Anglo-American collaboration in special intelligence came somewhat later than that in other forms of SIGINT. All collaboration in intelligence was based, one must remember, on mutual advantage rather than on altruism. As it progressed, confidence grew, but neither country released anything to the other unless the action would benefit itself either directly or indirectly by the uses to which the other would put the assistance. When the likelihood of war applied primarily to the Japanese Empire, U.S.-British exchanges of information on Japan in the cryptologic field went further than in other areas. Benefits had always to be balanced against risks, and while the U.S. remained a non-belligerent, Britain seemingly viewed U.S. security secondary to the protection required for special intelligence.

The United States moved toward belligerency by stages. Before the surrender by France in 1940, the U.S. was a neutral hoping to escape the conflict. Gradually it faced the prospect that not only France but Great Britain would succumb to Nazi German dominance unless the U.S. became an "arsenal of democracy" and furnished Churchill the tools which he needed to "finish the war." Delivery of the material made in America led to protection of convoys by units of the U.S. Navy along with the Royal Canadian Navy as far as a "chop" line in mid-Atlantic. The formula for American action became "all aid short of war." By 1941 the U.S. government had given the Nazi fuehrer occasion for declaring war against the U.S. had he needed an excuse. He would do so, though, only if more advantageous to him than refraining. Preparedness for war in the United States lagged far behind the requirements of readiness to meet such an action.

Steps in that direction included staff talks in the Washington area at which, in the supposition that the U.S. had gone to war, a grand strategy and certain major preparatory measures were agreed.

For the British, World War II had its phases too. While deprived of support by Belgium and France in 1940, and while acquiring Fascist Italy as a declared adversary, the British "stood alone," awaiting a probable invasion by the Germans. By surviving German bombing attacks, the British discouraged the invaders and diverted them to a vast attack on the Soviet Union in June 1941. As the Germans became enmeshed in a winter offensive there, the Japanese seized an opportunity (that may have seemed to be fleeting) to benefit from the involvement of the British, French, and Dutch as well as the United States in the European conflict. Pearl Harbor was quickly followed by Japanese triumphs in a selected part of the British and Dutch empires and by Japanese threats to isolate Australia and New Zealand. When Hitler and Mussolini declared war against the United States – while Japan withheld parallel action against the Soviet Union – the alignments of World War II were established.

During the stages of change in the war situation of the future Anglo-American Allies, certain factors that affected collaboration in special intelligence appeared. The producers of SIGINT in both governments were contributors to general intelligence used in connection with foreign policy, ground, air, and sea operations of the armed services, and various forms of warfare included within the concept of "total war" that then prevailed. The army and air forces of the United States had one SIGINT organization, although the air components tended to become independent as the war continued. The U.S. Navy Department had a second SIGINT organization. Among the British, the Foreign Office, Admiralty, War Office, and Air Ministry kept their principal SIGINT producers under one umbrella, the GCCS; the latter distinguished sharply between special intelligence and "Y" intelligence. Different authorities controlled the two.

In the United States the same authorities, divided between army and navy organizations, controlled production of SIGINT of all grades. The Army's Signal Intelligence Service (later Signal Security Agency) tried to meet the requirements of the Military Intelligence Service. In the Navy, OP-20-G became part of an extensive reorganization in the months after the Pearl Harbor attack. OP-20-G tried to enhance its services to those engaged in the Battle of the Atlantic, while devoting most of its growing resources to Japanese matters.

The initiative to effect Anglo-American collaboration in SIGINT was taken by the British government through its ambassador in Washington, Lord Lothian, who in July 1940 suggested to President Franklin Roosevelt certain exchanges of broad and general scope.^[107] Churchill, who had been prime minister for about two months of a country left alone to face grave danger, offered to exchange certain British scientific and technological accomplishments that had military application in return for access to American industrial productivity, particularly in the field of ultra-shortwave radio emitters. While not requiring any other quid pro quo, the proposal expressed a hope that the U.S. government would exchange scientific and technical information as well as allowing British procurement of military equipment.

On 11 July 1940 the cabinet adopted a position expressed by Secretary of War Henry L. Stimson (in which Secretary of the Navy Frank Knox concurred), that the U.S. government

Secretary Stimson advocated that any British procurement arising from the agreement to exchange information be subject to approval by the U.S. Army or Navy Departments, in order to insure that it would not interfere with American programs of procurement.

The exchanges began, after the arrival of a British technical mission headed by Sir Henry Tizard, at a session on 28 August 1940. The representatives of the U. S. Army had been authorized to furnish not only "technical information on munitions, devices, or processes of manufacture owned by the U.S. Government," but also "cryptanalytic information," not, however, to include any information about our own codes, ciphers, and methods of cryptography. Intelligence gained abroad could be given to the British, but U.S. patent rights and trade secrets were to be protected.^[109]

The president sought to obtain American military assessment of the British ability to withstand the Axis powers in addition to the counsel being provided him by Colonel William D. Donovan. In August 1940 Brigadier General George V. Strong, chief, War Plans Division; Brigadier General Delos Emmons, USAAF; and Rear Admiral Robert Ghormley, assistant chief of Naval Operations, went to England (at about the same time that the Tizard mission was in Washington) to observe and to discuss ways of assisting each other. Their soundings continued while, at home, Army and Navy SIGINT officials considered the possibility of exchanging cryptologic technical information and SIGINT products with the British government.

From London, General Strong, by cable of 5 September 1940, asked the chief of staff if the Army would agree to a full exchange with the British of German, Italian and Japanese code and cryptographic information. The U.S. Navy (Captain L. F. Safford, Chief, OP-20-G) was unwilling to exchange more than intercepts; the U.S. Army, on the other hand, responded favorably to General Strong's query, though not to a continuous exchange of intercepts. By January 1941 contacts between GCCS and both G-2, SIS and OP-20-G were in progress, under the canopy of the Tizard mission and by an authorization to the National Defense Research Committee to deal with crypto-logic exchanges.

Captain Laurance F. Safford, USN (Photograph from the NSA History Collection)

Actual exchanges of information with respect to cryptanalysis were somewhat gingerly approached. The British were then beginning to produce SIGINT from high-level communications of the German Air Force and Army, but their successes were limited, hard won, and often required amplification from other sources. German submarine communications enciphered by an Enigma machine presented analytic problems of immense difficulty, soon to be increased, on which the effort was intense. GCCS was working also on Abwehr (German Secret Intelligence) traffic and on intermediate cryptosystems of the German, Italian, Japanese, and other governments, and achieving important successes.

