A hapless cloak-and-dagger operation in Lisbon set off a chain of intrigue that was part of “Wild Bill” Donovan’s ultimate downfall.
The story that came to be known as the Lisbon affair is one of the most enduring cautionary tales in modern intelligence history. It became a legendary warning of the damage that can be done when overeager amateurs, in their zeal to gather enemy secrets, inadvertently tip off the enemy to critical leaks in their own security. No less a personage than the army’s chief of staff, Gen. George C. Marshall, argued to the president that the ham-handed attempt by agents of the Office of Strategic Services to steal cipher material out of a wastebasket in the Japanese embassy in Lisbon in April 1943 had dealt a fatal blow to American intelligence. Alerted that its codes were now insecure, Marshall insisted, Tokyo had responded by changing the cipher used by its military attachés around the world, and the replacement system had so flummoxed American code breakers that a full year later they remained unable to read these vital messages. Only dumb luck, or the obtuseness of the Japanese, the story went, had prevented an even more sweeping upgrade of Japanese diplomatic ciphers that would have resulted in a total blackout of this critical intelligence source during the war.
The affair reverberated for years afterward in American intelligence circles. It helped convince the Joint Chiefs of Staff to restrict OSS operations and keep its flamboyant director, William “Wild Bill” Donovan, on a short leash. It helped reinforce the popular image of Donovan as a loose cannon and his agents as incompetent amateurs. And ultimately it contributed to the abolition of the OSS at the end of the war.
But the actual story behind what happened in Lisbon in the spring of 1943 is a far more Byzantine tale of intrigue, in which the real rivals were not so much the Allies and the Axis but warring spy agencies within the American capital. Many of the facts that Marshall, and later President Harry Truman, firmly believed about the case turned out to be false or exaggerated.
The real truth is that to those in the U.S. Army and the State Department who were out to get Donovan, what began as a trivial incident in Lisbon proved a golden opportunity they were not about to let slip through their hands. And had it not been for an almost bizarre series of spy vs. spy mishaps in which American and Axis code breakers, each reading the other’s diplomatic messages, amplified and distorted the original wastebasket incident, the whole matter would rightly have been relegated to an obscure footnote to the annals of intelligence history. As it was, it shook the highest levels of power in Washington and threatened to upset a delicate and highly secret arrangement between the Americans and British on sharing the greatest intelligence coup of the war, the breaking of the German Enigma cipher machine.
From its creation in June 1942, the Office of Strategic Services had attracted the enmity of other intelligence agencies that considered the new espionage organization a threat to their status and prerogatives. After all, why would the country need a new service unless the old ones weren’t up to the job? And if the old services weren’t up to the job, then why keep them?
Part of the problem was William Donovan himself. A Medal of Honor winner in the First World War and a Wall Street lawyer and sometime Republican office holder between the wars, Wild Bill had been sent to London by President Franklin Roosevelt in 1940 to evaluate Britain’s prospects for successfully resisting a Germany that was now master of the European continent. Donovan had returned convinced of British resolution and the need for the United States to improve to its own security. Especially impressed by what he was allowed to see of British intelligence services, Donovan had been converted to the belief that a large and centralized intelligence service dedicated to collecting, collating, and disseminating information from around the world was a prerequisite for national security, a position he pushed on Roosevelt at every opportunity.
Like many converts Donovan became a true believer whose passion and purpose irritated skeptics, particularly those in established intelligence organizations such as the army’s Military Intelligence Division, the Office of Naval Intelligence, and the Federal Bureau of Investigation. When, in 1941, the president created a new intelligence office, the Coordinator of Information, and appointed Donovan its director, the older organizations withheld cooperation and sought every opportunity to undermine the interloper. The hostility continued even after Pearl Harbor demonstrated the deficiencies of American intelligence. When Roosevelt established the OSS as a replacement for the COI, and gave the office the chief responsibility for overseas intelligence work, the enmity and jealousy only grew.
In the bureaucratic jungle of wartime Washington, the OSS faced no more dangerous and implacable opponent than Brig. Gen. George V. Strong, the assistant chief of staff for military intelligence. In part it was personal. One of the last serving officers who had actually campaigned in the Indian Wars, Strong was notorious for his volatile temper and combative assertiveness in protecting the reputation and programs of military intelligence. In Washington, he had earned the nickname “King George” for his imperious manner. Jealous of Donovan’s meteoric rise and resentful of his political connections, particularly his direct access to the White House, Strong seized any opportunity to embarrass the OSS director and confound his plans. Strong had earlier fought a knockdown battle to prevent the COI from receiving any intelligence derived from breaking coded messages, arguing that Donovan and his minions were too unreliable to be entrusted with such valuable secrets.
