by DAVID KAHN
Such occasions were rare. Direct telegraph connections were soon established between Pokorny’s group and Deubner’s; together they laid open virtually every Russian cryptogram that their posts intercepted. And they were guaranteed a good harvest when the headquarters of a Russian army was given permission to use radio for its front-line activities because its linemen were busy with repair work.
Thus it was that the Central Powers learned from Russian wireless that the Grand Duke Nicholas was forming a huge phalanx of seven armies to rumble into the industrialized heart of Silesia in east-central Europe. By the end of October, the picture of the composition, disposition, and strength of the Russian forces that Hindenburg and Ludendorff had before them could not have differed much from the official one at Stavka. Only the date of the advance was unknown, but the Germans assumed that it would take a little time before this ponderous Russian steamroller could get up momentum. They determined to seize the initiative and attack first in the hope of throwing a monkey wrench into the steamroller’s mechanism.
Ludendorff’s plan was characteristically bold. He removed a German army from the defenses blocking the invader and poised it in the north for a plunge downwards into the right side of the Russian wedge. On November 11, the point of this dagger—an army under Mackensen—began to pierce the Russian flank. At 2:10 p.m. the next day, the chief of staff of one of the Russian armies under attack transmitted a long radiogram which the Central Powers intercepted. In addition to mentioning the date of the projected Russian advance, it specified the line of demarcation between his army and a neighbor—always a zone of weakness. This message lay, cryptanalyzed and translated, on the desks of the German headquarters for the Eastern Front at Posen by the next afternoon.
It was immediately forwarded to Mackensen. At 7:30 p.m., with this picture of the Russian dispositions before him, he telephoned his order for the next day to his subordinates. It called for an all-out attack, concentrating on the meeting line of the two armies in the hope of driving them apart and breaking through.
He achieved a massive success. The Russian forces were split; they pulled back hastily to the south. Mackensen shoved the dagger in up to the hilt. At the same time, Ludendorff pinned the front Russian armies in combat and sent a corps to turn the Russian left flank. He hoped to effect another Tannen-berg—a double envelopment. In sharp fighting around Lodz, the German forces drove their enemy back, abetted by a constant stream of cryptanalyzed intelligence. On November 15, for example, the German command learned that four corps were to reinforce Russian troops at the Ner and Bzura rivers and that another corps was to cross to the left bank of the Vistula at Plozk. These details enabled the Germans to maneuver each day as if in a war game.
By now the Russians were changing the key to the order of the cipher alphabets—not the alphabets themselves—each day. The cryptanalysts kept pace. On November 18, it appeared that the Germans had won their victory when the cryptanalysts solved a message ordering a Russian retreat from Lodz. But the rejoicing at headquarters was cut short when the codebreakers read a message from Grand Duke Nicholas countermanding the order and directing his forces to fight on despite their difficult position. The flow of radio intelligence continued unabated, and on the 19th Mackensen even delayed giving an order until intercepted information was received.
The next day a premonitory fear chilled the intercept services when they picked up a message from a liaison officer of the Russian 4th Army to a colleague, warning that the Germans had the Russian cipher key. The Russians had captured a German cipher key, and they apparently assumed that one of theirs had likewise fallen into German hands. A new key was instituted—and this time the entire set of cipher equivalents was changed. A curtain of silence descended upon the Eastern Front.
Feverishly, Deubner and Pokorny, who was assisted by Lieutenant Colonel Heinrich Zemanek and Lieutenant Viktor von Marchesetti, grappled with the new key as the intercept posts sucked in every scrap of Russian wireless. The moment could not have been worse. The battle around Lodz raged at its peak, and just as Ludendorff was about to consummate his envelopment with his inferior forces but his superior intelligence, that intelligence was abruptly blanked out. Deprived of his eyes and ears, he did not know of the Russian reinforcements that began to cut off the deeply sunk point of Mackensen’s dagger. By the 21st, the point had been isolated, and the envelopers were themselves enveloped. A guards division and two cavalry corps were encircled by Russian forces with no apparent hope of escaping. The Russians exultantly ordered up trains to carry off the prisoners.
