The Woman Who Smashed Codes

by Jason Fagone

  It’s best to do the counting in a systematic way. You might start by drawing a thing called a frequency table. You chop the cryptogram into its component parts and sort them according to the letters by which they’re surrounded. It looks like this:

  Though it may look like gibberish, it’s a powerful tool—“the Real Stuff,” in Elizebeth’s phrase—because with a quick glance down the columns, you can identify the most frequent letter groups in the cryptogram and the letters that come before and after them. Letters in a given language are like children in a kindergarten class; they have affinities, cluster in cliques. In the lunch line, one kid likes to walk behind a second kid and in front of a third kid while a fourth sits off in the corner, eating from a paper bag. What you’re really looking at when you look at this frequency table is a picture of “certain internal relations in the English alphabet,” as Elizebeth and William would put it. You’re looking at the structure of the underlying language itself.

  Now you have some grip on the puzzle. You can begin to peel back the skin of the message, to see familiar shapes in the strangeness. Like with a crossword puzzle, there’s no direct, guaranteed route to solving a cryptogram. The solver has to make educated guesses, plug in letters and see if they lead to recognizable words, backtrack and erase if a guess is wrong, try a new letter.

  Elizebeth quickly got the hang of it, plowing through messages and counting letters, although it felt completely new and weird to her, a totally different way of looking at language than what she was used to. All her life she had celebrated the improbable bigness of language, the long-lunged galaxy that exploded out from the small dense point of the alphabet, the twenty-six humble letters. In college she trained herself to hear the rhythms of playwrights and poets, the syllables that slip from the tongue in patterns. Tennyson:

  There lives more faith in honest doubt,

  Believe me, than in half the creeds.

  There LIVES more FAITH in HON-est DOUBT,

  Be-LIEVE me, than in HALF the CREEDS.

  But before, she had gone no further than chopping lines into meters. She left the words in their boxes, intact. Codebreaking required more drastic measures. Now Elizebeth had to shake the words until they spilled their letters. To rip, rupture, puncture, chisel, scissor, smash, and scoop up the rubble in her arms. To chip off flakes from the smooth rock of the message and place them in piles and ask questions about them. It involved a kind of hard-hearted analytic violence that she had never contemplated before. It was reaching into the red body of the text until the hands dripped with blood.

  Ahhhh!

  The first few messages she broke, real military messages, had been intercepted from the Mexican army. Like most military cipher messages, they were written in blocks of five letters, like TZYTV RGFQF MQFHC, in order to fuzz out the original lengths of the words and therefore make the messages harder for adversaries to break. Elizebeth counted the letters, drew her frequency tables, consulted materials on the frequencies of various letters and letter combinations in the Spanish language, scratched her guesses into the graph paper, and there, right there—she saw things that started to look like words. A lovely shape pried out of the murk, glistening.

  The process gave her a sensation of power that was electric and new and made her want to keep going. It was nothing like working on the biliteral ciphers with Mrs. Gallup. Here there was no mystery, no squinting through a looking glass at the curls of italic letters and trying to sort them into categories based on vague criteria never fully explained. Here the method was sharp and clear, a series of small and logical steps that built toward a goal. “The thrill of your life,” Elizebeth said later, describing how it felt to solve a message. “The skeletons of words leap out, and make you jump.”

  And she was never alone. That was the other thrill. She and William operated as a team. During the day they were never more than a few feet apart, handing papers back and forth, checking each other’s work, asking questions when stuck, keeping up a friendly patter, “calling out” letters on their sheets in the “word-equivalent” alphabet commonly used by the U.S. Army: Able for the letter A, Boy for B, Cast for C, Dock, Easy, Fox, George, Have, Item, Jig, King, Love. If Elizebeth needed to read the ciphertext FVGEQ, she would call out, “Fox! Vice! George! Easy! Quack!”