But it was never just about personalities. Strong considered the very existence of the OSS an affront to army intelligence, and would have felt so no matter who was the director. Diligently patrolling the boundaries of his bureaucratic turf, the assistant chief of staff had fought unsuccessfully to block the creation of the new intelligence entity in the spring of 1942, and then worked to undermine its reputation. He dismissed Wild Bill’s crew as dilettantes, misfits, and shirkers. But secretly he feared that a successful, independent competitor would eventually divert attention, resources, and missions from the army’s own intelligence service.
Strong wanted and expected the OSS to fail, and in the spring of 1943 failure was in the air. At that same moment, the OSS was trying to dampen the flames from another embarrassing incident: the American ambassador in Spain had called for an investigation and a curtailment of OSS clandestine activities after the spy agency sent a shipment of pistols to the OSS station there in a crate addressed to the ambassador—and accompanied by a manifest clearly stating its contents. Donovan, called on the carpet by the Joint Chiefs of Staff, was forced to restrict his service’s operations in Spain. It was a low point for the OSS, and Strong undoubtedly saw in the Lisbon imbroglio an opportunity to further humiliate Donovan and weaken his already beleaguered agency.
The Lisbon affair began when the OSS station in the Portuguese capital recruited a local citizen who worked inside the Japanese embassy as a messenger for the naval attaché. Sometime in early April 1943, this source delivered to his American controller some crumpled papers he had retrieved from the wastebasket of his employer. Written on the pages were both numbers and Japanese characters. Believing the writing represented a cipher, the Lisbon station forwarded the materials to OSS headquarters in Washington. On May 20, the OSS passed the pages to the Signal Corps, the service arm that included the Signal Security Service, the army’s top-secret code breaking organization.
Four days later the Signal Corps responded, thanking the OSS for its effort, but disclaiming any interest in the papers. The pages did indeed contain a cipher, but one already well known to army code breakers. Used only for routine, low-grade communications, the cipher was of so little importance that the code breakers had some time earlier abandoned any interest in it. For intelligence purposes the material from Lisbon was worthless. The purloined papers were buried in the archive and everyone considered the matter ended. Everyone was wrong.
The first sign of trouble came in the beginning of July, when the U.S. Army code breakers decrypted a flurry of messages passing between Tokyo and its embassies in Lisbon and Madrid. The messages reported that the Japanese had just learned from Italian intelligence that the Americans had somehow penetrated the Lisbon embassy and might have gained access to its ciphers. The messages ordered the ambassador in the Portuguese capital to report immediately on the security measures in place to protect his ciphers. The ambassador in Madrid was ordered to dispatch a senior officer to review security arrangements in Lisbon.
The initial alarm that this event set off in American intelligence circles was genuine enough. America’s greatest intelligence coup of the war—it was one of the most closely-held secrets of the American government—was the cracking of the Japanese cipher machine code-named Purple, used to encrypt Tokyo’s highest-level diplomatic communications. It was an astonishing feat of mathematical cryptanalysis: the U.S. Army code breakers had reconstructed the inner workings of the machine without ever once laying eyes on one, just by analyzing patterns in the coded messages it produced. The intelligence gleaned from reading the most secret communications of the Japanese foreign ministry was given the code name Magic, reflecting the unprecedented access it gave Washington to its enemy’s plans and actions. Magic in particular provided a front row seat from which to observe Japanese intelligence efforts, much of which was run out of Japanese embassies and consulates and revealed in Japanese diplomatic traffic.
If Japan suspected that any of its ciphers had been compromised, then Tokyo might very well replace all of its current cryptosystems with new ones, including Purple. The Signal Security Service had recently noted that Japan’s foreign ministry had suddenly stopped using a medium-grade cipher known to the code breakers as J-19. It might just have been a coincidence—but it also might have been the first of a wholesale replacement of Japanese ciphers ordered in the wake of the Italian warning.
On July 3, Col. Carter Clarke, chief of Special Branch—the office in the Signal Security Service responsible for disseminating communications intelligence within the War Department— informed General Strong about the pilfered papers and the alarm in Tokyo, Lisbon, and Madrid. He also raised the possibility that it was the OSS operation that had jeopardized the wall of secrecy surrounding American code breaking successes.
If Clarke had hoped to set off warning bells, he certainly pushed the right button.
Strong at once launched a personal investigation, questioning OSS managers and requiring the service to produce all records concerning the Lisbon operation immediately. Strong almost immediately muddied the waters—probably more from confusion than malice, however. The scraps of paper fished from the wastebasket by the OSS agent pertained to a cipher used by the Japanese naval attaché. But the general got it into his head that the OSS had stolen an entirely different cipher, used by Japanese army attachés in Axis and neutral capitals to keep Tokyo informed of their espionage activities. American code breakers had cracked this important and difficult cipher, a coup that Strong feared was now in jeopardy. Outraged, he fired off a series of memos to General Marshall condemning the “ill-advised and amateurish” activities of the OSS, warning that such shenanigans “have so alarmed the Japanese that it is an even money bet that the codes employed by the Japanese are in imminent danger of being changed,” and concluding that the OSS was no less than “a menace to the security of the nation.”
