THE CODEBREAKERS

by DAVID KAHN

  Neither before nor since have criminals ever used such extensive and complicated systems of secret writing, and neither before nor since has so strong an effort been mounted against them. Mrs. Friedman fought in this arena as the champion of the law, at first singlehandedly and then, late in the game, as the leader of a small band. Appropriately enough, she crowned this work with solutions of messages that helped destroy one of the largest smuggling rings in the United States. While she had the satisfaction of a job well done, Woodcock’s letter must have come as a very welcome appreciation. It was probably not lost upon her that he had prosecuted in the name of the United States of America.

  In 1934, some of Mrs. Friedman’s earlier solutions helped to extricate the United States from an embarrassing diplomatic tangle and to establish a point of international law. Among the intercepts that she had solved during her trip to Houston in 1929 were 23 that had no connection with any case then under investigation by the United States Attorney there. Between October 2, 1928, and March 15, 1929, they had been exchanged between the cable address CARMELHA, which belonged to the well-known liquor import-export firm of C. A. Melhado in Belize, and the New York cable addresses HARFORAN and MOCANA, both of which were unregistered. The American authorities had intercepted them as they passed over the radio circuits of the Tropical Radio Company at New Orleans. A typical message was this:

  HBA69 6 Wireless—NS Belize BH 29 427P

  MOCANA

  NEW YORK

  YOJVY RYKIP PAHNY KOWAG JAJHA FYNIG IKUMV

  Mrs. Friedman had little difficulty in discovering that they had been encoded in Bentley’s Complete Phrase Code and that the placode group had been replaced by the codegroup five places forward in the codebook. This message’s plaintext then read: Arrived. Some repairs necessary. Will leave 2d February. Telegraph instructions. None of the messages carried any signature.

  On her way back to Washington, she stopped off at New Orleans and gave her solutions to Edson J. Shamhart, supervising customs agent there. Sham-hart practically leaped out of his chair with a whoop. He had been trying to help the State Department resolve the difficult case of the schooner I’m Alone. This handsome two-master had been built in Nova Scotia in 1924 for the liquor trade, which she had plied vigorously and profitably. In 1928 she was sold to new owners, but her mission remained the same. On March 20, 1929, when the Coast Guard cutter Wolcott, suspecting her of running liquor, ordered her to heave to, she refused. The ships’ positions then were given by the boatswain in command of Wolcott as 10.8 miles from shore, well within the 12-mile territorial limit, and by the captain of I’m Alone as 15 miles from shore. A chase began. Wolcott fired a warning shot, then turned her guns on the schooner itself, damaging her sails and slightly wounding the captain. Shoal water caused Wolcott to fall back, however, and it was not until March 23 that she and another cutter, Dexter’, again caught up with I’m Alone. This time, at a point 220 miles from shore and well out of American jurisdiction, Dexter poured shells into I’m Alone and sank her with her Canadian flag waving.

  Relations between the United States and Canada were already somewhat strained over the liquor question, since the exporting of liquor was entirely legal in Canada, and the I’m Alone sinking exacerbated feelings. A member of the Canadian Parliament even claimed that the attack was an act of war if it had been carried out under official instructions. Canada demanded $386,000 for loss of the vessel and its cargo. This claim was based on the presumption that the vessel was Canadian-owned. American officials contended that the ship had been challenged in territorial waters, that hot and continuous pursuit had ensued, and that international admiralty custom sanctioned her sinking. In addition, they strongly suspected that the schooner belonged to Americans, in which case the affront of sinking a Canadian flag could be adjusted between friendly powers by a formal apology and a token cash indemnity.

