decoded messages between Tokyo and the Japanese embassy, and in early 1929 Herbert Hoover's new Secretary of State, Henry Stimson, had removed his department's funding upon hearing about it. Years later, in his autobiography, he justified his action by saying that diplomats should not engage in such activities because "gentlemen do not read each other's mail." The Army quietly transferred the activities from the State Department to a new unit in the War Department, the Signal Intelligence Service (SIS). Friedman took charge as chief civilian cryptanalyst.
One of William's principal tasks was to train a new generation of experts to make and break codes, because all of the Black Chamber's personnel had dispersed. He started in the spring of 1930 with four spaces. He selected Frank Rowlett, a teacher from a small town in Virginia; Solomon Kullback and Abraham Sinkov, mathematicians from New York City; and John Hurt, the nephew of a congressman and an expert in Japanese.
The new team combed the files of the Black Chamber for useful information on Japanese cryptology. They read Friedman's booklets and worked through problems he concocted. He gave them assignments such as solving the Hebern cipher machine. "His teaching was such that we developed on our own," recalls Abraham Sinkov. "He just looked in from time to time to see how we were doing."
They operated on slim budgets--bringing their own pencils and writing on the back of old weather reports to save paper--but within two years they had progressed enough for Friedman to present them with the real thing: encrypted radio communications between Japanese diplomats.
By this time the Japanese knew that America had been reading secret diplomatic radio traffic. To protect its highest-level communications, Japan quickly switched to the new electromechanical cipher machine, "Angooki Taipu A." or "Red" to the Americans. It used a combination of wheels, rings and a rotor to scramble clear messages into a complex cipher. In 1935 the Army ordered Friedman's group to concentrate on breaking the system; by 1936 they were able to read the messages.
The passion for cryptology spilled over into the Friedmans' home life. William created a cipher game for children called "The Game of Secrecy" and tried it out on his son, John. He printed holiday greeting cards with cipher messages.
The most inventive of these pastimes were the progressive dinner parties hosted by the Friedmans. They divided their guests--a unique blend of codebreakers, newspapermen and scientists--into teams, and held the first course at one restaurant. "While they were eating, the restaurant owner gave them a piece of paper containing a clue about the next place to go," recalls John Friedman. "They'd go to five or six restaurants.... The first team to return home, won a prize."
William amused his children with enigmatic behavior. "He would come down to Sunday breakfast with six different ties on," remembers his daughter, Barbara. "He had a terrific sense of humor and whimsy, and women absolutely adored him."
Friedman had to leave most of his social life behind, however, when the Japanese decided to replace the Red machine with something much more secure. Labeled "Purple" by the SIS, the machine first contributed messages to Japan's radio circuits in February 1939. Unlike its predecessors, Purple used a type of switch identical to those used in automatic telephone exchanges to accomplish the main work of encipherment.
While dozens of others did the detailed statistical analysis necessary, William labored alone in his office on portions of the problem. He worked obsessively. The fear of failure--of what the Japanese were up to behind the screen of Purple--followed him home each night. "He'd be up until 2 or 3 in the morning," Elizebeth recalled later. "Sometimes I'd awaken and find him down in the kitchen making a Dagwood sandwich in the middle of the night." She remembered that her husband never discussed his difficulties with Purple, not even on the day in the fall of 1940 when the deciphering first succeeded. (For years, all he would tell his son, John, about his job was that he "worked for the Army.")
The team spent a total of 18 months puzzling out the mechanism and the nest of wiring from their calculations son how the original machine converted plaintext to cipher. Then, out of $684.65 worth of parts, they went ahead and built their own reconstruction of the Purple machine.
The Purple disclosures (codenamed "Magic") proved extremely valuable during the war. Ironically, says David Kahn, Magic "had its greatest effect on the war against Germany, not Japan, because we were getting a great deal of information that the Japanese ambassador in Berlin was sending ii back to Tokyo." For example, the ambassador, Baron Oshima, described Germany's new jet fighters in detail, and he listed German troop strength in the Balkans. Oshima even cabled a close description of the Normandy defenses in late 1943. Unfortunately, Magic did not clearly reveal the Japanese intention to attack Pearl Harbor --signals only indicated an impending crisis between America and Japan.
