Code Warriors

by Stephen Budiansky

  “The only thing you’re proving” by continuing to drop agents into Soviet-controlled territory, acidly remarked Truscott’s aide at one confrontational meeting with the local CIA chief, “is the law of gravity.”6

  But still CIA officials tried to defend the effort. “Even if they don’t send back good intelligence, we’re causing the Russians a lot of headaches,” insisted one. Dulles, confronted the following year by an aide who told him bluntly, “Our operations have failed and there is no alternative to offer,” shot back, “You can’t say that.” Later, when the failures became truly undeniable, Dulles, who would become CIA director in 1952, still tried to justify it all. “At least we’re getting the kind of experience we need for the next war,” he insisted.7

  In fact, as it would later be discovered, every single one of the 100 agents dispatched by E-boat to the Baltics by the SIS from 1944 to 1954 was under Moscow’s control, wittingly or unwittingly, from the very start. So too were the 150 agents parachuted into Ukraine from 1949 to 1954—at least those not immediately captured and shot after having been lured into the web of KGB deception. So too was an entire Polish “resistance” movement fabricated by the Soviets in 1949 that duped the CIA into sending radios, weapons, supplies, agents, and a million dollars in gold—all humiliatingly exposed in a mocking two-hour diatribe broadcast on Polish radio in December 1952. It was not until 1956 that SIS finally abandoned the last of these fiascoes, an operation that trained agents in Turkey to be parachuted into the Soviet Caucasus. Dozens went in. The failure rate was 100 percent.8

  Most horrific of all was the fate of scores of Albanian émigrés recruited by SIS and CIA from 1947 to 1952 to gather intelligence and lay the groundwork for an uprising against the Communist government of Enver Hoxha. “Operation Valuable,” it was optimistically named. All were betrayed by Kim Philby, who in many cases knew in advance the exact time and location of their landing by sea or air, even the names of individual men, and passed the information to his Soviet controllers, who in turn passed it to the Albanians, who were ready and waiting to ambush the infiltrators. Some were killed in a hail of bullets the moment they touched the ground or came ashore; others were hunted down and butchered by the police and militia—burned to death in a barricaded building, dragged to a bloody pulp tied behind a jeep—or captured, tortured, and put on display in show trials in which they “confessed” their guilt in zombie-like voices.9

  “There was not one successful operation,” said a U.S. Army intelligence officer in Germany, looking over the whole sordid history of the effort to slip spies behind the Iron Curtain. Anthony Cavendish, who ran many of the SIS parachute infiltrations, admitted afterward that they “must seem now nothing more than a catalogue of disasters.”10 A 1955 review ordered by President Eisenhower of intelligence sources that might be able to provide strategic warning of a Soviet nuclear attack on the United States found that virtually no useful information had been obtained from “classic covert operations inside Russia”:

  The security zones at the border, the general restrictions in the interior, the thousands of security police, and the innumerable informers among the population are brutally effective in limiting the infiltration, exfiltration, and usefulness of agents. Therefore, we must more and more depend upon science and technology.11

  “Pitiful,” concluded George Kistiakowsky, the Ukrainian-born Harvard physical chemist who had led a key program in the Manhattan Project and in the 1950s became Eisenhower’s science adviser, helping to shape U.S. defense and intelligence strategy. “The time of Mata Hari has passed.”12

  1

  The Russian Problem

  The men and women who aimed to supplant Mata Hari with the steady, efficient, reliable, and ever so much safer methods of science and technology began their first tentative foray into what they would always call, with a certain clinical detachment, “the Russian problem” on February 1, 1943. Almost all of the codebreakers who would work over the next several years to achieve the first breaks into the labyrinth of Russian communications secrets were newcomers to the business—which made them no different from the other thirteen thousand new recruits, military and civilian, who would by the war’s end swell the explosively growing ranks of the U.S. Army and Navy signals intelligence headquarters in Washington.1 The variety of their backgrounds was extraordinary: career officers and new draftees, young women math majors just out of Smith or Vassar, partners of white-shoe New York law firms, electrical engineers from MIT, the entire ship’s band from the battleship California after it was torpedoed by the Japanese in the attack on Pearl Harbor, winners of puzzle competitions, radio hobbyists, farm boys from Wisconsin, world-traveling ex-missionaries, and one of the world’s foremost experts on the cuneiform tablets of ancient Assyria.