But the next day, Pokorny’s group finally subdued the new Russian alphabets, and the intelligence once again began streaming into German headquarters. Intercepts soon revealed a weak spot at Brzeziny in the ring of Russians. Ludendorff’s headquarters radioed this information to the trapped commanders, who, grouping their forces densely and fighting hard, broke out on the 25th and reached safety, bringing with them 10,000 prisoners. General Lietzmann, commander of the guards division, won the title “Lion of Brzeziny” for the brilliant escape; the cryptanalysts who had showed him how best to use his fangs and claws purred with amusement in their secret lairs.
This harrowing episode, resulting from a fortuitous change of key, balked the Germans of a decisive victory, but they had succeeded in throwing the vaunted Russian steamroller out of gear. Never again did it threaten German soil. The Central Powers pressed forward, still reading Russian cryptograms, and on December 6 the soldiers of the Czar evacuated Lodz, the second city and the industrial capital of Poland. Eight days later they again made a wholesale change of alphabets in their digit cipher. Solution again required several days, and when it was completed the Austro-German command learned that the Russians planned to dig in for the winter along the Nida River. Soon thereafter they gave up the old cipher altogether.
When activity quickened in the spring of 1915, the Russians were using a simple Caesar cipher.* The multiplicity of tables used by different armies in the old cipher, the daily shift of keys, had evidently proved too difficult to handle for the half-illiterate muzhiks. The Austrian and German crypt-analytic organizations saw right through this transparent new cipher and read the indications of a projected Russian invasion of East and West Prussia. Then began what Colonel Max Ronge, head of Austro-Hungarian intelligence, called “the most brilliant period of our interception services.” Enormous quantities of intelligence were sluiced from the Pokorny and Deubner groups into the offices of the operations staffs of the German and Austro-Hungarian commands. Helped by this, they parried the first tentative Russian advances, and then themselves swept through the whole enemy line in a rapid onslaught that penetrated 80 miles in two weeks.
Time after time, their solutions enabled the Central Powers to take steps which were so perfectly the right thing to do in each tactical situation that the Russian general staff was mystified by its opponents’ apparent clairvoyance. Once the Germans fell back just two days before an overwhelming assault was to be launched; had they remained in place, their position would quickly have become critical. After the Germans captured Lodz, the Russians pondered the precision of the enemy moves and decided that the Germans must have obtained intelligence from air reconnaissance.
Eventually, however, the conviction grew that the foe must be reading their ciphers. They did not suspect cryptanalysis. Spies, they thought, must have sold them to the Austrians, and in a wave of spy-mania they persecuted officers with German names—none of whom, Ronge said, had ever given anything to him. The Russians changed their cipher at the height of the enemy’s spring offensive, but this caused the cipher clerks more trouble than it did the crypt-analysts, for almost all messages of May 15 were unintelligible to their recipients and most of those of the 16th as well.
The summer-long Russian retreat finally came to a halt at the end of September on a defensible position deep within their own territory. By then Russia had lost 750,000 men as prisoners and untold hundreds of thousands more as casualtie
s. She simply threw more men into the war. She seemed to adhere to the same policy in cryptography—and with the same lack of success. On December 20, 1915, she put her 13th cipher into operation. The Austrian and German cryptanalysts recognized it at once as having been used elsewhere on the front, and during the inconclusive battles before and after New Year’s Day kept up with the enemy situation hour by hour. On June 16, 1916, the Russians began using their first code, a small one of about 300 groups. This development may have been influenced by the French, who had learned about the German solution of Russian messages from their own cryptanalyses and had passed the news to their allies. Or it may have resulted from Russia’s own intercept service; just how well Russia did in military cryptanalysis is not known, but she did set up direction-finding stations in mid-1916 and started an intercept school at Nicolaieff.
