The Secret Sentry Read online

Page 3


  At the same time that Sauterne was solved, GCHQ began producing the first intelligence derived from its solution of another Russian army cipher machine system, which the British called Coleridge and which was used to encrypt traffic on Russian army radioteletype networks in the European half of the Soviet Union.31Alexander led the cryptanalytic attack on Coleridge. He had returned to code-breaking work after a brief, unhappy stint working as a financier in London because he could notstand a job “that involved a black jacket and striped trousers.”32Assisting Alexander on the other side of the Atlantic was a team of U.S. Navy code breakers led by one of the best machine cryptanalysts in America, Francis “Frank” Raven. A 1934 graduate of Yale University, Raven had worked as the assistant manager of the Allegheny Ludlum Steel Company in Pittsburgh before joining the navy COMINT organization in 1942. An incredibly talented cryptanalyst, during the war he had been instrumental in solving a number of Japanesenavy cipher machine systems.33 The Coleridge decrypts were found to contain reams of administrative traffic for the Soviet military, but when analyzed, they yielded vitally important information about its order of battle, training activities, and logistical matters.34

  At about the same time, the Anglo-American cryptanalysts made their first entry into a third Russian cipher machine system, designated Longfellow. By July 1946, a copy of the Longfellow cipher machine had been constructed by U.S. Navy cryptanalysts in Washington, D.C., based on technical specifications provided by the British cryptanalysts who had solved the system, but the solution of the cipher settings used on the Longfellow machine required several more months of work. Finally, in February 1947 a team of British cryptanalysts led by Gerry Morgan and a team of U.S. Navy analysts in Washington, headed by Commander Howard Campaigne, together solved the encryption system used by the Soviet army’s Longfellow cipher machine system.35

  But the value of the decrypts of Longfellow traffic that were just beginning to be produced in the spring of 1947 was eclipsed by the ever-rising volume of translations being produced across the Atlantic at GCHQ through the exploitation of the Coleridge cipher machine. These decrypts proved to be so valuable that, according to a report by the U.S. Navy liaison officer assigned to GCHQ, Coleridge was “the most important, high-level system from which current intelligence may be produced and is so in fact regarded here.”36

  The net result was that by the spring of 1947, translations of decrypted messages from all three systems were being produced in quantity. At Arlington Hall, the ASA cryptanalysts alone were churning out 341 decrypts a day, seven days a week, 365 days a year, most of which were derived from Russian radio intercepts. 37 By early 1949, more than 12,500 translations of decrypted Russian army radio messages had been published by ASA and sent to intelligence consumers in Washington.38

  The Anglo-American cryptanalysts were also experiencing considerable success in solving the cipher systems used by the Soviet navy. By early 1947, a number of Russian navy ciphers used in the Far East had been successfully solved, largely because the two Russian fleets operating in the Pacific were forced by geography to use radio to communicate with Moscow instead of secure teletype landlines. This allowed U.S. Navy listening posts in the Far East to easily intercept the radio traffic sent between these headquarters and Moscow. There was also some success in reading the cipher systems used by the Soviet fleets in the Baltic Sea, as well as the ciphers used by the Black Sea fleet and the Caspian Sea flotilla. By February 1949, U.S. Navy cryptanalysts had produced more than twenty-one thousand decrypts of Soviet naval message traffic, which was almost double the number of decrypts of Russian army traffic produced by ASA.39

  A number of the Soviet air force’s operational ciphers were also quickly solved. In 1947, ASA cryptanalysts solved one of the operational cipher systems used by the Soviet air force headquarters in Moscow to communicate with its subordinate commands throughout the Soviet Union and Eastern Eu rope, as well as several variants of this system.40In the Far East, U.S. Army cryptanalysts in Japan were reading the encrypted radio traffic of the Soviet Ninth Air Army at Ussurijsk/Vozdvizhenka and the Tenth Air Army at Khabarovsk.41