  It was demanding work. Each solution had to be checked. Errors corrected. A single miscopied letter could wreck hours of effort. You got tired and needed a friend to look at your page while you rubbed your eyes. Each learned to recognize signs of fatigue in the other and knew when to suggest that it was time for a break. The less you had to think about, the better and more accurate your work. Elizebeth and William used the same kind of pencil, the same kind of paper, and never deviated from these choices. They liked pencils with soft lead and big erasers, the eraser end seeing as much action as the lead end.

  Cast! Easy! Jig! King! Opal! They called out letters all day long like teachers taking attendance at a strange school. Pup! X-ray! Vice! Love! Sail! The pencils at Riverbank were plentiful and free, black with white erasers, and doubled as advertising tools; a cipher alphabet was printed on each pencil in white letters, along with RIVERBANK LABORATORIES—GENEVA, ILLINOIS.

  Mike, she called out, a smile playing at the corners of her mouth. Zed. Rush. Fox. Zed.

  Watch, he called out, grinning. Dock. Yoke. Pup. Easy.

  The paper they used was graph paper with a grid of quarter-inch squares. One letter per square. They never threw anything out. “Work sheets SHOULD NOT BE DESTROYED,” the pair would soon write in one of several scientific papers about their discoveries. Worksheets “form a necessary part of the record pertaining to the solution of the problem. No work is too insignificant to discard, therefore it should be done well from the start.”

  Tare. Yoke. George. George. Able.

  Unit. King. Nan. Zed. Boy.

  No one told them how to set up this workflow, and no one told them they had to collaborate. They simply found, by trial and error, that collaboration made things go faster, that “a group of two operators, working harmoniously as a unit, can accomplish more than four operators working singly. Different minds, centered on the same problem, will supplement and check each other; errors will be found quickly; interchange of ideas will bring results rapidly. In short, two minds, ‘with but a single thought,’ bring to bear upon a given subject that concentration of effort and facility of treatment which is not possible for one mind alone.”

  Although William and Elizebeth solved their first batches of military messages using techniques they learned from Hitt’s manual, they soon exhausted its teachings. Elizebeth filled the margins with her own notes in blue pen, comments about sentences she found imprecise, things Hitt had gotten slightly wrong, things he could have explained better or had left unexplored. (She underlined a sentence on page 85 and wrote next to it, “This is poorly expressed.”) She and William had reached the cordon of what was known, the edge of the map. From here on, they would need to invent new techniques—to become scientists, explorers, pushing into a wild land.

  One way of thinking about science is that it’s a check against the natural human tendency to see patterns that might not be there. It’s a way of knowing when a pattern is real and when it’s a trick of your mind. Elizebeth and William had begun at Riverbank by looking for the false patterns of Mrs. Gallup. But now, over the next several years, they found ways of seeing true patterns. It was as if they had been tossed into a raging river of delusion without knowing how to swim and figured out how to save themselves from drowning, clinging to each other the whole time. This struggle made them stronger than they could have ever imagined. They climbed out of the river transformed, with new powers, shaking the water from their backs, and then took off at speed, racing across the mountains and through the swamps of an undiscovered continent.

  Between 1917 and 1920, George Fabyan used Riverbank’s vanity press to publish eight pamphlets that described new kinds of codebreaking strategies. These were little
books with unassuming titles on plain white covers. Today they are considered to be the foundation stones of the modern science of cryptology. Known as the Riverbank Publications, they “rise up like a landmark in the history of cryptology,” writes the historian David Kahn. “Nearly all of them broke new ground, and mastery of the information they first set forth is still regarded as the prerequisite for a higher cryptologic education.”

  The eight Riverbank Publications are commonly attributed to William alone, with two exceptions. Inside his personal copy of one paper, Riverbank No. 21, Methods for the Reconstruction of Primary Alphabets, William wrote in black ink beneath the title, “By Elizebeth S. Friedman and William F. Friedman.” A second paper, Methods for the Solution of Running Key Ciphers, never included her name, but she and William always told colleagues it was a joint effort.