Codebreaking definitely changed the course of the war. The Allies made three major breakthroughs in all: the Japanese diplomatic cipher, Purple: the main German military cipher called "Enigma," which was broken by the Polish and British; and the Japanese fleet code, broken by the U.S. Navy, which made possible many American victories including the Battle of Midway. What made the breaking of Purple so extraordinary, says author Louis Kruh, is that the SIS had no pieces of the machine to study. When the team studied genuine Purple components after the war, he adds, they found that, out of all the thousands of soldered connections, "only two wiring connections turned out to have been interchanged."
Due largely to overwork on Purple, William suffered a mental breakdown at the turn of the year, and spent the next three months recovering in Walter Reed Hospital. Though the hospital recommended he return to duty, shortly afterward the Army discharged him from service. He served out the war as a civilian, organizing attacks on new cipher systems and rapidly training more cipher personnel.
"The war years were horrendous," recalls his daughter. "He was very, very involved in his work. I remember being worried to death about Dad." Attacks of depression severe enough to require psychiatric treatment or hospitalization would return again.
Wartime associates remember William as demanding and brilliant. "When working on a problem," recalls Frank Rowlett, "he was inclined to develop a plan of attack which he usually followed meticulously until the problem was either solved or it could be established that the attack would not be successful. His memory was excellent and he had a most unusual ability to grasp the complexity of a problem." Lambros Callimahos, a concert flutist who became a codebreaker in 1942, remembered Friedman as the "dapper figure with the Adolphe Menjou mustache, the characteristic bow tie, and the two-tone black-and-white shoes." Friedman was notoriously meticulous, even to the point of keeping carbon copies of handwritten notes. Judy Friedman, his daughter-in-law, remembers her he first weekend visit with the family. John had forewarned her that if his father failed to criticize her language, that meant she hadn't made the grade. Throughout the weekend, there was not a single comment from William about her speech. Judy was crushed. Then, upon leaving the house, he leaned close and whispered, "My dear, you made five grammatical errors this weekend," and proceeded to elaborate on them. Nothing could have pleased her more.
After the war, the Army restored William's rank and he served as a top ranking cryptanalyst in the various agencies leading up to the present National Security Agency (NSA). In 1956 Congress awarded him $100,000, partly in compensation for many of his inventions that were so sensitive and useful that the military had kept them off the market. One of these, a mechanism he had developed with Frank Rowlett, was the basis of the SIGABA machine, which the U.S. military used as its top-level enciphering device during World War II. The Axis never penetrated its messages.
William retired officially from the NSA in 1955 but continued to undertake various special and highly secret missions. His relations with the NSA were stormy at times; he believed the agency sometimes mistreated him and that it snooped excessively into citizens' communications. The NSA, in turn, occasionally saw him as a security risk: in 1958 three NSA employees appeared at his Capitol Hill house and carried off a stack of papers.
Elizebeth finished her professional career as a consultant for the International Monetary Fund, setting up a secure communications system there.
Toward the end of their careers (William died in 1969 and Elizebeth in 1980), the couple returned to the unlikely project that had first seduced them into the world of cryptography -the Shakespeare problem. Their incisive book, titled The Shakespearean Ciphers Examined, won them a Folger Shakespeare Library award in 1955. And in a classic demonstration of their life's work, Elizebeth and William Friedman included a hidden biliteral cipher on page 257 of the book. Buried in an italicized phrase, using two different typefaces, is their final verdict about the whole controversy: "I did not write the plays. E Bacon."
The Codebreakers: The Story of Secret Writing
by David Kahn, Macmillan, 1967
The Shakespearean Ciphers Examined
by William F. and Elizebeth S. Friedman,
Cambridge University Press, 1957
The Man Who Broke Purple
by Ronald Clark, Little, Brown, 1977
Machine Cryptography and Modern Cryptanalysis
by Cipher A. Deavours and Louis Kruh,
Artech House (Dedham, Massachusetts), 1985
The Index of Coincidence and Its Applications in Cryptography
by William F. Friedman, (Riverbank publications, Vol. 9),
Aegean Part Press (Laguna Hills, California), 1979
The Code Book
by Simon Singh