  In June 1943, Cecil Phillips was an eighteen-year-old chemistry student at the University of North Carolina who had just been rejected for the draft because of flat feet (“to my great pleasure and surprise,” he later admitted); with no plans for the summer, he wandered into the U.S. Employment Service office in his hometown of Asheville, in the Blue Ridge Mountains, to see if he could get a job. The person there told him there was a lieutenant from the Army Signal Corps over at the town post office who had a large quota of clerk positions to fill.

  “How would you like to go to Washington and be a cryptographer?” the lieutenant asked him.

  “That sounds interesting,” Phillips answered.

  The lieutenant, clearly surprised that someone actually knew what he was talking about, blurted out, “You mean you know what that means?”

  Phillips did, having once owned a Little Orphan Annie decoder ring, though that was about as far as his knowledge went. But it was good enough for the lieutenant, who administered him a general aptitude test on the spot, signed him up as a $1,440-a-year GS-2 junior clerk, and told him to report in a week to an address in Arlington, Virginia.2

  Brought in under nearly identical circumstances was a young home economics instructor, Gene Grabeel, who was teaching high school near Lynchburg in central Virginia and dissatisfied with her job when she met a young Army officer in the post office who was looking for college graduates to go to work at an undisclosed location near Washington, to do a job he could not offer any details about. (The officer, an infantry lieutenant who just days earlier had been posted with the First Army at Governors Island, New York, did not know himself what the work involved. Driven largely by the need to process volumes of Japanese army traffic that had suddenly become readable due to breakthroughs in several systems, the Army would hire four thousand new civilian employees for its signals intelligence operation in 1943 alone. In the rush to meet such burgeoning manpower requirements, the recruiters were as green as everyone else. The lieutenant had been ordered to report to Arlington Hall on Monday the week of Thanksgiving in November 1942; he spent the next day filling out administrative paperwork; Wednesday he was given a crash course on recruiting procedures; and on Thanksgiving morning he found himself at the post office in Lynchburg trying to collect warm bodies, without even having had a chance to find out what his new outfit did.) Grabeel had been thinking about trying to get a job with the federal government and asked her father what he thought of the idea. He told her she might as well “go to Washington for six months and shuffle papers.” She was off to the capital as soon as she found a replacement teacher to take over for her.3

  Arlington Hall Junior College had been a finishing school for girls before the Army abruptly seized it under emergency war powers in June 1942. The site was convenient to the Pentagon, the colossal new Army and Navy headquarters, soon to be the largest office building in the world, which was arising on a marshy flat along the Potomac River less than two miles away, just across Arlington National Cemetery. The school’s one hundred acres also offered security and ample room for expansion. One of the Arlington Hall recruiters found a prewar postcard of the school, depicting a stately residence hall, manicured lawns, tennis courts, and indoor and outdo
or horseback riding arenas and stables, and shamelessly deployed it in his recruiting pitch. What the new arrivals found instead were two huge, dreary, hastily erected warehouselike office blocks built in the traditional U.S. government style with long corridors and a series of perpendicular wings, surrounded by barbed wire and guardhouses. Workers sat side by side at long tables arrayed in rows. Air-conditioning remained a dream for the future; the buildings were sweltering in the humid Washington summers and overrun year-round with legions of rodents. There were no horses in the deserted stables, but there was a drill field and barracks for the enlisted men. The lieutenant who had paraded the prewar postcard avoided eye contact with his recruits when he later encountered them at Arlington Hall.4

  More than 70 percent of the staff at Arlington Hall were civilians, and by the war’s end more than 90 percent of those were women. A similar balance of the sexes quickly took hold at the Navy’s signals intelligence headquarters, across the Potomac River. The Navy had a deep tradition of never permitting a situation to arise where an officer might have to take orders from a civilian, and insisted on putting all of its new hires in uniform. But with its establishment in summer 1942 of the WAVES—Women Accepted for Voluntary Emergency Service, which allowed women to serve in the Navy as officers and enlisted personnel—the service was also able to freely recruit women for codebreaking duty, and some 80 percent of its cryptanalysts by the war’s end were female.5

