THE CODEBREAKERS

by DAVID KAHN

  This was only one of a number of Confederate cryptograms solved by this triumvirate, who, being barely out of their teens, were probably the youngest wartime cryptanalysts in history. The solution did not help Grant take Vicksburg, but it provided the three young men with a Confederate keyword, of which the South apparently used only three during the war. Early in 1865, J. B. Devoe, acting master of the United States Navy, was reporting to the Assistant Secretary of the Navy the two known keywords—MANCHESTER BLUFF and COMPLETE VICTORY (a phrase the Confederates clung to long after that cherished hope had dissipated)—and confessing that “the new key is not known.” But the youngsters’ most important solution dealt not with military but with political affairs.

  In December of 1863, Postmaster Abram Wakeman of New York spotted an envelope addressed to Alexander Keith, Jr., in Halifax, Nova Scotia, who was known to be in frequent communication with rebel agents. Wakeman turned it over to the Secretary of War, who found that the letter inside was written in a complicated mixture of symbol ciphers. After War Department clerks puzzled over the mysterious signs in vain for two days, the cryptogram was given to the “Sacred Three,” as Bates, Chandler, and Tinker liked to call themselves. They determined to do what the clerks could not.

  They ascertained that the unknown encipherer had intermingled five different kinds of signs plus ordinary letters as substitutes in the letter. But he had imprudently marked off the words with commas and confined himself to a single set of signs within each word. The letter patterns of the plaintext consequently showed through. One 6-letter word repeated its second and sixth letters. It was followed by a 4-letter word that in turn was followed by the cleartext phrase reaches you. The three deduced that the sequence should read before this reaches you. Bates recognized the ciphertext signs involved as those of the pigpen cipher, which had been used as a price marker in the Pittsburgh store in which he had worked as a boy. This permitted prompt reconstruction of the entire pigpen alphabet, driving a substantial wedge into the cryptogram. The identification of signs in the dateline as standing for “N.Y. Dec. 18, 1863” yielded further values, and, working in this way, the three—with the President hovering about anxiously—unlocked the cipher in about four hours. It read:

  The Confederate agents’ message, solved by Tinker, Chandler, and Bates

  N Y Dec 18 1863

  Hon J P Benjamin Secretary of State Richmond Va

  Willis is here The two steamers will leave here about Christmas Lamar and Bowers left here via Bermuda two weeks ago 12000 rifled muskets came duly to hand and were shipped to Halifax as instructed We will be able to seize the other two steamers as per programme Trowbridge has followed the Presidents orders We will have Briggs under arrest before this reaches you Cost $2000 We want more money How shall we draw Bills are forwarded to Slidell and rects recd Write as before

  J H C

  A special cabinet meeting was called, and by 7:30 that evening Assistant Secretary of War Charles A. Dana had started for New York to take charge of an investigation. Two days later, another cryptogram addressed to Keith was intercepted and promptly solved. “Say to Memminger,” it read, “that Hilton will have the machines all finished and dies all cut ready for shipping by the first of January The engraving of the plates is superb.” Christopher G. Memminger was the Confederate Secretary of the Treasury; the letter made it clear that plates for printing rebel currency were being made in New York. Hilton, the engraver, was easily located in lower Manhattan, and on the last day of the year the U.S. marshall raided his plant, seizing the plates, machinery, dies, and several million dollars worth of already-printed bonds and money. The plot was broken up, the Confederacy deprived of badly needed plates for printing its paper money. For their central role in all this, the three junior cryptanalysts each received the handsome raise of $25 a month.

  The men in gray, who sometimes could not read their own messages, could never solve the Union’s. The ravings of the Delphic oracle must have seemed more clear than messages in the federal route transposition. Though many of the North’s estimated 6,500,000 telegrams were in cipher, though the Confederates tapped the Union wires, though their cavalry raids must have captured parallel plain and cipher copies of messages, though the system had intrinsic weaknesses—though they had all these clues, the rebels never sorted out the Yankee word-thicket. This would be incredible if they had not vouched for it themselves by publishing a number of messages in their newspapers with a general request for solution. Even the capture of two of the ciphers themselves—No. 12 in July of 1864 and No. 1 in September—failed to help. The Yankees simply got out a new list of routes and jargon words, and the result was always more than the rebels could handle.

  Appomattox itself did not still the cryptologic reverberations of the Civil War. In the trunk of John Wilkes Booth, found in his room at the National Hotel after he was shot, officials discovered a Vigenère tableau. This was introduced into evidence at the trial of the eight Southern sympathizers charged with conspiring to assassinate the President in an obvious attempt to link them with the actual killer, though no one testified to their use of the cipher. The prosecution then sought to show that the crime had been instigated by the Confederate government by exhibiting a rebel “cipher reel,” which Major Eckert averred to be identical with the Booth cipher. This curiosity, captured on a shelf in the Richmond office of Judah P. Benjamin, Confederate Secretary of State, simply consisted of a Vigenère tableau wrapped around a cylinder; over this, an arm supported two indicators that presumably pointed out the letters. It deciphered no messages at the trial. The burlesque reached a climax when a North Carolina pile-driver named Charles Deuel described how he and a friend solved a cipher that he found floating in the water near where he was working. The plaintext, signed “No. 5,” began: “I am happy to inform you that Pet has done his work well. He is safe, and Old Abe is in hell.” What connection all these displays had with the accused was never made clear, but they were hanged anyway.

  At about the same time that Booth and others were being hunted down and captured, Jefferson Davis was using the third Vigenère key to compose the last official cryptogram of the Confederacy. Sent to his secretary on April 24, almost two weeks after Lee’s surrender, it was a message of futile defiance ordering “active operations to be resumed in forty-eight hours.” No one knows who chose this final key of the Confederacy, or why, but in view of Davis’ own impending fall and the black days of Reconstruction that lay just ahead, it gleams as the most somberly prophetic in the whole history of cryptology: COME RETRIBUTION.

  On the morning of Monday, October 7, 1878, the New York Tribune trumpeted forth one of the great scoops of American journalism. Under the two-column headline “The Captured Cipher Telegrams,” the lead story of the day blared the plaintext of cryptogram after cryptogram that the Tribune had solved. The messages, which hearkened back to the most famous electoral dispute in American history, were the first to play a vital role in American politics.

  After the popular votes were counted in the presidential election of 1876, the Democratic candidate, Samuel J. Tilden, held a clear lead of 250,000 ballots over his Republican opponent, Rutherford B. Hayes. But which way the deciding electoral college vote went depended on which of the double and conflicting returns from Florida, Louisiana, South Carolina, and Oregon were accepted as valid. Congress created a special electoral commission to settle the matter; by a straight party vote of 8 to 7, it awarded all 22 contested electoral votes to Hayes. This gave him a majority of 1 in the college—and the Presidency.

  During the tumultuous legislative session that followed, a Congressional committee was appointed to look into persistent Democratic rumors of Republican purchase of electors’ votes. As part of its investigation, the committee subpoenaed 641 political telegrams out of the 29,275 that had clattered back and forth between politicians and their agents in the four states—the vast majority having been burned by Western Union to publicize the privacy of the correspondence entrusted to it. A
large bundle of the impounded wires kicked around the committee room during the summer of 1878, and, through a complicated chain beginning with a committee messenger and ending with the Republican National Chairman, 27 of the telegrams in cipher were leaked to the Republican-leaning Tribune in the hope that they might embarrass the Democrats.

  A few weeks earlier, Manton Marble, one of Tilden’s closest political advisors, had written an open letter to the New York Sun contrasting dark Republican practices with Tilden’s station in “the keen bright sunlight of publicity.” Whitelaw Reid, the Tribune’s brilliant editor, took a suggestion of the Republican chairman and inserted the cipher telegrams in editorials as subtle commentaries on Marble’s letter. The Democrats squirmed as the Tribune staff played impishly upon the ciphertexts. Was this cryptic mumbo-jumbo the vaunted Democratic candor? But as more and more dispatches poured in upon Reid from other G.O.P. sympathizers, he conceived a broader scheme. Reckoning that any negotiations that had to be conducted beneath the cloak of cipher would mightily discomfit the Democrats if drawn from under that cover, he set to work to get them read.

  Prompted by hints in the editorials, numerous subscribers offered suggestions for their solution. Schuyler Colfax, who had been Vice President during Grant’s first term and had been interested in cryptology since his teens, referred Reid to several magazine articles on the subject, but they proved useless. William M. Evarts, the Secretary of State, had a good idea: get a student of mathematics to unearth the law on which the messages were based. But this only promised; it did not produce. Reid even tried the approach direct when he ran into Tilden at fashionable Saratoga that August: “I told him that we had all the cipher dispatches that went between his house and Florida, and asked him, laughingly, for the key. I told him we couldn’t make head or tail to them, and wanted him to help us. He smiled and blushed, innocent as a baby, and passed on.” Things were getting nowhere.

  Meanwhile, the Detroit Post had learned from a former business partner of J. N. H. Patrick, one of the Democratic agents, that the Democrats had couched their electoral communications to Oregon in the same dictionary code that Patrick had used in his mining ventures. The encoder had looked up the word in the edition of the Household English Dictionary that was published at London in 1876, noted the word’s numerical position on the page, and took the corresponding word four pages to the front of the book as the code equivalent. The decoder had reversed the process. For instance, the most damning of the Oregon messages read, in codetext:

  J. N. H. PATRICK

  BY VIZIER ASSOCIATION INNOCUOUS TO NEGLIGENCE CUNNING MINUTELY PREVIOUSLY READMIT DOLTISH TO PURCHASED AFAR ACT WITH CUNNING AFAR SACRISTY UNWEIGHED AFAR POINTER TIGRESS CUTTLE SUPERANNUATED SYLLABUS DILATORINESS MISAPPREHENSION CONTRABAND KOUNTZE BISCULOUS TOP USHER SPINIFEROUS ANSWER

  The first codeword, by, is on page 30 as word 29. The decoder counted towards the back to page 34, where the 29th word is certificate. The entire plaintext read:

  Portland, Nov. 28, [1876]

  W. T. Pelton, New York

  Certificate will be issued to one Democrat. Must purchase Republican elector to recognize and act with Democrat and secure vote and prevent trouble. Deposit ten thousand dollars my credit Kountze Brothers, Twelve Wall Street. Answer.

  J. N. H. Patrick

  On September 4, one of the Tribune’s editors, John R. G. Hassard, basing his work on the Detroit Post’s revelation, set forth 3½ columns of cryptograms and translations that showed that the Democrats had sought to buy a Republican elector for $10,000 and that the deal had fallen through only through delays in transmission.

  But the Household English Dictionary was not the key so urgently desired to the messages from the other three states. With no outside help forthcoming for their solution, Reid set his staff to work on the problem in earnest.

  Hassard, then 42, had become managing editor in all but name on the death of Horace Greeley in 1872. A tall, lanky man with sandy hair, side-whiskers, and hazel eyes, always spruce, with a no-nonsense manner, he was gifted with a charm of style and breadth of culture that showed in his graceful editorials. He had converted to Catholicism at 15—a courageous act in the heyday of Know-Nothingism—and, after graduating from St. John’s College at the head of his class, abandoned his plans for the priesthood only because of ill health. He served as secretary to the first archbishop of New York, John Hughes, whose biography he later wrote. His dispatches to the Tribune from Bayreuth on the premiere of the Nibelungen Ring series in 1876 did more to bring Wagner’s music to America than perhaps anything else up to that time. Hassard took on the challenge of the cryptograms himself, and worked on them so uninterruptedly that a cold hung on and developed into tuberculosis. He spent the next ten years in search of health, but succumbed in 1888.

  Soon after Hassard started his task, another member of the Tribune staff became interested and took up the puzzles independently. This was Colonel William M. Grosvenor, who had become economic editor of the Tribune in 1875, three years earlier. A burly, forceful man, then 43, with bristly eyebrows, long hair and beard, and a leonine head, he had demonstrated his statistical skill while editor of the St. Louis Democrat by making an elaborate comparison between the whisky production of the St. Louis distillers and the revenue accruing therefrom to the government. It clearly indicated fraud on the part of the liquor interest and led to exposure of the notorious Whisky Ring. A native of Massachusetts, he had commanded a regiment of Negro troops in the Civil War. Grosvenor later became editor of the prestigious Dun’s Review and was frequently consulted by government experts on tariff and currency legislation. His integrity in these matters was so great that on one occasion his advice cost him a fortune in the stock of a printing firm. He was gifted in mathematics and languages, was one of the most expert billiard players in New York, could carry on three games of chess simultaneously, and more than held his own at tennis and whist. He died in 1900.

  Grosvenor and Hassard, each in his own home, wrestled with the riddles and obstinacies of an unfamiliar science. They could not have known it, but not only were they mastering a problem that had repulsed many, they were also breaking new ground in that science.

  A cipher telegram offering the electoral votes of Florida for $200,000, solved and published by the New York Tribune

  “They both did extremely well,” Reid said later, “worked independently, compared notes loyally and altogether cooperated in a charming way in a highly important piece of work. Hassard was a little earlier in the field, and to that extent deserves special credit; but Grosvenor was equally keen, and, as well as I can now remember it, about equally successful. Sometimes he and Hassard would attack the same despatch on different lines, and after being foiled again and again, would finally reach the solution the same evening, Hassard in Eighteenth Street and Grosvenor out at Englewood.”

  Unknown to them, a young mathematician from the U.S. Naval Observatory in Washington had been solving some of the specimen ciphers that Reid had published in the Marble-baiting editorials. This was Edward S. Holden, 31, who had graduated third in the West Point class of 1870 and had gone to the Naval Observatory three years later. In 1879, the year following his work on the cipher telegrams, he was appointed librarian there. In 1885, he became president of the University of California and director of the Lick Observatory, relinquishing the presidency in 1888 on completion of the observatory. He founded the Astronomical Society of the Pacific, organized five eclipse expeditions, and edited the observatory’s publications. From 1901 until his death in 1914 he was librarian at West Point, adding 30,000 volumes to the collection, cataloging it, and issuing many bibliographies.

  Holden had been attracted by the “novel and ingenious character” of the cryptograms. “By September 7, 1878,” he said later, “I was in possession of a rule by which any key to the most difficult and ingenious of these … could infallibly be found.” He approached the Tribune, which had liked Evarts’ idea of hiring a mathematician, and Hassard sent on a quantity
of dispatches. But Hassard and Grosvenor had independently reached the theory of solution that Holden had, and furthermore had solved some messages before he did. None of Holden’s solutions reached the Tribune before Hassard and Grosvenor had solved those messages, Reid said, and in general his work was regarded as corroborative.

  The most important messages, and those to which the new theory of solution applied, were enciphered in a form of word transposition grievously deteriorated from the excellent Civil War system that had evidently inspired it. Only four keys were employed, one each for telegrams of 15, 20, 25, and 30 words, with longer telegrams being enciphered in parts by two or more keys. Sometimes deciphering keys were used to encipher. Code disguised some of the proper names and important words. The enciphering key for 25 words (18, 12, 6, 25, 14, 1, 16, 11, 21, 5, 19, 2, 17, 24, 9, 22, 7, 4, 10, 8, 23, 20, 3, 13, 15) served to encipher this honest offer of corruption from Tallahassee:

  In the code list, BOLIVIA stood for proposition, RUSSIA for Tilden, LONDON for canvassing board, FRANCE for Governor Stearns, MOSELLE for two, GLASGOW for hundred, EDINBURGH for thousand, and MOSES for Manton Marble. As transmitted to New York, the message read:

  CERTIFICATE REQUIRED TO MOSES DECISION HAVE LONDON HOUR FOR BOLIVIA OF JUST AND EDINBURGH AT MOSELLE HAND A ANY OVER GLASGOW FRANCE RECEIVED RUSSIA OF

  The reply to that is extant; it was both clear and in clear: “Telegram here. Proposition too high.”

  The Hassard-Grosvenor-Holden theory of solution of messages like this fed upon the great quantity of dispatches in each key. It is now considered the general solution for all transposition ciphers, because it works on any transposition whenever two or more cryptograms of the same length in the same key are available for analysis. The method, which they developed empirically for the first time in cryptology, has become known as “multiple anagram-ming,” and though Holden did not use that term, he gave a good description of the technique: “There is one way, and only one way, in which the general problem can be solved, and that is to take two messages, A and B, of the same number of words, and to number the words in each; then to arrange message A with its words in an order which will make sense, and to arrange the words of message B in the same order. There will be one order—and only one—in which the two messages will simultaneously make sense. This is the key.”