In the United States, while the navy worked on a Japanese naval attaché system, the Army SIS in the summer of 1940 was nearing the end of an arduous eighteen-month quest for an electromechanical analog device able to decipher the highest level cryptosystem used in Japanese diplomatic communications. Having in 1939 displaced a system previously known in Washington as RED, the new one was labeled PURPLE. Early in October 1940 came the first deciphering successes with "the PURPLE machine." A wiring diagram, technical data, and a considerable body of intercepted traffic were thereupon delivered by agreement to the navy cryptanalysts for use on work in progress on the same target. Secure navy engineering facilities then constructed six PURPLE machines, of which three went to the army, one to GCCS, and two to the navy.

As part of the exchanges in the cryptologic field, the British agreed to receive at GCCS two U.S. Army and two U.S. Navy cryptologists who would escort the PURPLE machine that the U. S. was giving to the British. They crossed the Atlantic from Halifax on a British warship. Representatives from each SIGINT organization, each a reserve officer, were designated. The Navy Department sent Lieutenant Robert Weeks and Ensign Prescott Currier; the Army had planned to send its leading expert, Lieutenant Colonel William F. Friedman, but illness caused his orders to be revoked, and Captain Abraham Sinkov thus became the senior Army officer, accompanied by Lieutenant Leo Rosen. At GCCS, where special machines to rapidly test analysts' ideas were in use, the American gift could be appreciated. Washington officials expected these officers to discover and learn all that they could to benefit the work to be done at home.

Lieutenant Leo Rosen (Photograph from the NSA History Collection)

Captain Abraham Sinkov Photograph from the NSA History Collection)

Captain Sinkov's report described how GCCS was organized and how its output was facilitated by interception, direction finding (DF), and technical radio intelligence (called by the British, Wireless Technical Intelligence or WTI). Both Army and Navy officers had observed much that was outside the realm of special intelligence; they knew only of the existence of special intelligence and were pledged not to reveal its existence except to specifically named individuals.^[110]

The ability of the analysts at GCCS to read German communications enciphered in some of the cryptosystems that used the Enigma machine was sufficient in 1940 to promise that the limited defenses of the United Kingdom would be used with maximum effectiveness against any invasion across the Channel. Broader successes in the coming months could be expected at GCCS. The products, special intelligence, were kept utterly secret by elaborate security precautions. Awareness by the Germans could deprive the British of what amounted to a "secret weapon" essential to their survival.

When OP-20-G started trying to read German naval Enigma traffic early in 1941, considerable progress would have been facilitated by sharing the lessons of experience already learned at Bletchley Park. Technical SIGINT collaboration then was confined to "Y" intelligence. Intelligence provided by the Admiralty to the Navy Department may have included some SI in a disguised form. As late as August 1941, when Commander A. G. Denniston, then head of GCCS, conferred in Washington with officers of the U.S. Navy and U.S. Army SIGINT organizations,^[111] he could not fail to observe how far from a consolidated, unrestricted commitment of American resources to preparedness still prevailed. The division of American public opinion over the right course of the United States concerning the war in Europe remained deep and emotional. Prominent were not only a "Committee to Defend America by Aiding the Allies" but also an "America First" committee.

Probably at the time of Commander Denniston's visit if not earlier, OP-20-G received from GCCS diagrams of wirings and wheels of an Enigma machine and descriptions of the wheel movements during encipherments. Such material was subjected to preliminary research by Mrs. Agnes P. Driscoll and others, but was allowed to drop out of sight.^[112] Methods of solution that had worked on lower-grade systems could not be applied successfully. They were much too slow. Machine processing was required to cope effectively with such complex machine encipherings, a fact realized slowly at OP-20-G.

Although the U.S. Navy had been participating with the Royal Canadian Navy and the Royal Navy in protecting transatlantic convoys long before the attack on Pearl Harbor, German submarines in January 1942 commenced a campaign of sinkings along the eastern seaboard which the U.S. Navy was not able to oppose effectively. When the means of retaliation had been sufficiently increased through better information about the enemy's situation and centralized control of our shipping and escorts, the Germans found the campaign too costly and returned to other areas of the Atlantic.

Left to right: Alistair G. Denniston, Director, GCCS; Professor E. R. Vincent; and Brigadier John H. Tiltman in London, 1942 (Photograph from the NSA History Collection)

Commander Rodger Winn, a sagacious British barrister turned Royal Navy Reserve officer, visited the United States in the summer of 1942 and succeeded in persuading navy authorities to institute a system of escorted convoys from the Caribbean to northern ports and to link that system with the convoys to European destinations. He also convinced navy authorities that something akin to the Operational Intelligence Centre (OIC) in the Admiralty, would strengthen American anti-submarine operations. The British OIC was soon visited by Lieutenant Commander Kenneth Knowles, who was to become the officer-in-charge of parallel activities in Main Navy, Washington. His unit, under a successively modified administrative organization, was the Atlantic Section, Combat Intelligence Division, U.S. Fleet (F-21) and eventually, a unit of the Tenth Fleet (FX).^[113] Its submarine tracking room used intelligence from all sources, and could advise the convoy and routing unit of the submarine situation affecting a convoy's course.

Collaboration in the production of special intelligence tends to be focused on the invention and manufacture of high-speed machines for use by army, navy, and GCCS analysts. Some of the books published about how Ultra intelligence was used to win the war leave the impression that, on both sides of the Atlantic, men of genius conceived of mechanisms that enabled decryption to be performed by tenders of machines analogous, perhaps, to the shift in textile manufacture from hand-weaving to factory output. Instead of being watchers of bobbins, the analysts acted more like musicians whose instruments provided music when properly played, but which otherwise yielded only noise.

Several types of a machine usually called a "bombe" emerged during the war from the work of men who applied electromechanical technology that was then novel. The British in 1939 acquired from Polish and French collaborating cryptologists the prototype of a bombe,^[114] and went on to devise their own. The Enigma in use from 1939 to 1941 had three wheels, or rotors, and could be additionally elaborated so that the paths of a maze of wiring between the same plaintext letter and its final cipher equivalent could differ several thousand times. The bombe enabled the analysts to discover the way in which the controls of the sending and receiving Enigmas had been set for an enciphered message to be converted to plaintext German. It was possible for an enciphered message to be a reencipherment necessitating a second deciphering with special keys available to limited recipients.

The three-wheeled Enigma was partially replaced on 1 February 1942 by a fourth-wheel version for some German naval communications. When the four wheels were used, GCCS lost the ability to decipher most Atlantic submarine traffic for at least nine months in that year. The intelligence shared by the British Admiralty with the Navy Department with respect to protecting convoys and weakening the German submarine offensives abruptly declined in quality. Certain navy authorities, unaware of the reasons, concluded that the British were not meeting the obligations that they had assumed for full reciprocity in exchanges of cryptologic techniques. If the American cryptologists started as junior partners in producing SIGINT (other than Japanese), they intended to become equally competent by catching up. The material given to 0P-20-G, including samples of plaintext messages with data about the Enigma, had been so quietly provided and so secretly retained there, that British action was probably unknown to the complainers.