  The joke making the rounds in Washington the first year of the war was that if the Army and Navy could capture enemy territory as fast as they were seizing it in the nation’s capital the fighting would be over in a couple of weeks. Not long after the Army claimed Arlington Hall, the Navy took possession of its own girls’ school, Mount Vernon Academy on Nebraska Avenue in northwest Washington. In February 1943 the Naval Communications Intelligence Section (known as Op-20-G in the arcane numbering system the Navy’s bureaucratic administrators had devised to designate its myriad branches and offices) moved into its new home at what was now officially called the Naval Communications Annex; construction crews went to work at once ripping out the school’s graceful colonnaded walkways to make room for functional buildings; the headmistress’s residence, with its elegant polished hardwood floors, was converted into the post exchange, selling Cokes and cigarettes; and a double line of wire fences went up around the perimeter, guarded by marines patrolling with submachine guns.6

  —

  That summer there began arriving at Nebraska Avenue’s Building 4 the first of what would soon be a phalanx of one hundred massive electromechanical calculating machines. Built by the National Cash Register company at its factory in Dayton, Ohio, at a staggering total cost of $6 million, the “bombes,” as they were called, weighed two and a half tons apiece and housed sixty-four motor-driven wheels whose electrical contacts spun at speeds of up to 1725 rpm; when they were all running together they drew a quarter of a megawatt of electricity, enough to power a thousand homes.7

  Their internal electrical logic was the genius of the eccentric British mathematician Alan Turing. Building on prewar work by a small team of Polish cryptanalysts who shared their results with the British just weeks before the Nazi invasion, Turing in the fall of 1939 developed a comprehensive mathematical solution to the Nazis’ legendary Enigma cipher machine. By guessing a few probable words contained in an enciphered Enigma message, he showed how to eliminate in one mathematical-logical leap about 1014 of the permutations that the Enigma’s scrambling machinery relied upon to baffle any would-be codebreaker. That left only a few hundred thousand possibilities to test to recover the unique daily setting of the scrambling rotors used on each of the high-level radio networks that employed the Enigma, the key to unlocking thousands of extremely secret signals a day: orders to U-boats prowling the Atlantic, reports on the dispositions and plans of Rommel’s troops in the North African desert or the state of the Nazis’ Western Wall defenses along the coast of France. The whirring wheels of the U.S. Navy’s bombes, recreating the internal wiring of the rotors of the actual Enigma machines, could in twenty minutes try every possible starting setting of the Enigma (each of the three or four rotors of the Enigma could be set at twenty-six different positions, which meant there were 26 × 26 × 26 × 26 = 456,976 possibilities in the case of the four-rotor version used by the German U-boats). When the bombe reached a position consistent with a chain of letters built up from a short sequence of matching plaintext and cipher text, a circuit was completed through a series of interconnected cables used to program the bombe for each test, causing an electrical relay to trip and triggering a clutch and brake that would bring the device jolting to a halt, revealing the daily jackpot.8

  At the height of the Battle of the Atlantic nearly half of the staff at Nebraska Avenue was working on the Enigma problem; most of the rest were trying to keep up with the huge volume of Japanese navy traffic and the ever-changing complexities of its code systems, most of which used codebooks rather than cipher machines like the Enigma.9, *1 IBM punch card machines turned out to be well suited to the exhaustive cataloging and searching required to break into these Japanese codes. The process involved combing through intercepted messages to hunt for repetitions of their four-or five-digit numerical cipher groups—a possible sign that two different messages had been prepared with the same string of obscuring key. Compared to a modern computer, punch card machines were undeniably primitive, but they could carry out massive data searches that would have overwhelmed a human being. Cards punched with the code groups of tens or even hundreds of thousands of messages could be automatically placed in numerical order by an IBM card sorting machine, then printed out in massive catalogs by a printing tabulator to be scanned by eye for any repetitions. The Japanese army codes worked much the same way as the Japanese navy codes, and by 1943 Arlington Hall and Nebraska Avenue were operating hundreds of IBM machines and paying the company three-quarters of a million dollars a year in rental fees, while burning through hundreds of thousands of punch cards a month.10