  However, there’s evidence that Elizebeth was involved with more than just the two papers. The original typewritten and hand-edited drafts of the Riverbank Publications are now held by the manuscript division of the New York Public Library, and her handwriting is all over them. William seems to have written a lot of the technical sections, with the drafts marked up by both of them, Elizebeth’s comments interspersed with his, while Elizebeth wrote and researched the historical sections, which he edited in a similar fashion.

  They worked as a team in most matters and the soon-to-be-legendary papers were no different. In a 1918 letter to Elizebeth, William referred to the early Riverbank Publications as “our pamphlets”—our, not my. And other Riverbank workers contributed as well: men and women, codebreakers and translators. The publications were “a piece of work that was done by the staff,” Elizebeth said later. “No one person was mentioned as the sole conqueror or anything like that. Everybody worked together.” This is as far as she ever went in claiming a piece of the credit. Today it’s hard to know exactly what she did, because she wanted it that way. “Mrs. Friedman had a tendency to see that the record made little or no mention of her contribution to a number of their joint efforts,” the custodian of the Friedmans’ personal papers wrote in 1981 to a researcher interested in Elizebeth. “And therefore it will be difficult to get a clear picture of her exact role.”

  Why hide her role? Partly it was expected at the time, that the man was the scientist and the woman the helpmate, but Riverbank was a bubble world where the usual rules didn’t apply. Fabyan had no trouble championing the work of women, as he proved with his zealous promotion of Mrs. Gallup. A more likely explanation is that Elizebeth was trying to help William win a battle with Fabyan over the copyrights of the Riverbank Publications. At first, Fabyan didn’t even let William place his name on the covers of the pamphlets, only on the inside pages, and Fabyan registered the copyright under his own name. He said he saw no ethical problem because he had paid for the research. “It may be egotism on my part,” Fabyan told William, “but so long as I pay the fiddler, I am going to have the privilege of selecting a few of the tunes.”

  It was hard enough for William—a credentialed scientist, a genetics Ph.D.—to get credit for the work. He and Elizebeth may have decided it would be doubly hard to convince Fabyan to share credit with her, too.

  Whatever the case, the Riverbank Publications, and the breakthroughs they describe, still seem incredible today. Seven of the eight pamphlets were written in the space of two years, in a little cottage in the middle of Illinois, the cryptologic equivalent of Albert Einstein’s annus mirabilis, when Einstein rewrote the language of light, mass, and time in the space of a single year, at age twenty-six, while working as a patent clerk in Switzerland, staring out the window of his office and bouncing ideas off a fellow clerk. This is the achievement that the NSA interviewer in 1976, Virginia Valaki, kept begging Elizebeth to explain: How? Elizebeth gave unsatisfying answers, noted in the transcript:

  “That World War I leapt on, and so many things happened so fast. . . .”

  “Nothing was ever as carefully executed as that. It was sort of on a day-to-day basis. You did what you could with what you had to do it with.”

  “I don’t think I remember offhand. I was too busy either getting on this swing or getting off that one. ((Laughs.))”

  “I feel no confidence whatever to speak on that point; wouldn’t have the faintest idea what to say.”

  The likely truth: it only looked improbable in retrospect. At the time they didn’t know what was supposed to be hard, and there was no one around to tell them. They didn’t see themselves inventing a new science. They were playing the game day-to-day as best they could, as Fabyan always said. They were just trying to solve messages as they poured in and not get stuck.

  The mail from Washington contained a frothy mix of messages from all over, a zoo of alphabets that had to be studied and classified. There were two main animal kingdoms of cipher, “transposition” and “substitution.” A transposition cipher was like Scrabble, a jumbling of the same letters into a new order. A substitution cipher was a swapping of letters. Each kingdom contained a diverse multitude of beasts that had to be tamed in different ways, and there was always a time crunch, someone demanding a quick answer. Invention under pressure.