The U.S. Army SIGINT organization in the throes of rapid growth, particularly in 1942, intended to acquire competence to produce SIGINT from all kinds of encrypted communications. It could not acquire German military traffic of sufficient quality and quantity for efficient processing from its own intercept stations, and it expected reciprocity – from GCCS for American contributions since 1940 concerning Japanese SIGINT – to include all kinds of material concerning German communications. Those expectations were for a time disappointed. The Military Intelligence Division, War Department, headed in 1942 by Major General George V. Strong, concluded that it was not being given all the SIGINT to which it was entitled and that it could not allow a condition of dependence on GCCS to persist indefinitely. No U. S. Army forces were even scheduled for operations against the Germans until the decision in July 1942 to seize French North Africa as a base for future operations in the Mediterranean region. The first landings there could not be made until the following November. During Allied discussions in 1942 of a cross-Channel attack to establish a second front in Western Europe to aid the Soviet struggle on the Eastern Front, and during the planning for Operation TORCH in Northwest Africa, the necessary intelligence was for the most part of British origin. During the Casablanca Conference in January 1943, the same situation persisted.

Bombe deck at OP-20-G, May 1945 (Photograph from the NSA History Collection)

In 1942 the U.S. Navy Department's pressure on the British authorities led, finally, to an accommodation. Colonel John Tiltman, BA, as a temporary representative of GCCS in Washington, convinced his superiors in the U.K. that the U.S. Navy's "Y" authorities were determined to develop the ability to produce special intelligence to offset the British failure to furnish it. He was authorized to explain that the Royal Navy itself was being denied the kind of German naval intelligence that had once been available and that GCCS had not been able to produce it since the advent of the four-wheeled Enigma for Atlantic submarine communications. The closest possible collaboration by the British in producing and using "Y" intelligence could be expected, but recovery of special intelligence of first-rate quality, as before, seemed likely to depend upon a successful development of a high-speed bombe. GCCS had obtained authorization to exchange information about such a development, and would supply one bombe to OP-20-G.

By arrangement, two U.S. Navy officers, Lieutenant R. B. Ely and Lieutenant Junior Grade J. J. Eachus, went in June to GCCS in part to observe how the British made use of an expert analytical research group and, in part, to learn all that they could about the processing of naval Enigma traffic. Lieutenant Commander Howard T. Engstrom and others concerned with rapid analytical machines (RAMs) prepared the two visitors for their trip. Ely returned in August, Eachus in October, after having stayed to see tests of the British devices then under development. Those tests and others that followed were discouraging.

OP-20-G began a project, after Navy Department approval on 10 September 1942, to develop and then manufacture a device designed to operate at high speeds and to discover Enigma settings by testing possible cribs. That was to be the navy's bombe for the four-wheeled Enigma. As outlined by Commander Joseph Wenger a week earlier, the project would cost $2,000,000 (already available at the Bureau of Ships) and take about five months from the first model to reach a production rate of about one per day. Operating them would eventually require 30,000 square feet of space and 500 operators. (The exact total number to be fabricated was left for later determination. It was once set at 96, then increased to 112, plus more to be shipped to the UK for GCCS.) The Navy's contract with the National Cash Register Corporation of Dayton, Ohio, was surrounded by extraordinary security controls and given the highest priority and precedence in obtaining essential materials, after an appeal to President Roosevelt by Admiral Ernest King. Lieutenant Commander Ralph Meader, USN, was placed in charge of what was designated the Naval Computing Machine Laboratory at the Dayton plant. Joseph R. Desch of NCR Corporation was the contractor's principal engineer involved. Waves were sent to Dayton to assist in assembling and testing the first models. While the work there advanced, a special building for bombe operations was prepared at the Naval Communications Annex in Washington.

Commander Joseph Wenger (Photograph from the NSA History Collection)

By June 1943 production models were available for tests which they passed well. Output reached four or more per week, and analytic operations were maintained in Dayton until September. Commander B. F. Roeder, USN, was put in charge of a new special unit of OP-20-G to process Enigma traffic and to disseminate results. Quarters for the Waves who would run the bombes were constructed across the street from the compound of the Naval Communications Annex in Washington. In September the bombes began working there as the transfer from Dayton became gradually complete. When the daily keys in German naval traffic had been ascertained, the bombes became available for other tasks, some of which GCCS requested.

The terms of collaboration in special intelligence by OP-20-G and GCCS were recorded in a written agreement negotiated in Washington for GCCS by Sir Edward Travis, its head, and reported to his superior by Commander J. N. Wenger, USN, on 1 October 1942. It was signed the next day. The British consented to full exchange of intercepts, keys, "menus," cribs, and all technical data applicable to German Navy and, in particular, German U-boat communications. The United States undertook to match the British primary responsibility for German naval SIGINT by assuming parallel responsibilities for the Japanese naval SIGINT problem. OP-20-G would send to GCCS the Japanese material intercepted at American stations and all American recoveries of cipher keys and codes to help GCCS retain its capabilities in Japanese SIGINT production.

In developing and using analytical equipment, GCCS would respond to requests for technical assistance and would send to OP-20-G certain technical personnel to obtain information about the American high-speed project. GCCS would also obtain certain items of special equipment that were manufactured in the United States.

By the collaboration thus defined, the intelligence needs of both the Navy Department and the Admiralty would be served. The arrangements might be expected to benefit the Allied protection of transatlantic convoys over the new sea lanes for supply and reinforcement of the Allied campaigns in the Mediteranean area.

While the Navy bombe project was in the developing stage, the Army too became restive at what seemed to be a withholding of reciprocal assistance by the British at GCCS. In May 1942 Major Solomon Kullback began a visit of several months at GCCS. He studied the cryptanalytic methods used at GCCS for treating a wide range of enemy cryptographic systems. Though he was shown operations in progress on Enigma traffic, it was not done in a way that enabled him to become an expert at that as he did with other cryptosystems.

Probably before the navy actually launched its bombe project, the army SIS had decided to design a bombe that used relay switching instead of the rotary type to be employed in British and U.S. Navy versions. On 30 September 1942 the Army Signal Corps contracted with Bell Telephone Laboratories (BTL) to produce a test sample. Its subunits would be a single-frame apparatus analogous in its operations to those of a German three-wheeled Enigma. By November 1942 BTL could demonstrate successfully the sample model. BTL then undertook to complete 144 frames. To the uninitiated this bombe looked like metal shelving stocked with the insides of radio receiver sets in an orderly and interconnected arrangement.