  Unfortunately, they had little success in tracing the owners—until Mrs. Friedman walked into Shamhart’s office with her translations of messages to and from New York. For the dates of arrival and orders to sail and quantities of alcohol listed in the messages agreed perfectly with the sailing dates of I’m Alone from Belize and with the quantities of Scotch, rye, and malt in her manifests. Some detective work discovered that the cable addresses MOCANA and HARFORAN belonged to one Dan Hogan, a New York bootlegger and half-owner of I’m Alone, who was arrested and convicted. In the winter of 1934-35, a Canadian and an American justice of the two countries’ Supreme Courts sat as a court of arbitration in the I’m Alone case. They agreed that the doctrine of hot pursuit—the rule that lets a policeman who begins chasing a speeder in his own state arrest him in another—held in international law, settling this moot point for the first time. They granted Canada $50,666 for the flag insult (the United States also formally apologized), but ruled against the Canadian claim for restitution to the owners of I’m Alone. At the time they decided, among the items of evidence before them were the 23 solutions of Elizebeth Smith Friedman.

  The great Teapot Dome oil scandal of the 1920s revealed corruption and rottenness in the dealings of some of the highest ranking and wealthiest men in America. The list began with the Secretary of the Interior, Albert B. Fall, who had leased out for development the rich Teapot Dome naval oil reserve in Wyoming to Harry F. Sinclair and the Elk Hills reserve to Edward C. Doheny of the Pan American Petroleum Company. This he did in secret on the ground of national security.

  When this came to light early in 1924, the Senate Public Lands Committee began an investigation. It soon obtained an admission from Edward B. McLean, editor-president of the Washington Post, that he had lent Fall $100,000 for which he had neither canceled checks nor stubs nor any kind of receipt. McLean denied knowing Sinclair or anything at all about the oil reserves matter. But then the committee dug up a number of messages in code, mostly to and from McLean, one to Doheny. It turned them over to the Army for an attempt at solution. On March 4, 1924, the committee heard sworn testimony concerning their readings from the chief of the code section in the Signal Corps, William F. Friedman. His task had not been too difficult, because he had access to all three of the codes used. The Doheny message used a private code of the Pan American Petroleum Company; the company furnished a decode of the message to the committee, who gave it to Friedman, who merely checked it. The McLean messages were encoded either in Bentley’s Complete Phrase Code, used straight, or, surprisingly, in a code of the Department of Justice’s Bureau of Investigation (forerunner of the F.B.I. ). It gave rise to such messages as the one to McLean on January 9, 1924, whose odd-sounding language tickled the fancy of many newspaper readers:

  ZEV HOCUSING IMAGERY COMMENSAL ABAD OPAQUE HOSIER LECTIONARY STOP CLOT PRATTLER LAMB JAGUAR ROVED TIMEPIECE NUDITY STOP HOCUSING LECTIONARY CHINCHILLA PETERNET BEDAGGLED RIP RALE OVERSHADE QUAKE STOP….

  The Justice Department code partook of the characteristics of both a one-part and a two-part code. Like the German diplomatic code 13040, which figured in the Zimmermann telegram, it was a hybrid. Thus, in the encoding section, while the plaintext ran alphabetically from a to z, the initial letters of the codewords ran from R to W, from N to Q, and from A to M. The breaks occurred in the middle of the plaintext e’s and i’s. Thus eight was WIPPEN and end was NAUTCH, interrogation was QUAKE (appropriately enough) and is was ACERBATE. The January 9 message thus read: Zevely thinks trend of investigation favorable to you. Not impressed with Walsh as cross-examiner. Thinks you need have little apprehension about forthcoming interrogation….

  Friedman read the McLean messages by “means of a code book which I obtained at the Department of Justice from Mr. [William J.] Burns,” director of the Bureau of Investigation. (The codeword for burns in that code was SNIVELING—how that must have given the punsters a field day!) And how had McLean gotten this official code? Soon after Warren G. Harding was elected President, McLean explained, he himself was made a special agent of the Department of Justice and was given a little card, a badge, and the Just
ice Department code. The sender of the January 9 message, William O. Duckstein, McLean’s private secretary, was a former Justice agent who seems simply to have kept the book. Burns later took the stand to explain to the committee that when he came into the Justice Department he found that the old code was so widely held that he had a new one made up, only two copies of which were given to each branch. He instructed the agents in charge “that whenever an agent went out on a more important case, where it would require the use of an absolutely secret code in order to communicate with either me or the local office, they were to use this new secret code. The other code is used by the agents in the field when they get into a little town and do not care to write out a telegram that the operator could spread. Mr. McLean had a code book. Every agent that asked for one could have one; but not so with the new code.”

  Friedman’s decodings of the McLean messages showed McLean exhibiting an intense interest in a matter about which he had said he knew nothing. The plaintext readings tended to suggest that McLean’s $100,000 loan to Fall was in fact involved in the oil reserves matter and was not just some pleasant isolated transaction. It was soon discovered that the money had really come from Doheny. Fall was later convicted of accepting a $100,000 bribe and served a year and a day in jail. Doheny and Sinclair were acquitted of trying to bribe a public official, but their leases on the reserves were canceled.

  Cryptology did not supply an essential link to these events. Its contribution seems peripheral. Friedman’s testimony spiced the proceedings with the mystery of code and cipher and helped revitalize flagging public interest in the exposé—and public interest is, in the long run, the only force for honesty in government.

  Criminals, like anyone else, will bother with codes and ciphers only when they have to, and the only time they really need them is in international smuggling, when illegal movements have to be coordinated in secrecy over great distances. In 1934, Swedish police gave cryptologist Yves Gyldén some cipher messages sent by smugglers, a typical one of which read: 16 48 59 74 29 53 99 32 86 28 60 0 St-a 55 67 29 07 28 67 55 44 46 63 80 90 02 99 06 03 15 05 74 59 69 00. He quickly determined that it was a rather simple homophonic substitution and that the plaintext read: Överlämna 28600 St-a allt klart henom [error for genom] spärren, with the 00 meaning period. As a result of these solutions, completed August 19, the police lay in wait for the fast, low Kismet, whose crew discovered that things were not “allt klart” when they were nabbed on September 26 with 5,000 liters of illegal liquor.

  The greatest era of international smuggling—Prohibition—created the greatest era of criminal cryptology. Complicated criminal systems of code and cipher are now rare. They survive only in the drug traffic—the only major international smuggling racket still in existence. For example, two heroin smugglers, John D. Voyatzis and Elias Eliopoulos, encoded their messages in the Universal Trade Code, then moved the center digit to the front, and replaced the new digit with the equivalent five-letter codeword in that book. Thus the plaintext sold was encoded to 58853; this became 85853, which was turned into XIQWD. Much simpler systems exist side by side with the better ones. In 1955 some dope traffickers that used the aircraft of the British Overseas Airway Corporation as their transport wired some accomplices in Bahrein, ORDERING 19 COULD MANAGE MORE IF AVAILABLE, which meant Arrive Bahrein 19th. Could carry more opium than last time.

  Today, most criminal cryptology springs from the efforts of bookmakers to conceal or nullify evidence of their illegal activity. The systems are highly specialized and suitable only for bookmaking. Usually they combine encipherment of numbers with abbreviations for recording bets and payoffs; they do not have sufficient scope to encipher ordinary plaintext. Their solution demands as great a knowledge of the various illegal forms of gambling as the bookie’s own, for the “plaintext” will be only a series of numbers!

  One highly popular game is policy, or the numbers racket. The bettor puts his money—which may be as little as a dime—on a three-digit number. If the number turns up, he wins 666 times his wager. These odds considerably favor the house. Many variations and combinations exist beyond this basic bet. He can play any permutation of the three-digit number, winning correspondingly less if one of the six permutations shows up. (Usually the three-digit number is determined by a syndicate from the last digits of the total amount wagered on three successive horse races at a nearby track.)

  Perhaps the world’s expert in solving policy codes was Abraham P. Chess, who, before his transfer to another bureau of the New York City government, frequently broke them for the New York City Police Department as a sideline to his normal work as an attorney in the department’s legal bureau. He had become interested in cryptology when he read Poe’s “The Gold-Bug” at 18, but got into criminal cryptanalysis quite by accident.

  In 1940, a New York City detective spent the greater part of a day watching a policy collector take bets and, it appeared, note them down. But when the detective made his arrest, he found to his surprise that the notes constituted not the usual bet record, but several sheets of music, neatly staffed and scored, complete with treble clefs, slurs, and hold and crescendo marks. Without evidence that the collector was actually engaged in bet-taking rather than some odd but wholly legal activity, the arrest would not stick. The detective felt sure that the music constituted some kind of code, but the New York City Police Department laboratories had no cryptanalysts. Then another detective remembered that a young lawyer in the legal bureau was interested in cryptology, and the “music” was sent to 30-year-old Abe Chess.