  In room 2409 at Arlington Hall, a brilliant thirty-four-year-old former Japanese linguist and cryptanalyst named Meredith Knox Gardner was making spectacular progress solving the ciphers that had been used during World War II by the Soviet civilian intelligence service (its military counterpart was the GRU), then called the NKGB, to communicate with its rezidenturas in the United States. In later years, this work would be part of Venona program. In December 1946, Gardner solved part of a 1944 NKGB message that gave the names of some of the more prominent American scientists working on the Manhattan Project, the American war time atomic bomb program. The decrypt was deemed so important that army chief of staff Omar Bradley was personally briefed on the contents of the message. Five months later, in May 1947, Gardner solved part of a message sent from the NKGB’s New York rezidentura on December 13, 1944, which showed that an agent within the U.S. Army General Staff in Washington had provided the Soviets with highly classified military information. Unfortunately, Gardner was not able to deduce anything further as to the agent’s true identity from the fragmentary decrypt. By August 1947, new decrypts provided the first evidence that an extensive Soviet spy ring was operating in Australia during World War II, which set off alarm bells in both Washington and London. Gardner was able to report that the decrypts contained the cryptonyms of dozens, perhaps hundreds, of Soviet agents operating in the United States, Australia, and Sweden during the war. But the report also clearly showed that Gardner had only made partial headway into the Soviet codebook, and that the results of his work were still very fragmentary.42

  Taken together, these decrypts opened up a wide array of Soviet military and civilian targets for exploitation by the information-starved intelligence analysts in both Washington and London. An NSA historical monograph notes, “ASA in the post–World War II period had broken messages used by the Soviet armed forces, police and industry, and was building a remarkably complete picture of the Soviet national security posture.”43This is confirmed by material obtained by researchers from the former KGB archives in Moscow, which reveals that the Anglo-American COMINT organizations were deriving from these decrypts a great deal of valuable intelligence about the strength and capabilities of the Soviet armed forces, the production capacity of various branches of Soviet industry, and even the super-secret work that the Soviets were conducting in the field of atomic energy.44

  Former NSA officials have stated in interviews that the first postwar crisis in which COMINT played an important role was the 1948 Berlin Crisis.45 Ultimately, it was COMINT that showed that the Soviets had no intention of launching an attack on West Berlin or West Germany. The initial stage of the Berlin Crisis was actually a Russian feint.46COMINT also provided valuable data during the second part of the crisis, when on June 26, 1948, the Soviet’s cut off all access to West Berlin, forcing the United States and Britain to begin a massive airlift to keep West Berlin supplied with foodstuffs and coal for heating. Careful monitoring of Soviet communications indicated that the Russians would not interfere with the airlift.47

  Black Friday

  During President Truman’s October 1948 nationwide whistle-stop train tour in his uphill battle for reelection against Governor Thomas Dewey, the U.S. government was at a virtualstandstill. On the afternoon of Friday, October 29, just as Truman was preparing to deliver a fiery campaign speech at the Brooklyn Academy of Music in New York City, the Russian government and military executed a massive change of virtually all of their cipher systems. On that day, referred to within NSA as Black Friday, and continuing for several months thereafter, all of the cipher systems used on Soviet military and internal-security radio networks, including all mainline Soviet military, naval, and police radio nets, were changed to new, unbreakable systems. The Russians also changed all their radio call signs and operating frequencies and replaced all of the cipher machines that the Americans and British
had solved, and even some they hadn’t, with newer and more sophisticated cipher machines that were to defy the ability of American and British cryptanalysts to solve them for almost thirty years, until the tenure of Admiral Bobby Ray Inman in the late 1970s.48