In the winter of 1942–1943, GCCS had as a bombe building project one that involved ultimately choosing one of two methods for coping with encipherments from a four-wheeled Enigma. The U.S. Navy had another. The U.S. Army was well along on developing its bombe applicable to a three-wheeled Enigma. Officials of SIS and OP-20-G conferred in January 1943 on the progress made thus far with the Army's apparatus, and the possibility of incorporating certain features in its further development. At that time, seventy-two frames (or half the total number) were expected to be ready early in April 1943.

Pursuant to the agreement with the Navy in October 1942, GCCS sent one of its leading cryptologists, Dr. Alan M. Turing, to the United States in December 1942. He had been closely concerned with the devices for machine processing that had been developed by GCCS. On 21 December he was escorted to the Naval Computing Machine Laboratory by several officials of OP-20-G and by Major Geoffrey Stevens, the British SIGINT Liaison Officer stationed by agreement at Arlington Hall Station.. The navy bombe as it then stood was of uncertain merit as he saw it, though he may have reconsidered and thought better of it later on.

When Dr. Turing's authorization to visit the Bell Telephone Laboratories and observe Army cryptologic projects there was sought from a junior American officer, he was suspected of trying to slip in somewhat clandestinely. His request was refused until the issue of his eligibility had worked its way through official suspicion at various levels to become the subject of correspondence between General Marshall and Field Marshal Sir John Dill, the head of the British Joint Staff Mission in Washington. On 5 February 1943 he and Major Stevens were finally received at BTL and were shown the army bombe and other devices then under development.^[115]

Dr. Alan M. Turing, British cryptologist (Photograph from the NSA History Collection)

During the Anglo-American discussion preceding Dr. Turing's visit to BTL, some U.S. Army SIGINT officials expressed doubt that the British had been exchanging cryptologic materials with complete reciprocity, as agreed. When the army identified materials allegedly withheld, the British insisted that they had all indeed been released to the American cryptanalysts who had come to GCCS. The British maintained that, to fulfill their part of a "full and free exchange," they did not have to allow such materials to be used in Washington. The correspondence with Sir John Dill induced General Marshall's deputy to acknowledge that sending "us" the Enigma materials wanted by the U.S. Army did indeed involve an increased hazard of compromise that need not be incurred. Instead, he conceded that arrangements should remain as they then were.^[116]

The British concluded that at the Bell Telephone Laboratories the U.S. Army was experimenting with what they believed to be a more efficient development of the same "highspeed analyzer" that had been shown at GCCS to several American officers. The prime minister had authorized this disclosure during their visit to GCCS in April-May 1941, and at later dates to Brigadier General Frank Stoner and Lieutenant Colonel George Bicher of Headquarters, ETOUSA. Major Solomon Kullback and Captain Roy D. Johnson of the SIS had subsequently worked at GCCS on the Enigma cryptanalytic problem, and had learned of the more recent developments. In the eyes of the British, the experiments at BTL should have been disclosed to them without waiting for ultimate success.

The Turing episode is important as an indicator of the U.S. Army's degree of self-confidence and the British appraisal of a proper partnership. Dr. Turing was a young mathematical genius whose doctorate had been won at Princeton University in 1938, after two years of leave from his Fellowship at King's College, Cambridge University. He had then declined the opportunity to work with the famous mathematician, John von Neumann, at Princeton as an assistant in order to resume the role of a Cambridge don. He had already produced, before the war, a paper with ideas concerning an electronic machine which could calculate automatically – a paper read widely among mathematicians because of its demonstration that some problems were not mathematically solvable even though mathematically stated.

GCCS had recruited him very soon after the war began, and at Bletchley Park he was known as "the Prof." because of his abstracted ways; he had shown various practical qualities too. They ranged from padlocking and chaining his tea mug to a radiator in his office to devising electromechanical means for sifting the cryptographic variables in enemy encipherments.

The records do not show whether the method by which his access to the Bell Telephone Laboratories as first sought was really intended to avoid drawing attention from those who ought to have been approached for approval. It was not the only time that Turing relied for necessary credentials on administrative channels which ignored requirements or took too much for granted. While the obstacles to his visit were being removed, he seemed to have found time to visit Princeton and to have enjoyed access to food more nourishing than Britain's wartime rationing permitted at home.

The main consequence of the episode was undoubtedly the demonstration that, when taken to the highest army level, the specifics of collaboration in cryptologic matters would follow the view of GCCS and not of G-2. Moreover, in spite of the navy's intention to protect its highest COMSEC system, the decision was made by the chief of staff of the Army to allow Turing to see such a cipher machine under development. In the discussion, General Marshall set a precedent for accepting the British view that all unnecessary risks of a compromise by transmitting SIGINT materials out of the United Kingdom should be avoided.

When, a few weeks later, the U.S. Army tried to obtain Ultra materials from the British comparable to what had already been released to the U.S. Navy's OP-20-G, the British declined. GCCS, supported by the Chiefs of Staff Committee, then insisted that its SIGINT collaboration with the U.S. Navy, including some SI, had been necessary in order to fight the German Navy in the Atlantic.

The availability of enough of the new army's rapid analytical machines (RAMs) for SIS work on German Army and Air Force Enigma communications could be foreseen in February 1943. Mr. Friedman on 8 February then advised Colonel W. P. Corderman, commanding officer at Arlington Hall, that the SIS could by 1 April 1943 begin assembling and installing "our E-solving machinery" and could expect to be operating soon afterward. Needed then would be trained operators, traffic, and traffic-intelligence materials, and certain special information and complex special procedures already developed at GCCS. Special cryptographic equipment for use in fully secure communications channels would also be required. While some German Army Enigma traffic was being intercepted in 1942–43 at Vint Hill Farms Station, or at an Army intercept station in Newfoundland and another in Iceland, the amount was small and the reliability of interception was inadequate for regular work. British intercepts would be essential to the program at Arlington Hall Station.

The British position in January 1943 at the time of the "Turing affair" had been that there should be no "exploitation" in the U.S. of vitally Top Secret traffic unless the British became convinced, as in the case of German submarine communications, that such exploitation in the United States was necessary. Colonel Tiltman persevered in upholding that position in conversations with Mr. Friedman. The latter insisted that the U.S. Army apparatus bore no external or internal resemblance to the British bombes or associated equipment and asserted that it was capable of solving several other types of cryptographic traffic problems. He proposed to expand U.S. interception of Enigma communications in Northwest Africa and to establish new Washington-Algiers telecommunications cryptochannels.^[117]

William F. Friedman at his desk at Arlington Hall Station (Photograph from the NSA History Collection)

The U.S. Army proposal to the British on 23 February 1943 meant that a second center for producing some German Army and Air Force Ultra would be established in the Washington area. Enigma traffic that had been intercepted by American and British operators, and then analyzed and interpreted, would require transatlantic communications for delivery of special intelligence from the U.S. to destinations in the theaters of war.