  One day in early 1917 a heavyset man showed up at Riverbank on a mission all the way from Scotland Yard, the police headquarters in London. He had been referred to Riverbank by the U.S. Department of Justice. Fabyan barked an introduction at Elizebeth and William, and the detective opened a briefcase. Stacks of messages spilled out. He said the messages had been intercepted by British postal censors, and the recipients included as many as two hundred individuals in India, then a British colony. Scotland Yard suspected an attempt by Germany to spark a revolution among Hindu separatists, but no one knew for sure. All the detective knew were the names of a few of the suspects, which he told to Elizebeth and William.

  The young codebreakers looked at the messages and found them “quite baffling.” They were written in numbers, grouped together in shorter or longer blocks:

  38425 24736 47575 93826

  97-2-14

  35-1-17

  73-5-3

  82-4-3

  Elizebeth and William assumed from these groupings that the separatists were using three different codes. The blocks of five numbers looked like a simple type of codes based on a rectangular grid of letters:

  1

  2

  3

  4

  5

  6

  7

  1

  A

  B

  C

  D

  E

  F

  G

  2

  H

  I

  J

  K

  L

  M

  N

  3

  O

  P

  Q

  R

  S

  T

  U

  4

  V

  W

  X

  Y

  Z

  The grid turns a letter into a number (C is 13), which then has a number added to it, based on a prearranged key word. If the word is LAMP, the value of L (25) might be added to C (13), making 38. Elizebeth and William deduced the key word and solved the messages by analyzing frequent numbers and the intervals between them.

  Thinking about the second set of numbers (97-2-14), which were confined to a single long message in the detective’s trunk, Elizebeth and William noticed that the middle number was always either a 1 or a 2. This was a clue that the conspirators were using a specific, yet-unknown book to encrypt their messages and that the book had two columns of type, like a dictionary. The numbers likely pointed to words at certain locations in this mysterious book. For instance, in a sequence like 97-2-14, 97 meant the page number, 2 meant the right-hand column, and 14 meant the fourteenth word in the column. Applying similar logic to the third set of numbers in the detective’s messages (73-5-3, 82-4-3), the young codebreakers deduced that the numbers pointed
to individual letters within a different book possessed by the conspirators. The recipient of the letter, seeing the number 73-5-3, would turn to the seventy-third page of the book, go to the fifth line, and write down the third letter in the line.

  Of course, this wasn’t enough information to solve the messages, and the young codebreakers at Riverbank weren’t sure they could go further: If the letters and words had been selected from specific books, and Elizebeth and William didn’t know the names of the books or have copies of them, what was the use?

  “For a time,” William wrote, “it looked like an insurmountable task.”

  But they wrote down all the numbers in order, searched for repetitions, thought about it some more, and found a foothold. A Harvard professor had recently counted the words in a long English text, and the prairie codebreakers read his study. Of 100,000 total words, only 10,161 were unique, and just 10 words accounted for 26,677 of the 100,000: “the,” “of,” “and,” “to,” “a,” “in,” “that,” “it,” “is,” “I.” “You can’t convey much intelligence using only these words,” William wrote, “and yet you can’t construct a long, intelligible, unambiguous message without using them over and over again.”

  Turning to the numbers that seemed to come from a dictionary, the codebreakers reasoned that a word at the beginning of the alphabet, like “and,” would correspond to a code group beginning with a lower number (1, 2, 3, 4) than a word that came higher in the alphabet, like “the.” This insight helped them solve the most frequent words in the messages, and from there it was possible to work out others: If 97-2-14 was YOU, then 99-2-17 must be a word close to “you” in the dictionary, perhaps “your,” “young,” or “youth.” Elizebeth and William ended up solving 95 percent of the message in this manner, without ever seeing a copy of the conspirators’ dictionary. As for the last set of numbers, the ones that seemed to refer to letters in an unknown book, they used a similar process, matching frequent code groups with frequent letters and pairs of letters, and reverse-engineering the text of the book as they went. Whenever they discovered a new letter of plaintext, it told them more about the content of the book, and whenever they pieced together a new line in the book, it told them more about the plaintexts of the messages. One of the conspirators’ notes read, in part: