THE CODEBREAKERS

by DAVID KAHN

  * As corroboration, it might be noted that Painvin won a prize as a young ‘ cello player, that Mauborgne and Kunze both play the violin at least passably, and that Hitchings taught music.

  * All times are local times. This would be May 4 in Hawaii.

  * Whether this solution was made in cooperation or in competition with the Ango Kenkyu Han, the Foreign Ministry’s cryptanalytic section, is not known.

  * This was later proved to be a canard, but its authenticity was accepted at the time.

  * Actually December 11. Marshall was referring to Yoshikawa’s message of December 3, Honolulu to Tokyo, setting up Kühn’s signalling system.

  18

  *

  ALTHOUGH SECRET WRITING appears in Russia in the simple letter-substitutions of 12th- and 13th-century manuscripts, akin to those of medieval France and Germany, political cryptography seems to have first come to the country under the Westernizing influence of Peter the Great.

  The most direct evidence outside the nation’s archives lies in the records of England’s Decyphering Branch, whose first Russian solution is dated 1719—the 37th year of Peter’s reign. This accords well with what one might expect. Peter was fascinated by technical arts of all kinds; he not only studied them but picked up their tools and worked at them. He perhaps heard about codes and ciphers during his visits to Holland and England in 1697-98 and Paris in 1717. It was a time when official nomenclators were formally employed by the emergent nations of Europe—and when cryptanalysts were paid to solve them. If he himself did not import cryptography to Russia, the seeds might have been planted by the foreigners that Peter imported for the governmental reforms that began in 1712. The new structure was modeled on Sweden’s, and perhaps included a cipher office, for by that time Sweden had had more than a century and a half of cryptographic experience—employing, for example, a one-part code of almost 4,000 groups in 1700. Secret writing thus might well have been among the new and useful practices that Peter adopted in transforming Russia from the semibarbarism of Ivan the Terrible to a modern state.

  The first ciphers used by Peter’s ambassadors in London were as primitive as his country then was, and had no more security than first ciphers usually possess: they were monalphabetic substitutions. The plaintext was replaced with secret symbols hardly less bizarre to the insular English eye than the original Cyrillic letters themselves. Such systems served at least until 1728. In the reign of Peter’s strong-willed daughter, Elizabeth, Russian cryptography suddenly blossomed forth with all the maturity of Europe’s best. In 1754, the Russian ambassadors to England employed a two-part nomenclator of 3,500 elements, including homophones. It was in French, which was then not only the language of diplomacy but also the tongue cultivated in most of the courts of Europe, nowhere more slavishly than in Russia’s. (Sweden, too, was using French-language codes.) Other, smaller two-part nomenclators followed at frequent intervals: one of 900 elements appeared in 1755, and still another of 1,000 in 1761.

  Russian monalphabetic key, recovered by England’s Decyphering Branch, 1728

  The next year Catherine II—she who was to make her country the chief continental power of Europe and to become known as “the Great”—ascended the imperial throne. Six years later, the codemakers of St. Petersburg experimented with Russian for a two-part nomenclator of 1,500 elements. By 1780 they had returned to French. It was on a worksheet for this code that an English decypherer noted “many nulls beginning and ending sentences”—authoritative testimony to the craft of the Russian cryptographers. In 1784 they tried something new: a kind of voluntary superencipherment in which the initial numbers 1, 2, 3, or 4 of a codegroup could be replaced at will by 6, 7, 8, or 9, respectively, or left unchanged. Thus que was represented by either 3126 or 8126. This may have been some kind of an economy move, for the underlying nomenclator was one-part; if so, the Russian experts saw quickly that it was a false economy, for the system could not provide enough security, and they discarded it the very next year for a new two-part nomenclator.

  Part of English solution of Russian dispatch encoded with full nomenclator, 1700s

  Annual changes, in fact, may have been routine. One French-language chiffre général existed for 1798 and another for 1799; the presence of a Russian-language general cipher for 1798 in addition indicates the open-handedness of the Foreign Ministry in cryptography. Sometimes nomenclators were changed or canceled before the year was out if they were suspected of being compromised. On January 22, 1800, the Foreign Minister, Count Nikita Petrovich Panin, ordered his ambassador in Berlin not to use the 1799 general cipher, which was thought to have been carried off by the enemy with the baggage of a Russian general during the French revolutionary wars. A similar suspicion may have caused the Foreign Ministry to discontinue a code used by their ambassadors in Madrid and Lisbon after only about ten months of use.

  The Russians exercised great cryptographic prudence. Panin warned the Berlin ambassador: “Your confidential reports must always be ciphered with one of the new keys, even when you use a courier.” As an added precaution, he wrote many of his own dispatches in invisible ink beneath a cover-text. This also had the advantage that development of the ink would indicate rather pointedly that the letter had been tampered with. Once he wrote to Berlin that “Not having at hand the sympathetic ink that I have been using, I used lemon juice today in the attached confidential letter; consequently, instead of dipping it into aqua fortis, it must be heated.” All this sophistication suggests that the Czarina’s cryptographers learned their techniques in the only way they really can be learned—through cryptanalysis.

  For among the Western innovations that had come to the new Russia was the exceedingly valuable one of black chambers. Situated, like those of England, France, and Austria, in the post offices, they employed the full battery of expert openers, seal-forgers, translators, and cryptanalysts. At least some of the latter appear to have been German, probably hired by Peter, and their descendants seem to have maintained a monopoly in this field for generations.

  The black chambers were in operation as early as Elizabeth’s reign, and the French ambassador, the Marquis de la Chétardie, knew full well that they were opening his dispatches. But they were enciphered, and, in the manner of diplomats everywhere, he felt safe because he thought that the Russians were too dumb to break his cipher. He may have been right about Russians, but three Germans in the black chamber were making mince pie out of it. He erred in writing home with a deplorable lack of gallantry about the Czarina, remarking that she was “given entirely to her pleasures” and was “so frivolous and so dissipated.” The interceptions were seen as a matter of course by Count Aleksey Bestuzhev-Ryumin, grand chancellor of the imperial court. He had been waiting to strike back at Chetardie, who had organized a cabal against him because of his Anglophile tendencies. He showed the solutions to Elizabeth, who, blinded by her own French leanings, refused to believe them until he deciphered them in her presence. The next day, June 17, 1744, as Chétardie entered his residence, he was handed a note ordering him to leave Russia in 24 hours. He protested; a Russian began reading him his dispatches. “That’s enough,” he said, and started to pack.

  At the turn of the century, cryptanalytic information was still informing Russian foreign policy. Foreign Minister Panin wrote on March 26, 1800, from St. Petersburg to his ambassador in Berlin: “We possess the ciphers of the correspondence of the king [of Prussia] with his charge d’affaires here: in case you suspect [Prussian Foreign Minister Count Christian von] Haugwitz of bad faith, it is only necessary to get him to write here on the subject in question under some pretext, and as soon as his or his king’s dispatch is deciphered, I will not fail to apprise you of its content.”

  Twelve years later, Russian cryptanalysis played an obbligato to the grand symphony of the Russian winter in inflicting the first defeat on the hitherto unconquerable Napoleon. That military genius, though not quite the cryptologic moron that it has been the fashion to portray him as being, certai
nly did not fully appreciate the importance of a tough cryptography. He depended upon a single, easy-to-solve system during most of his campaigns, including the Russian; this was his petit chiffre, a nomenclator of about 200 groups. Even without his generals’ predilection for partial encipherments, the Napoleonic cryptograms must have crumpled before the assault of the Russian cryptanalysts. How the solutions helped the Russians is not known, but that they must have been of some assistance is indicated by the fact that the victorious Czar, Alexander I, cited them himself when reminiscing about the war. At a state dinner that he gave in Paris years later for the marshals of France, he mentioned having read secret French dispatches. Marshal Macdonald, who had commanded a corps for Napoleon, recalled that one of the French generals had defected and said, “It is not surprising that Your Majesty was able to decipher them; someone gave you the key.” Alexander denied it. “He assumed a serious air,” Macdonald related, “placed one hand on his heart and raised the other. ‘No,’ he replied, ‘I give you my word of honor.’ ” His cryptanalysts would have been proud of so stout a defense of their honor.

  During the nineteenth century, cryptanalysts functioned as one of the Czar’s chief tools of despotism. Libertarian movements were growing increasingly restive and radical. One way in which the Okhrana, the notorious secret police, kept tabs on underground workers was to have the black chambers read the letters and telegrams of suspects—as well as most foreign mail and a random selection of the domestic post, too.

  Permanent black chambers were established in the post offices of St. Petersburg, Moscow, Warsaw, Odessa, Kiev, Kharkhov, Riga, Vilna, Tomsk, and Tiflis; temporary ones were set up elsewhere when needed. Most of the experts were foreigners, though Russian subjects; a fair number were Germans who spoke Russian with a heavy accent, apparently because they secluded themselves for security’s sake from their neighbors. Though they mostly worked from watch lists, they became so sensitive to the nuances of clandestine correspondence that they could detect a suspicious letter from an insignificant blot on an envelope, a line under a name, the odd formation of an address. Letters were usually opened with steam or a hot wire or blade under the wax seal, but one dedicated employee, Karl Zievert, in charge of the Kiev office (and later convicted of being an Austrian spy), invented a device that eliminated all possibility of telltale wrinkles or scorches. It consisted simply of a thin round, polished, flexible stick about the size and diameter of a knitting needle, split down half its length. Zievert would slide this under the flap of an envelope at the corner, catch the letter in the slit, furl the paper around the needle, and then draw it out without noticeably distending the envelope!

  Ciphers posed few problems for the official meddlers. The black chambers forwarded the cryptograms they had light-fingered to the Okhrana, whose specialist in cryptanalysis, Zybine, displayed almost uncanny powers. The former Okhrana head in Moscow, P. Zavarzine, has given a vivid portrait of him: he was fortyish, tall, thin, swarthy, with long hair separated by a part, and with a lively and piercing look. “He was a fanatic, not to say a maniac, for his work. Simple ciphers he cleared up at a glance, but complicated ciphers placed him in a state almost of a trance from which he did not emerge until the problem was resolved,” said the police chief.

  Zavarzine had to send for him once in 1911 to solve an intercept, written in secret ink and consisting largely of fractions, which nobody in the Moscow section could make out (apparently the black chambers of Okhrana prefectures had staff members who could solve the simpler ciphers). Zybine arrived the next morning from St. Petersburg and barely greeted Zavarzine before asking for the letter. An official gave him a copy. He wanted the original. He instantly started for the post office to get it, but was told that it had already been sent on. Zavarzine lent him his desk, and soon Zybine was totally immersed in his work, scribbling rapidly on papers spread before him. When Zavarzine returned to invite the cryptanalyst to dinner, he had to call him twice before he answered, and to insist before he came. At table, Zybine, still in a trance, downed a bowlful of soup, then turned the plate over and tried to write on its back. When the pencil would not take, he started on his cuffs—all the time completely ignoring his hosts. Suddenly he leaped from his chair and shouted, “Tishe idiote, dalshe budiote!”

  After which he sat down, relaxed, and ate his dinner like a normal man. He explained to Zavarzine that the repetitions of the letters had given him the clue. The proverb that he had shouted, meaning “Who walks softly goes far,” served as the key for the cipher. The phrase was written vertically. Each of its letters headed a Caesar alphabet (in Russian) that extended out to the right; these rows were numbered. The ciphertext fractions that formed the ciphertext were composed by taking as numerator the number of the row of the plaintext letter and as denominator the position of the letter in the row. Thus 1/3 would mean the third letter in the first row; since this row was headed by T, 1/3 would represent Φ. The cipher, a weak homophonic substitution, served the underground as one of its standard systems. The message told of sending some cardboard boxes, undoubtedly loaded with explosives, to Kiev when the Czar was planning a visit there. Zavarzine promptly slapped shadows onto the addressees of the letter and kept them from blowing up their Little Father.

  Zybine said that he had been defeated by a cryptogram only once, in a letter sent by an Austrian spy. “But that was a long time ago,” he told Zavarzine. “Today it wouldn’t happen.” The last head of the Okhrana, Alexei T. Vassilyev, also speaks of Zybine, though he does not give his name. In one case, a raid on a house in Sevastopol uncovered a sheet of paper covered with figures. Vassilyev gave it to Zybine, who suggested that the chief telegraph to Sevastopol for a list of all books found in the house. A short time after Zybine received it, he placed the solution before Vassilyev; it was based on The Duel by Aleksandr Kuprin—which was, appropriately, a novel of protest against the Russian military class. Zybine got a raise and a decoration for that job. On another occasion, he cracked another terrorist missive as soon as he had learned from Vassilyev the price of a pound of dynamite! Vassilyev, who seems to have been a little awed by Zybine’s mysterious faculty, says that the cryptanalyst could pick out nulls and nonsignificant lines at a glance.

  The most popular cipher of the Russian underground seems to have derived from the prisons in which so many of its leaders had to serve time. Intercommunication among the inmates was strictly forbidden. But the prisoners, languishing in the tomblike solitude of their gloomy stone casements, with nothing to occupy their minds, had the patience, perseverance, and ingenuity to outwit their jailers. They knocked, using the number of taps to indicate the rows and columns of a simple checkerboard, like the original Polybius square, sometimes 6 × 6 to accommodate the 35 letters of the old Russian alphabet, more often five across and six down, with the alternate letter forms eliminated. In English, the checkerboard would take this form:

  Thus hello would become 23 15 31 31 34. Prisoners quickly memorized the proper numbers and “talked” at from 10 to 15 words a minute. The system was universal in the penal institutions of Russia, with felons as well as political convicts employing it.

  One of its advantages was that it afforded communication by a great variety of media—anything that could be dotted, knotted, pierced, flashed, or indicate numerals in any way could be pressed into service. It often concealed a message within an innocuous handwritten letter. The ciphertext numbers were indicated by the number of letters written together; breaks in the count were indicated by minute and almost imperceptible spaces, much as occur naturally in many persons’ handwriting. Spaces between words were bridged by having the last letter of a word end in an upstroke if the count was to continue, in a downstoke if the end of the word coincided with the end of a count. This subtle means, in which the cover-text bears no relation to the underlying message, and so does not have to strain to make sense, frequently bootlegged secrets in and out of prisons, and undoubtedly past the noses of the black chamber experts, until they finally caug
ht on.

  The popular cipher that the checkerboard inspired is named for the Nihilists, the anarchistic opponents of the czarist regime, who may have invented it. The Nihilist cipher converts both the plaintext and a repeating keyword into numerical form via the checkerboard, and then adds them together to produce the ciphertext. If the keyword is ARISE, or 11 42 24 43 15, the plaintext Bomb Winter Palace would be enciphered like this:

  Occasional three-digit groups will occur, as 55 + 54 = 109. The cipher is a kind of modified numerical Vigenère with additional weaknesses that simplify solution. It would not have baffled a Zybine very long. Yet this basic system—the adding of a key to a checkerboard substitution, though with important improvements—survived through the years to become the primary form of secret communication for Russian undercover agents.

  Only one other department of Russian officialdom coddled cryptology as did the Ministry of the Interior’s Okhrana: the Foreign Ministry. It employed six or seven codes, most of only 1,000 elements, the more important of which were superenciphered by a table of 30 number alphabets. Keys varied from day to day, and deliberate “errors” were reportedly made to muddy the statistics of enemy cryptanalysts; these naturally had to be eradicated by the cipher clerks before they could read their own messages. The wily Russians also employed a code that they knew had been solved by other nations to keep the foreign cryptanalysts happy and productive and away from the important codes. Still, Russian codes were read, either by bribery or by solution. At least one of the cryptanalyses of a Russian diplomatic code was made through the classic entry of guessing that a message ended with a full stop.

  Inheriting, perhaps, the cryptanalytic service that had solved the dispatches of diplomats during the times of Elizabeth and of Catherine the Great, the Foreign Ministry impartially read the coded messages of friends and foes alike: Turkey and Austria-Hungary in the latter category, France and England in the former, and Sweden, a neutral, in neither. Just before World War I, the Foreign Office cryptologic organization was streamlined by Aleksandr A. Savinsky, chief of the ministry’s cabinet from 1901 to 1910. He placed the cryptanalysts directly under the minister, introduced new codes, and promulgated strict regulations for their employment.