At that time, the Allies controlled no part of the European continent, and neither partner was ready to risk an Allied attempt to invade across the English Channel for lack of sufficient strength to remain. The British could still consider the United States to be a junior partner whose potentialities remained uncertain in the light of current events in Tunisia. They quite naturally recoiled from the prospect of relinquishing control over an instrument of war that they had produced and on which they depended so heavily. U.S. Army authorities knew that special intelligence was available to Americans at Allied Force Headquarters at Algiers and at ETOUSA in England, and would continue to be available to them on the principle of the "need-to-know." It was the ability to decrypt Enigma traffic independently that, in the long run, they believed they must acquire. They appreciated that the security of Ultra was fragile but believed that the risk in producing it in the U.S. was not too great.

Earlier agreements to exchange traffic with the British (and, as of 15 January 1943, with the Canadians) were bringing material of solid benefit to the U.S. Army. The traffic was chiefly diplomatic and Japanese Army communications. At British sites, communications originating in neutral and German-dominated countries in Europe could be copied when no American station could hear them. In Canada, Australia, and British India, Japanese Army material copied by the British and furnished in the exchanges amounted to about thirty percent of that available to American analysts. The amount of duplication did not exceed what was warranted in order to eliminate garbles and to verify doubtful information.^[118]

The attitudes of individuals at lower levels of the chain of command were more nationalistic than that of those near the top. When Captain Edward G. Hastings, RN, discussed the American request with Colonel Carter Clarke and Colonel Corderman, for example, his remarks could be treated as more or less an ultimatum – the Americans should withdraw their proposal or Anglo-American collaboration in SIGINT might cease. Cooler heads, in particular Lieutenant Colonel Telford Taylor of G-2, recognized that collaboration would not be allowed to founder if the issue were to be taken high enough. He also believed that the current British position would eventually soften so that American capabilities could later be expanded gradually under arrangements that would then be agreeable.

Consultations in London, in which the British Chiefs of Staff Committee was asked to consider a British reply to the U.S. Army proposal, and in which the U.S. military attaché and the principal U.S. SIGINT officer in London, Colonel George A. Bicher, participated, produced an "unshakable answer" from the British. American participation in the United Kingdom would be welcome; no exploitation of Enigma traffic performed in the United States would be accepted. The British were able to find support for their position in the words of General Marshall's letter at the time of the "Turing affair."

During the negotiations, the U.S. SIGINT officer at Allied Force Headquarters in Algiers, Lieutenant Colonel Harold G. Hayes, was asked to report whether SI went from GCCS to AFHQ in Algiers and, if so, whether it came to him. He replied that the "Y" Service in which he was engaged was quite separate, that SI went directly to General Eisenhower's chief of intelligence at AFHQ via a special radio link and that Hayes himself could not prudently even inquire about SI until he knew the substance of any agreement about it between GCCS and the SIS, Washington.

Brigadier General Harold G. Hayes, 1955 (Photograph from the NSA History Collection)

The Americans wished to avoid an ultimatum from either ally to the other, but they insisted that the U.S. Army must become capable of producing SIGINT from German Army and German Air Force Enigma communications. Again the issue was resolved by negotiations in Washington in which Sir Edward Travis, head of GCCS, came to an understanding with Colonel W. P. Corderman, Sig C., in terms that were supplemented by arrangements satisfactory to the Special Branch, G-2, War Department. The agreement was signed by Travis and Corderman on 17 May 1943. It conformed closely to a draft approved by the British Chiefs of Staff Committee and offered to General Marshall by Sir John Dill. General Marshall thus overruled General Strong's insistence on terms enabling the U.S. Army, forthwith, to start developing at home (with the assistance of GCCS) its own ability to produce SI. He acceded instead to the British agreement to receive a U.S. Army unit in the UK for training and production there that would be supervised by GCCS and for delivery of SI to the War Department by secure Britsh communications channels.

The new agreement distinguished between the American and the British names for the same things as follows:

Brigadier General W. Preston Corderman, 1954 (Photograph from the NSA History Collection)

The two organizations, GCCS and SIS, agreed to exchange completely all information concerning the detection, identification, and interception of signals from, and the solution of codes and ciphers used by, the military and air forces of the Axis powers, including the German Abwehr. The U.S. Army assumed as its main responsibility the reading of Japanese Army and Air Force codes and ciphers.

The British assumed as their main responsibility the reading of the codes and ciphers of German and Italian Army and Air Forces. Both countries agreed that special security regulations should apply to intelligence derived from enemy high-grade codes and ciphers. In that connection, both agreed to use their most secure cryptographic systems for the transmission of decodes and technical cryptanalytic data.

British or U.S. military or air commanders in chief would receive all the special intelligence necessary for the conduct of operations – obtaining it from either British or U.S. centers as mutually agreed. To insure dissemination of that sort, liaison officers were to be appointed and authorized to see all decodes. Distribution of SI would be held to a minimum and would be confined to those who needed the intelligence for the proper discharge of their duties. The same security regulations would govern all recipients of Special Intelligence A. Until modified by mutual agreement, the regulations in force in theaters of war where British forces were already operating would govern. Special Intelligence A was not to be intermingled with general intelligence from other sources unless that became imperative, and then all must be handled as if it were special intelligence A. Under no circumstances could special intelligence be transmitted in a cryptographic system that could be read by anyone except an authorized recipient. Special intelligence B was to become the basis for action, documents, or telegrams only when a different source could be presumed, and transmissions were to be in "absolutely secure" cryptosystems.

In Washington and London intelligence liaison officers of the two allies were to have access to all special intelligence and were to be free to select and forward whatever they deemed necessary.

Cooperation and coordination in "Y" between the U.S. Signal Intelligence Service and the British "Y" Service, at all levels, were also to prevail. Each was obligated to inform the other of the employment and scope of its "Y" effort in Allied theaters.

It was agreed that research into new technical methods of attack on German Enigma communications would continue to be conducted in Washington.

U.S. liaison officers at GCCS would select messages and summaries that they believed should be sent over existing British secure channels, either to G-2, Washington or to theater commanders.^[119] U.S. liaison officers in the War Office and Air Ministry would continue, as in the past, to handle SIGINT relevant to order of battle.