  Black Friday was an unmitigated disaster, inflicting massive and irreparable damage on the Anglo-American SIGINT organizations’ efforts against the USSR, killing off virtually all of the productive intelligence sources that were then available to them regarding what was going on inside the Soviet Union and rendering useless most of four years’ hard work by thousands of American and British cryptanalysts, linguists, and traffic analysts. The loss of so many critically important high-level intelligence sources in such a short space of time was, as NSA historians have aptly described it, “perhaps the most significant intelligence loss in U.S. history.” And more important, it marked the beginning of an eight-year period when reliable intelligence about what was occurring inside the USSR was practically non exis tent.49

  The sudden loss of so many productive intelligence sources was not the only damage that can be directly attributed to the Black Friday blackout. In the months that followed, the Anglo-American code breakers discovered that they now faced two new and seemingly insurmountable obstacles that threatened to keep them deaf, dumb, and blind for years. First, there was far less high-level Soviet government and military radio traffic than prior to Black Friday because the Russians had switched much of their military communication to telegraph lines or buried cables, which was a simple and effective way of keeping this traffic away from the American and British radio intercept operators. Moreover, the high-level Russian radio traffic that could still be intercepted was proving to be nearly impossible to crack because of the new cipher machines and unbreakable cipher systems that were introduced on all key radio circuits. The Russians also implemented tough communications security practices and procedures and draconian rules and regulations governing the encryption of radio communications traffic, and radio security discipline was suddenly rigorously and ruthlessly enforced. Facing potential death sentences for failing to comply with the new regulations, Russian radio operators suddenly began making fewer mistakes in the encoding and decoding of messages, and operator chatter disappeared almost completely from the airwaves. It was also at about this time that the Russian military and key Soviet government ministries began encrypting their telephone calls using a newly developed voice-scrambling device called Vhe Che (“High Frequency”), which further degraded the ability of the Anglo-American SIGINT personnel to access even low-level Soviet communications. It would eventually be discovered that the Russians had made their massive shift because William Weisband, a forty-year-old Russian linguist with ASA, had told the KGB everything that he knew about ASA’s Russian code-breaking efforts at Arlington Hall. (For reasons of security, Weisband was not put on trial for espionage.)

  Decades later, at a Central Intelligence Agency conference on Venona, Meredith Gardner, an intensely private and taciturn man, did not vent his feelings about Weisband, even though he had done grave damage to Gardner’s work on Venona. But Gardner’s boss, Frank Rowlett, was not so shy in an interview before his death, calling Weisband “the traitor that got away.”50

  Unfortunately, internecine warfare within the upper echelons of the U.S. intelligence community at the time got in the way of putting stronger security safeguards into effect— despite the damage that a middle-level employee like Weisband had done to America’s SIGINT effort. Four years later, a 1952 review found that “very little had been done” to implement the 1948 recommendations for strengthening security practices within the U.S. cryptologic community.51

  The Creation of the Armed Forces Security Agency

  At the same time that the U.S. and British intelligence communities were reeling from Black Friday, several new institutional actors shoved their way into the battered U.S. cryptologic community. On October 20, 1948, the newly in dependent U.S. Air Force formally activated its own COMINT collection organization, the U.S. Air Force Security Service (USAFSS).52It immediately became responsible for COMINT coverage of the entire Soviet air force and air defense system, including the strategic bombers of the Soviet Long Range Air Force. But the ability of USAFSS to perform this vital mission was practically non existent at the time owing to a severe shortage of manpower and equipment, largely because the U.S. Air Force headquarters staff in Washington was slow to provide the necessary resources that the COMINT organization so desperately needed. As a result, by the end of 1949, USAFSS was only operating thirty-five COMINT intercept positions in the U.S. and overseas, which was far short of what was expected of it. By December 1949, the situation was so serious that the chief of USAF Intelligence was forced to report that USAFSS’s COMINT capability was “presently negligible and will continue to be negligible for an unwarranted period of time unless immediate steps are taken to change the present low priority on equipment and personnel assigned to the Air Force Security Services.”53