  What had begun as little more than tiny back-office research groups (at the outbreak of World War II in September 1939 the Army’s Washington signals intelligence staff was nineteen people, the Navy’s thirty-six) had become veritable decryption factories, working round-the-clock shifts and tied to a sprawling global network of outstations that fed an uninterrupted stream of intercepted communications. Besides the thirteen thousand workers in Washington there were thousands more in the field manning a dozen principal intercept posts from Winter Harbor, Maine, to Bainbridge Island, Washington, and to points around the world as far-flung as the Aleutian Islands, the Canal Zone, Guam, and Recife, Brazil, each bristling with antennas and equipped with multiple shortwave receiver sets.11 Teams of enlisted men took down Morse code messages by hand and then retransmitted the copied traffic via encrypted landline or radio teleprinter links back to Washington or other cryptanalytic processing centers in Hawaii and Australia.*2

  By October 1943 Arlington Hall had up and running a semiautomated decryption processing line for Japanese army traffic that punched incoming teleprinter messages onto paper tape, converted the paper tape to IBM cards, matched the resulting decks of punch cards with other sets of cards punched with the corresponding sequence of cipher key, subtracted one from the other to reveal the underlying code groups and punched those on a third set of cards, and then used a library of cards containing code groups whose dictionary meanings had been recovered to print out the complete decoded message. In some cases Arlington Hall was reading a message before its intended Japanese recipient, who had to manually flip through his codebooks and work out the message on paper and pencil, could do so.12

  The size and extent of the American wartime cryptanalytic empire reflected the global reach of the conflict, but it also reflected the global nature of communications, and thus of intelligence opportunities ripe to be exploited. One of the most valuable sources of information on German preparations for the Allied D-Day landings would prove
to be the reports of Japan’s ambassador in Berlin, in cables radioed six thousand miles back to his government in Tokyo using the Japanese diplomatic system known to the American codebreakers as Purple, a cipher machine they had cracked in 1940, sight unseen, in one of their most stunning achievements of pure mathematical cryptanalysis.

  Some of the size of the wartime enterprise, to be sure, also reflected what the British liaison officer sent to Arlington Hall, Captain Geoffrey Stevens, diagnosed with ill-concealed irritation as the American genius for constructing “hopelessly overorganized” operations.13 The work involved a huge amount of painstaking drudgery, and the American solution was to parcel out tasks exactly like a factory assembly line. When Cecil Phillips arrived at Arlington Hall on June 22, 1943, his first job was to stamp the date on incoming messages. Having demonstrated his competence at that task, he was promoted to stapling. But his first boss saw something in the high school graduate that perhaps Phillips himself did not, and started setting aside an hour or two a day to teach him the rudiments of cryptanalysis. On May 1, 1944, Phillips was led to the back of one of the wings on the second floor of B Building, where a fifty-by-fifty-foot area had been partitioned off with plywood screens from the rest of the open wing. A small opening between the screens, just large enough for one person to squeeze through, led past a desk where an Army captain sat with his back to the entrance, keeping a sharp eye on the several dozen people at work at long tables.14

  Among them were Gene Grabeel, who had been assigned to this mysterious unit during her fourth week on the job; Richard Hallock, the Assyrian linguist and archaeologist from the University of Chicago, now an Army lieutenant; another newly commissioned lieutenant, Ferdinand Coudert, a partner in his family’s venerable New York international law firm, Coudert Brothers (among its clients had been the czar of Russia and the French and British governments), who held an MA degree in Slavic studies from Harvard and knew Russian, French, German, Serbo-Croatian, Bulgarian, and Japanese; and Frank Lewis and Genevieve Grotjan Feinstein, two of the comparatively veteran cryptanalysts of the Signal Intelligence Service, both of whom had been hired in a modest expansion of the service made possible with funds appropriated in response to the “limited national emergency” declared by FDR upon the outbreak of the war in Europe. Feinstein, a quiet mathematician, had made the crucial break into Purple in 1940; Lewis, a voluble polymath and musician, contributed to the solution of the Japanese army codes and a seemingly impenetrable German diplomatic one-time-pad system. (Lewis was to gain a devoted following in the outside world for the notoriously difficult cryptic crosswords he composed every week for six decades for the left-wing intellectual journal the Nation starting in 1947.) The captain in charge of the section was William Smith, a classmate of Coudert’s at Harvard who had recently been working as an editor for the famous one-volume Columbia Encyclopedia at Columbia University Press.15