The travail of the Central Powers cryptanalysts, who were unused to code, was simplified when some Russian commands, who were equally unfamiliar with it, continued using the old system. And their work was made almost mechanical when the headquarters of a Russian guard detachment that was being joined to the 8th Army compromised the new system by a message in clear. A great hubbub arose in the 8th Army; a new code was instituted; this one cryptanalysts solved without much trouble. By then they were reading up to 70 Russian dispatches a day. The German solutions seem to have been made in the radio stations of the various fortresses, to which Deubner communicated the keys as he solved them. Some of the Austrian cryptanalysis was done at Ronge’s Austro-Nord Penkala under the command of Captain Karl Boldeskul. Later in the war, when Pokorny was promoted to head of the whole Kriegschiffregruppe, the Russian subsection of the Dechiffrier-dienst at headquarters was taken over by von Marchesetti; in 1918 Rudolf Lippmann succeeded him.
On November 6, 1916, the Russian Army of the Danube suppressed the radio use of cipher No. 14 as known to the enemy, and on December 17 another cipher was called out of service because the radio station of the 1st Cossack Division had been captured. Four days later they returned to the air with a code that proved to be merely a slightly shifted version of one that had been instituted a week earlier. All these changes the cryptanalysts followed with contemptuous ease. The increasing disorganization of the Russian armies contaminated the radio services, and as discipline relaxed, garrulity increased. One day early in 1917, the Dechiffrierdienst solved 333 radiograms, from which it inferred that the Russian secret communications were rapidly disintegrating. In March the Czar was overthrown, in July an all-out offensive by the Russian armies collapsed, and in October the Bolsheviks, using the people’s overwhelming desire for peace, seized power and took Russia out of the war.
The way to this situation was opened primarily by Russia’s military failure. While this resulted largely from the lack of munitions, food, and supplies that the underindustrialized country could not supply, the tactical defeats inflicted by the Central Powers obviously played a conclusive role. And these victories of a David over a Goliath, though aided by superior German equipment, discipline, and logistics, were mainly engendered by cryptanalysis.
“We were always warned by the wireless messages of the Russian staff of the positions where troops were being concentrated for any new undertaking,” wrote Hoffmann. So complete was the intelligence that he could say: “Only once during the whole war were we taken by surprise on the Eastern Front by a Russian attack—it was on the Aa in the winter of 1916-17.” This dramatically underlines the importance of cryptanalysis in the outcome of the war in the East and in all that that entailed. Indeed, it may not be too much to claim that the establishment of Communist power, perhaps the supreme fact of contemporary history, was made possible to a significant degree by the cryptanalysis of czarist secret communications.
The consolidation of the Soviet regime permitted Lenin and his colleagues to turn not only to the difficult problems of running the world’s first socialist state but also to the traditional Communist activity of fomenting class struggle and the revolution of the proletariat. They felt justified in using subversion as well as the more orthodox methods of propaganda and political agitation in advancing Marxism in countries that had not yet reached Russia’s stage of historical development.
Some of their agents were mere mercenaries, some were Russians planted as spies, but many were native members of Communist parties who placed their quasi-religious dedication to that ideology above allegiance to country. These spies were soon sending quantities of information to, and receiving instructions from, Moscow, whence the impetus and control for the world revolution would come. In doing so during these early years before Communist espionage had stabilized itself they employed a wide variety of cipher systems.
In 1919, German Communists employed an irregular columnar transposition. One key was the second line of Heinrich Heine’s “Die Lorelei”: DASS ICH SO TRAURIG BIN: another was ACH, WENN DAS DER PETRUS WÜSSTE (Oh, if Peter only knew it). Three messages so enciphered were discovered in an airplane en route to the Soviet Union from Germany that was forced down in Latvia. The Latvian government, after failing to solve them, had turned them over to the American consul at Riga for help, and from him they eventually found their way to Yardley’s Black Chamber, where they were quickly solved. They began with a plea to Sendet geld (“Send money”), discussed the fiasco of a Communist conference in Holland, reported the arrest of the fiery German Communist Klara Zetkin, pleaded that “[Karl] Radek or [Nikolai] Bukharin [both intimates of Lenin] is absolutely needed here,” disclosed that “Radio station finally ready to send. Expert engaged … Guralski arrived here with money on his way to America,” and cautioned that “My name is now JAMES.” The solutions caused a stir in Washington, for they were among the first authentic documents dealing with Soviet international activities that came into the American government’s hands.
At about the same time, the U.S. Department of Justice began what must have been one of its first infiltrations into the Communist party, U.S.A. Undercover agent Francis A. Morrow of Camden, New Jersey, sent a steady stream of reports to the department while rising in the party to become secretary of a district committee. He got on close terms with the district organizer, and one day the latter, when a little tipsy, let Morrow help him in deciphering a message. The cipher that Morrow thus reported to the Justice Department was used by the party’s leadership to communicate with its organizers. It was based upon a United States postal money order, possession of which would not incriminate anyone. Its ciphertext appeared as a series of arithmetical fractions, whose numerator represented the line of type on the back of the money order blank, and the denominator the letter in that line. The system recalls the fractional one used by the Russian revolutionaries in czarist days, and was, in fact, probably taken over from that time, as were many Communist underground practices. For example, the use of the term “dubok,” which literally means “little oak,” to mean a hiding place for messages, and of “illness” to mean an arrest, originated before the Revolution.
This simple system vanished with the advent of a more highly organized Soviet intelligence apparatus. Russia’s Amtorg Trading Corporation, which set up offices in New York in 1924, controlled the first real Soviet espionage effort against the United States. Communications with the Soviet Union were naturally carried on in code, and, whatever system was used, it effectively protected the secrets of their American spies. Representative Hamilton Fish, Jr., of New York, chairman of a committee investigating Communist activities in the United States, subpoenaed 3,000 coded Amtorg telegrams in 1930 in the hope of learning more about those activities. The cryptanalysts of the Navy’s Code and Signal Section, to which he had submitted the cryptograms, reported that “the cipher used by the Amtorg is the most complicated and possesses the greatest secrecy within their [the Navy cryptanalysts’] knowledge.” Fish then gave the cablegrams to the War Department for solution; two years later, he complained on the floor of Congress, “Not one expert—and they had from six months to a year—succeeded in de
coding a single word of those cablegrams, although they had assured me they could decode them.”
In Copenhagen in 1934, a cipher disk with seven rings of digits around a mixed plaintext alphabet guarded the messages of the Danish Communist party. This key was captured in a police raid, however, after which the messages yielded rather easily to the cryptanalytic ministrations of Yves Gyldén, who was summoned from Stockholm.
It was during the Spanish Civil War, in which Russia actively aided the Loyalists, that a cryptographic element that had served the revolutionary predecessors of Lenin & Co. reappeared in a form both streamlined and more secure. This was the straddling checkerboard. Its straddling feature makes use of cipher equivalents of two different lengths—lengths usually of one digit and two digits; the two sets of equivalents are so constructed that the cryptographer can unambiguously separate them when they are run together. The cryptanalyst, however, not knowing which digits are singletons and which form pairs, may divide the ciphertext incorrectly, thereby “straddling” many of the true pairs and combining two singletons into a false pair. The device also reduces the length of the numerical text as compared with checkerboards in which all letters are replaced by numerical pairs. Straddling was first employed by the Argentis in some of their 16th-century papal ciphers (one wonders whether the atheistic Communists knew!).
The straddling checkerboard produces single-digit equivalents by leaving the side coordinate off one of the rows of the checkerboard. A letter in this row is enciphered by just the single coordinate above it. If ambiguity is to be avoided, none of these singletons can start a two-digit group. Hence none can be used as a side coordinate (which is read first). Using eight digits of the ten as singletons leaves two digits as side coordinates; each of these two side coordinates can then pair with the ten top coordinates (the singletons may serve in second position) to produce 20 two-digit groups. This configuration makes 28 ciphertext equivalents available for plaintext elements.