  Seven months later, on May 20, 1949, Secretary of Defense Louis Johnson issued a Top Secret directive creating the Armed Forces Security Agency (AFSA), which was given the responsibility for the direction and control of all U.S. communications intelligence and communications security activities except for tactical cryptologic activities, which remained under the control of the army, navy, and air force.54

  AFSA was a fatally flawed organization from its inception. Its funding was grossly inadequate when compared with the significantly higher level of funding given to the CIA, which had been created two years earlier in 1947.55 The military services then systematically stripped AFSA of virtually all of the authority that it had originally been granted. As a result, by the summer of 1950, AFSA found itself powerless and completely dependent on the military for all of its money, radio intercept facilities, personnel, equipment, communications, and logistical support.56Then, taking full advantage of AFSA’s weakened state, the military services got key portions of their COMINT missions exempted from its authority. With no means of compelling the other services to comply, including no control over the budgets of the three military SIGINT units, AFSA was forced to humble itself and negotiate on bent-knee agreements with the services that gave even more power away to them.57

  It is clear now that many of AFSA’s problems can be traced directly to its first director, Rear Admiral Earl Stone, who did not possess the combative personality desperately needed to force the branches of the military to cooperate in order to make AFSA work. By the time he left office in July 1951, astanding joke among his subordinates was that Stone’s authority extended only as far as the front door of his office, and even that was subject to debate.58Looking back on Stone’s sad two-year tenure as director of AFSA, one of his senior deputies, Captain Wesley Wright, said that the decision to give the job to Stone in the first place “was a horrible thing to do.”59

  Jack Gurin’s War

  Declassified documents make clear that AFSA’s legion of internal management woes, although serious, were the least of its problems. From the moment it was born, AFSA inherited, as a declassified NSA history puts it, “a Soviet problem that was in miserable shape.”60

  AFSA had only one source of intelligence left that offered any insight into what was going on inside the Soviet Union: intercepts of low-level, unencrypted Soviet administrative radio traffic and commercial tele grams, which were generally referred to as “plaintext” within the Anglo-American intelligence communities. A declassified NSA historical report notes, “Out of this devastation, Russian plaintext communications emerged as the principal source of intelligence on our primary Cold War adversary.”61Outside of plaintext, the only other source for information on what was going on behind the iron curtain came from Traffic Analysis, where analysts studied the now-unreadable intercepts to try to derive intelligence from the message “externals.”

  Plaintext intercepts had been ignored as an intelligence source since the end of World War II; after Black Friday, ever
ything changed. Since high-level Russian communications traffic could no longer be read, the previously deprecated Russian plaintext intercepts being processed in Arlington Hall’s room 1501-B suddenly became of critical importance for U.S. SIGINT. Overnight, the twenty-seven-year-old chief of the AFSA plaintext unit, Jacob “Jack” Gurin, became a leading figure within the U.S. intelligence community.62Now the world was beating a path to his door.

  The Blackout Curtain

  In addition to focusing on plain text intercepts, the other principal problem that the newly created AFSA had to confront was how to revamp itself and at the same time try to repair the damage caused by the Black Friday blackout. The U.S. Communications Intelligence Board quickly conducted a study, which determined that an additional 160 intercept positions and 650 intercept operators were needed just to meet minimum coverage requirements. The study also found that “currently allowed personnel are not sufficient for these and other important tasks.”63

  The question became, how should the scarce COMINT collection resources available be reallocated? In early 1949, the U.S. Army and Navy COMINT organizations began systematically diverting personnel and equipment resources away from non-Soviet targets in order to strengthen the Soviet COMINT effort. By the summer of 1949, 71 percent of all American radio intercept personnel and 60 percent of all COMINT processing personnel were working on the “Soviet problem”—at the expense of coverage of other countries, including AFSA’s targets in the Far East, most significantly mainland China. Declassified documents show that the number of AFSA analysts and linguists assigned to Asian problems had declined from 261 to 112 personnel by the end of 1949. Work on all other nations in the Far East was either abandoned completely or drastically reduced.64