A U.S. party in Great Britain would work at independently solving Enigma keys but would avoid duplication of effort. They would be furnished with British machines and be given instruction in all processing and would conduct complete processing to the extent they desired. Their decodes would be passed to Bletchley Park for emendation, translation, and distribution. Members of that American party would not be transferred elsewhere except for urgent reasons. Distribution of Allied special intelligence would be through specially trained British units at the headquarters of commanders-in-chief. If the commander-in-chief was an American, an American liaison officer would be attached to the unit "to overcome difficulties that may arise in regard to a difference in language." Ultimate control over matters of security and all dissemination of special intelligence was to be retained by the Director, GCCS.^[120]

The 17 May 1943 agreement was forwarded to the chief of staff, U.S. Army, by General Strong on 10 June 1943 with a recommendation that it be approved. On 15 June 1943 it was approved by the secretary of war.

On 23 July 1943 Colonel Bicher, the chief, SIS, ETOUSA, received from Washington the program for executing that part of the 17 May agreement that provided for a U.S. Army SIS contingent to learn how to produce SI in the UK.

After resolving matters with the Army, Commanders Travis and Wenger, along with other officers of OP-20-G, discussed the progress made on the two types of bombe for the four-wheeled Enigma that GCCS had under development and testing, and the situation in the United States, where the Naval Computing Machine Laboratory at Dayton, Ohio, already had a production model soon to be manufactured. On 18 May 1943 the British faced an uncertain performance by their new bombes in either version, while the Americans were confident that their machines worked well and would be in full production before the end of June.

In those circumstances it was agreed that the British would allocate tasks and provide materials to OP-20-G in the form previously shown, and that the Americans would keep GCCS informed of the number of operative bombes available and would, of course, return to the British all results as soon as possible.^[121]

Sir Edward Travis (Photograph courtesy of William F. Friedman Collection, George C. Marshall Library)

Even before the agreement had been concluded, Mr. William F. Friedman of SIS, now recovered in health but no longer in uniform, made a long visit to GCCS and to London. Lieutenant Colonel Alfred McCormack and Major Telford Taylor, from the Special Branch, G-2, accompanied him to England, arriving on 25 April 1943. In London they began by consulting Lieutenant Colonel George Bicher at Headquarters, ETOUSA, before he left for consultations in Washington. The dual nature of their positions in Army organization was reflected in courtesy calls on the assistant chief of staff, G-2, and the chief signal officer, ETOUSA. Their professional interest led them quickly to sessions with the deputy director, GCCS (Travis), and his superior officer (Sir Stewart Menzies), the chief, Secret Intelligence Service at the latter's office. There they learned that the British authorities expected them to proceed with their mission of observation without regard to the current controversy.

The three men often stayed together during their seven-week visit, but occasionally separated and compared notes afterward. Their movements took them not only to SIGINT stations in London and GCCS at Bletchley Park (BP) but to several outstations. Almost at once they went to Tidworth to see a U.S. Army SIGINT company in training. They sampled the SIGINT establishments at "Berkeley Street" and "BP," then devoted several weeks to thorough examinations of the systematic operations at each. It was soon apparent to them that U.S. intelligence officials in Washington had an inadequate understanding of the complexities of the British effort. Reports by earlier American visitors had not been studied widely enough. The British liaison officer may have been supposed to provide a full and intelligible picture to the responsible U.S. intelligence officials but had not. Whatever the reasons, Colonel McCormack set out to rectify deficiencies by industrious recording of observations on his own visits. All three tried to convey to their superiors (Colonel Carter Clarke of the Special Branch and Colonel W. Preston Corderman, CO, Arlington Hall) by carefully drafted messages, facts that they were convinced those officials at home should be taking into account in the negotiations over an GCCS agreement.

The main problems to be met in accordance with the terms of agreement would be

All Americans who were "in the Ultra picture" agreed that the U.S. Army must ultimately learn how to produce special intelligence for its own needs but in doing so ought not to hamper current production.

The three visitors were amazed by what the British had accomplished in the way of organization and expansion. Comparing their views of British production one evening, McCormack said: "It's not good. It's superb But it isn't military." The others agreed that at BP considerations of rank and grade were ignored in favor of "the best man for the job," of candor in technical cooperation, and of acceptance of personal eccentricities.

To Colonel Carter Clarke, Colonel McCormack cabled (in the spirit of BP):^[122]

The visitors discovered from the questions and instructions that came back to them while in London that their painstaking cabled reports were not convincing in Washington. Their superiors might not have been impressed by figures such as the many receiving sets within the United Kingdom alone by which Enigma intercept material was obtained, or the sixty-seven bombes already in operation of which forty-six were running twenty-four-hours per day on German Army and Air Force (three-wheeled Enigma) traffic. They might not have found noteworthy the estimate in May 1943 that 1,650 to 1,700 persons were at work on crypt-analysis and intelligence analysis of Enigma traffic, in addition to many more concerned with interception, special communications, and delivery. But if they ever understood the steps involved in processing Enigma, producing cribs, setting up machines for operations to verify the correspondence of letters, testing the arrangements, studying possibilities indicated by the machines, and deriving an understandable text from possible combinations, translating the text into intelligence, feeding the intelligence back into the general pot in order to relate it to logs and other data from traffic analysis in a search for more cribs, and other actions – then they might have been more agreeable with the conclusion of Colonel McCormack's message cited above. They might also have accepted his opinion that there was no likelihood that Arlington Hall soon could provide any significant amount of timely intelligence by working on the problems of Enigma keys that GCCS would define each day.

As McCormack noted, GCCS was both a producer of SIGINT information and an organization directly involved in supplying combat intelligence to field commanders. Although watch officers of each armed service selected items to go to their ministries (Admiralty, War Office, and Air Ministry) for treatment in the output of finished intelligence, GCCS itself sent any Ultra item relevant to a field commander's current operations to that commander by a direct communications system which GCCS controlled for that purpose. It sent appropriate items directly to Brigadier Sir Stewart Menzies, who quickly passed deserving items on to the prime minister.

Among the features of the British SIGINT operation that Colonel McCormack found of primary interest was the effort, within the security framework, to disseminate to that person all SIGINT affecting an individual's duty performance. Availability of SIGINT was not confined to the service with the greatest interest but extended to all three whenever their representatives in the Watch Room wished to have an item. Moreover, he could find no reason to believe that any items just went to British commanders when U.S./UK commanders were jointly involved. General Eisenhower, for example, was receiving anything pertinent to his area of command, and that information went also to General Sir Harold Alexander, his principal field commander.

Next in size to GCCS at Bletchley Park was CDR Denniston's station in London known as "Berkeley Street," where almost 250 persons were engaged in producing diplomatic intelligence. Everything released from Berkeley Street went to the Foreign Office and to the service ministries, while liaison officials from other government agencies selected what their chiefs might want. Items for Washington, New Delhi, Melbourne, and Cairo were distributed through GCCS; diplomatic intelligence based on SIGINT went to British diplomatic recipients through a system controlled by the Chief, SIS, rather than over standard diplomatic communications channels.