To support his accusations, Strong told Marshall that he had consulted George F. Kennan, the counselor of the American embassy in Lisbon, then in Washington on home leave; Kennan had confirmed to Strong that OSS officers in Portugal were rank amateurs whose efforts to penetrate the Japanese embassy were at best puerile and at worst disastrous, since their sources inside the embassy were almost certainly double agents who kept the Japanese informed of OSS activities. Strong demanded the recall of all OSS officers involved in the operation, an investigation by the Joint Chiefs of Staff of all OSS activities, and an explicit prohibition on OSS operations that might compromise sources handled by other intelligence elements.
Ironically, it was Kennan himself whose carelessness almost certainly was responsible for the Italians becoming aware of OSS penetration of the embassy. Kennan—who would later become famous as the author of a long analysis of Moscow’s intentions that formed the basis of the American cold war policy of containment of Soviet ambitions—was actually playing a double game. While still in Lisbon, Kennan had actually approved the operation in advance, been shown the documents that had been obtained, and had congratulated OSS officials there on their success. Somehow Kennan forgot to mention any of these facts to Strong during his investigation. Moreover, it is unlikely that Kennan, as a career diplomat, would have failed to inform his superiors by cable of at least the general outlines of the OSS operation. The trouble with that was that Italian military intelligence had broken the cipher used by the State Department to communicate with its embassies. The Italians almost surely intercepted and decrypted Kennan’s message to Washington revealing that the Japanese embassy had been penetrated—and cipher materials stolen.
also turned out that the OSS had never asked its mole in the embassy to target cryptographic material; the man had just spotted the papers in the trash and acted on his own. Moreover, the recall by the Japanese foreign ministry of cipher J-19 was not even a reaction to the warning from Italian intelligence, but part of a scheduled upgrade of diplomatic ciphers that the ministry had begun in late 1942. The new army attaché cipher had been introduced in February 1943, at least four weeks before the theft from the Japanese embassy, and a full four months before Tokyo heard from Italian intelligence about the possible penetration of its ciphers. Finally, by late summer 1943, the U.S. Army code breakers had intercepted messages passing between Tokyo and its diplomatic missions in Portugal and Spain indicating that after their initial excitement, the Japanese had convinced themselves that there had been no compromise of security. The ambassador in Lisbon informed Tokyo that embassy security measures, including wax seals placed nightly on the doors and windows of the embassy’s code room, were impenetrable.“This is a planted report to throw us off balance,” the foreign ministry in Tokyo concluded, and determined that there was no need to distribute new ciphers.
Yet all of this exculpatory evidence—which surely was known to Colonel Clarke and General Strong—was not circulated with the alacrity of the original charges: the Lisbon affair always had much less to do with protecting access to Japanese communications than with bureaucratic backstabbing.
But Strong may have had another reason at that particular moment for having such a short fuse over any implication that America could not properly handle cryptographic secrets. In the spring of 1943 Strong and his communications intelligence managers, including Colonel Clarke, were engaged in delicate negotiations with their British counterparts over access to German messages encrypted by the Enigma cipher machine. While prepared to share the intelligence results of their work against Enigma, the British were reluctant to allow Americans to participate in the technical process of cracking the German messages. Strong struggled throughout the spring of 1943 to convince the British to give his service a seat at the Enigma table. The fact that the U.S. Navy already sat at that exclusive table especially galled the combative general who, even in his best humor, was inclined to see a personal affront in every gesture or development. After months of difficult negotiations that strained the Anglo-American intelligence alliance, Strong overcame British reservations, particularly about British concerns that Americans were lax when it came to security, and in May 1943 signed an agreement providing for U.S. Army participation in Enigma operations. With the ink hardly dry on the document, the last thing that the general needed was an incident that suggested that top-secret cryptanalytic programs could be compromised by the actions of American agents in foreign lands.
In the short run the OSS survived Strong’s wrath, although to mollify the general Donovan issued an order prohibiting OSS personnel from seeking cryptographic material. The Lisbon affair, however, permanently sullied the agency’s reputation and provided ammunition for critics who would, after the war, convince President Truman to dismiss Donovan and abolish the OSS. The erstwhile spymaster learned that intelligence wars are fought on domestic as well as foreign fronts—and that dangerous opponents were to be found in the corridors of Washington as well as the streets and alleys of foreign capitals.
Originally published in the February 2008 issue of World War II Magazine. To subscribe, click here.