  When he tried it on a piano it proved to be virtually unplayable and absolutely unmusical. He soon observed that only ten different notes were used and that the measures were highly irregular. In seven hours he discovered that each note represented a number, the number being determined by the note’s position on the staff. The bottom line, normally E in music, stood for 1, the bottom space, normally F, for 2, the second line from the bottom, normally G, for 3, and so on to the first space over the scale, normally G, for 0. Each measure, set off by a bar, recorded a separate wager, and the repeat sign of two dots at the end of a measure indicated a combination play. Altogether, the score held 10,000 bets. Chess’s testimony in court brought about a conviction that, without his evidence of bet-taking, would have been impossible.

  From that first solution, Chess went on until by 1951 he attacked 56 such systems, or more than one a week. Their variety was astonishing. To represent numbers, the gamblers used Greek letters or Hebrew letters or even ancient Phoenician letters; or they invented characters or used the system that equates the numbers with a 10-letter word that has no repeated letters. If this keyword were CUMBERLAND, a policy bet of 25 cents on the number 137 would be written CMLUE—the number is ordinarily given first with the bet following, usually without a space between them. One system appeared at first to be shorthand notes in which some of the characters were identical with others except for their shading, some being dark and some light, just as in Pitman shorthand. Chess soon discovered that a light diagonal slanting from the upper left was 1, that the same diagonal written heavily was 2, and that the other numbers were coded similarly. Another policy collector’s notebook contained long, bizarre words like SINKKATYUNDEYO. Chess puzzled over this for almost three hours until he suddenly recalled that the suspect came from French West Africa. The “code” was nothing more than French numerals written phonetically as they are pronounced in the suspect’s dialect, except that zero was represented simply by the letter O. SINKKATYUNDEYO meant cinq quatre un deux zero, or, on number 541, a 20-cent bet. Chess compiled frequency tables for policy games, which the New York City Police Department still uses in its solutions. Chess eventually left the Police Department, but the work he started proved so valuable to law enforcement that today the department has a number of individuals to solve all the gamblers’ codes it gets. The Chief Inspector’s Investigations Unit, the Police Academy, the Police Laboratory, and the Police Commissioner’s Confidential Unit
all have cryptanalysts.

  Police departments not fortunate enough to have an Abe Chess—and this means nearly all of them—often call upon the F.B.I. for the solution of encoded bookmaking records. The F.B.I. does this work in its Cryptanalytical and Translation Section, whose existence it seeks, for some reason, to conceal; it has even lied about its location. In fact the section operates in a plain cream-colored concrete building with no markings except its street number—215 Pennsylvania Avenue, South East, in Washington, D.C. A marquee testifies to its former use as Frank Small’s auto agency, but the show window has been bricked up and even the owner of the liquor store next door knows only that it houses some hush-hush activity. Though the building stands almost within the shadow of the Capitol dome, it is so situated that no one can look into its windows. On the ground floor is a classroom and a green blackboard, sometimes covered with exercises in symmetry of position; on the upper floors are the offices. Entrance is through a courtyard in the rear.

  But if the F.B.I. hides the place where it effects its solutions of bookmaking codes for law-enforcement agencies, it does not hide its light under a bushel when it comes to telling of its successes.

  Cryptosystems involving bookmaking usually fall into one of two general categories. The most common system involves digital encryption, that is, reduction of most of the betting data to numbers and then enciphering these numbers. Tracks are assigned arbitrary number or letter equivalents, horses are identified by post position or racing sheet numbers, amounts and types of wagers are indicated by numbers significant by their position in the entries as well as their identity. Super-encipherment is frequently the substitution of letters or symbols, or combinations of these, for the code numbers. The second general class of “bookie” cryptosystems involves phonetic and related types of abbreviations, including use of foreign or corrupted language text.