McCormack's visits included the MI6's establishment at St. Albans devoted to training men and managing counterintelligence operations. There he found a small party (largely recruited from American academic faculties) from the U.S. Office of Strategic Services. They were apparently at an early stage, of their preparations for the eventual invasion of the continent across the Channel.

Mr. Friedman, Colonel McCormack, and Major Taylor saw at Berkeley Street the PURPLE machine that had crossed the Atlantic in 1941 and another one – bigger and not so dependable – that British engineers had since fabricated. From the point of view of U.S. intelligence, Berkeley Street might in the long run prove to be a source of greater value than Bletchley Park's.

While Mr. Friedman was at Bletchley Park, news came by radio of the Allied captures of Bizerte and Tunis on 8 May 1943, and on 18 May came word of the Corderman-Travis agreement reached in Washington. On 20 May he was for the first time shown Bletchley Park's bombe operations.

The visit to the U.K. by Mr. Friedman and Colonel McCormack ended on 12 June 1943 but Major Taylor remained to execute certain terms of the 17 May agreement that were of great concern to the Special Branch, MIS. He was to be the first of the intelligence (as distinguished from SIGINT technical) liaison officers at Bletchley Park. He was also to be officer-in-charge of others to come from the U.S. for similar G-2 duties, or to be inducted into one of the Special Liaison Units (SLU) for distribution of SI to commanders in the field.

Under the terms of the new agreement, one or more U.S. representatives would review the daily output of Ultra material at Berkeley Street, another station at Ryder Street, and Bletchley Park in order to select items of value for American intelligence.

The 'Yellow Project," as Arlington Hall's early work on Enigma had been called, was transformed by the new agreement. Captain (later Major) Roy D. Johnson, who prepared from November to April at GCCS^[123] to meet the responsibilities of operations officer of an Enigma analytic unit under Lieutenant Colonel Frank W. Rowlett, instead went back to England in October as officer in charge of an overseas Signal Security Detachment. The new "Operation BEECHNUT," as authorized in July, sent officers and enlisted men from Arlington Hall and Vint Hill Farms Stations for duty in three such detachments plus one headquarters unit that together became the Special Project Branch of the Signal Intelligence Division, ETOUSA. They included intercept operators, machine processors, and cryptanalysts. Major William P. Bundy became the operations officer of the Branch. The three Signal Security Detachments were numbered 6811, 6812, and 6813 and were separated of necessity.^[124] The intercept unit went to a site on the heath near Bexley in Kent and set up a station known as "Santa Fe." By dint of hard work they responded to training that enabled them before June 1944 to count about 100 reliable operators in the unit. In the latter part of 1944 they were the subject of a commendatory letter from a British consumer. The second detachment had working quarters at East-cote but were housed at Harrow-on-the Hill, seven miles away, until they could get a camp built, in part by their own labor. The analytical group began arriving in September and were taken in at Bletchley Park, where an individual's progress determined his transfer from training to operational work. The branch was slow in grasping the opportunity opened for it in July, largely because of the necessity of obtaining G-2 clearances before shipment from the U.S. and also because of the unexpected amount of transportation and housekeeping duties that their separate sites required. It was in March 1944 or later that the Special Project Branch could begin to look confidently toward its objective. At its peak strength, it numbered about 36 officers and 400 enlisted men, of whom some had been found in the theater.

The intercept operators of the 6811th Signal Security Detachment were at first inattentive to the wisdom of fully recording German communicators' chatter. An example of what could result from alertness was reported later by Albert Small,^[125] the SSA liaison officer at Bletchley Park. A traffic analyst noted in a log that one German operator, at a time of callsign changes, had asked another if a certain callsign was that of an individual with a well-known long and distinctive name. The man had been a sergeant, and could have been promoted to lieutenant. Trying his name as a crib in a signature, Bletchley Park soon broke the key for that day's transmissions.

U.S. cryptanalysts who went to GCCS to learn the ways in which work was organized and the methods by which results were attained found themselves engulfed in British nicknames and slang by which cryptosystems, analytic operations, processes, and units were identified. It illustrated the epigram about the British and the Americans as peoples who were separated by the same language. Foreign cryptographic systems, for example, were classified for identification in categories named by fish, flowers, animals, birds and insects, rather than by letter-and-number symbols, as they were at home. Colors were also used, as they had been in Washington.

Rapid analytical machines and auxiliary processing devices acquired distinctive names on which much ingenuity was expanded. The "bombe" – for which the Poles are credited with providing the name that the British and Americans retained -was made more efficient by using a "Duenna" in the latter part of the war. Resort to electronics rather than electromechanical operations produced for GCCS both a "Giant" and ultimately a "Colossus."

American visitors had to recognize the meanings of "blist," "cillies," and "Uncle Dudley," as well as "Duddery," and scores of other terms that served as oral shorthand. Such inventiveness was inherent, apparently, in the mental equipment needed in producing SIGINT. Machines devised by the Americans at home accumulated a roster of esoteric designators.

One comprehends the extent to which the use of special terms at GCCS amounted in effect to a code upon reading the following sample:

At "Berkeley Street," Taylor served until relieved late in July 1943 by a civilian, Mr. Roger Randolph. Captain Bancroft Littlefield followed Mr. Randolph in December, and about one year later, Captain Lewis T. Stone, Jr., took over from Captain Littlefield. At "Ryder Street," Major W. L. Calfee acted as U.S. SIGINT Liaison Officer.

The American Ultra officers who came to Bletchley Park for similar duties were the nucleus of a unit in Hut 3 that came to be designated at Bletchley Park as "3-US." Major Samuel McKee joined Taylor in selecting the first of a stream of "CX/MSS items" (Ultra) for transmission to Washington, always via British secure communications channels. The first messages were sent on 27 August 1943.

Principles of selection rested more on trial and error than on doctrine. Guidance from Washington was minimal. Almost at once, Taylor and McKee recognized that Ultra messages provided far better data on German Army and Air Force order of battle than the materials two U.S. liaison officers, assigned to duty at the War Office and the Air Ministry, had been obtaining from other types of SIGINT material. Their selections in Hut 3 soon overloaded the available British secure communication circuits and had to be restricted.

In Washington, where the MAGIC Daily Summary became so important to the conduct of the war, a narrowly distributed Military and Naval Supplement was used as the vehicle for SI. On 1 July 1944 it was replaced by a daily MAGIC European Summary prepared by the German Military Reports Branch of MIS. About ninety percent was based on special intelligence.

The five or more U.S. officers in Hut 3 were led by Major F. W. Hilles, while Taylor (promoted to lieutenant colonel) and McKee worked on arrangements for training American Special Security Officers in disseminating Ultra to U.S. commanders. The recruitment was done in Washington where the training there was preliminary; training at GCCS was the main, final stage, and chiefly accomplished in "3-US." All personnel were accounted for as the strength of "MIS, War Department, London Branch."^[126]

Several of the future Special Security Representatives completed their training by visits to the Mediterranean area commands where Special Liaison Units (SLUs) were functioning. The first visitor, Captain John F. B. Runnals, went to Algiers, La Marsa, Bari, and Caserta in November-December 1943 not long after the liberation of Naples. At AFHQ in Algiers he found the principal British Special Security Officer (SSO) in the theater, Lieutenant Colonel Robert Gore-Brown, from whom he obtained an explanation of the functions of nine SLUs in NATOUSA, six others in Egypt and the Near East, and a special communications network extending from "Windy Ridge" in the United Kingdom to Gibraltar, Algiers, La Marsa, Taranto, Bari, Foggia, Caserta, and Bastia. Further forward he ascertained the organization of SLU activities.

In April, when Major Lewis F. Powell, Jr., began his trip to the Mediterranean area, he carried instructions (from the chief of staff, Army) for certain U.S. air and army officers governing the protection and use of special intelligence. He was also to observe the SLU system as employed by the Mediterranean Allied Air Force and constituent commands.

The personnel of MIS, War Department, London Branch were of three categories: Advisors, Special Security Representatives (SSRs), and Special Security Officers (SSOs). The first type worked in Hut 3, Bletchley Park, or in a London British SIGINT station. The second and third types were to be assigned to the headquarters of U.S. commanders, as members of Special Liaison Units controlled by the British. They would work in offices, vans, or possibly briefly in tents, near a Special Communications Unit (SCU) of British RAF noncommissioned and enlisted communicators. The SCU in its own vehicle would be using a "Type X" machine – a cryptographic system that was reserved for the protection of special intelligence, in radio (or undersea cable) channels.

The SSRs were expected to be closely associated with the chief of intelligence, chief of staff, and commanding officer of a headquarters. The SSOs, depending on the preference of the SSR, would work for him, or with him. The SCUs kept in touch with the central station in the UK and delivered deciphered messages to an SSO. If the message was urgent, it would be taken at once to the SSR for action; otherwise, it would be handled as the others delivered at stated times, or used in daily briefings and summaries. The SSO office or van could be used as an Ultra library where officers "on the list" could consult files and maps showing information from SI. The maps not only presented enemy order of battle but the enemy's beliefs concerning the disposition of Allied forces.

During the planning to invade Normandy, the planners benefited from available special intelligence. The Eighth and Ninth Air Forces, then engaged in preinvasion bombing, also found Ultra useful in choosing targets and escorting the bombers. The tactical air commands that would support the armies prepared to make use of SI. Later forty-three American officers were assigned to Special Liaison Units. U.S. Army officers who served as special security representatives in the European Theater are listed below opposite the command headquarters to which they were accredited.

Future SSRs visiting in Algiers saw the segregated operational areas of the SLV and the SCV on the top floor of a hotel used as Allied Headquarters. They observed the receipt of Ultra messages from GCCS, their conversion into plain text, and the making of copies for delivery to persons on the Ultra list. They learned that record copies were filed and retained for certain periods, while others were recovered and destroyed after being read. SSRs read the messages and observed how they were used as sources of strategic or tactical intelligence. The pattern of briefings, mapping, summaries, and spot alerting was noted. They were told of the variety of security problems, including those arising from the association of officers uninformed about "Y," or unaware of SI, with those on the Ultra list, and the necessity of persuading commanders not to take military action traceable only to special intelligence. Those who visited in early 1944, or after the liberation of Rome, were able to note the circumstances and methods of SI service at 15th Army Group (Allied Armies in Italy), Mediterranean Allied Air Force, and U.S. Fifth Army.

Before the invasion of Normandy and while still in the UK, many SSRs had learned their way around the headquarters with which they would be associated in France. At the headquarters of U.S. Army or Air Force commands to which SLUs were provided, the presence of officers and enlisted men in British uniforms naturally aroused some attention, for they were conspicuously different from others. Their places of work were usually less prominent. Curiosity about them was deflected in some cases by imaginative titles for their units or, in the case of individuals, by the nature of their supposed specializations. On an intelligence staff a man would have had to be dull indeed not to recognize that a colleague with seemingly superlative powers of deduction might possibly have a special source of information. Awareness irritated some of the intelligence staff who were not on the SI list, and aroused envy in others, but most shelved their concern.

The term "special intelligence" was inclusive. Codewords like MAGIC and Ultra were preceded and followed by others. The War Department issued several successive Secret security directives for the protection of SIGINT with the titles: "Security of Radio Intercept Intelligence," then "Security of RABID Intelligence," then "Security of ULTRA DEXTER Intelligence within the European, North African, and Middle East Theaters of Operations."^[127]

On 24 August 1944 SHAEF issued revised instructions for protecting SIGINT of all grades. Besides specifying that American and British security practices were to be uniform and to follow certain methods, those instructions distinguished between crypt intelligence and traffic intelligence and assigned codewords as follows:

CIRO-PEARL – for medium-grade codes and ciphers.

PEARL – for low-grade codes and ciphers.

THUMB 1 – for traffic analysis, excluding direction finding.

THUMB 2 – for DF and other technical aids.

The regulation covered handling, and defined the recipients of different types of SIGINT.^[128]

To quote from a study prepared after the war, "Synthesis of Experiences in the Use of Ultra Intelligence by U.S. Army Field Commands in the European Theater of Operations,"^[129] concerning security measures:

The representatives, who worked and lived with Ultra in the field, were aware that it often had a direct operational value; one stated:

U.S. officers in Hut 3 selected the items to be passed directly to any of the U.S. commanders or to be used by the Intelligence Division of his staff in preparing briefings and summaries. At the commands, the way in which the items were used depended on the commander's attitude and the personalities of those directly involved. An intelligence briefing early each day could be considered normal. Ultra, if its role as the source could be masked, might furnish some of the information. If not, or perhaps in any case, a separate briefing to officers "on the list" could precede or follow the regular intelligence briefing. At some headquarters, such as that of General Jacob De-vers, CG, 6th Army Group, it was given in the SSO's own small office with maps that showed Ultra data. At General Bradley's 12th Army Group the Ultra unit, disguised by the title "Estimates and Appreciation Group," gave its own separate briefing in a trailer adjacent to that of the G-2.

Some SSRs served also as the "Y" officer in a staff SIGINT unit. Their knowledge of current Ultra enabled them to guide the work on "Y" material and in reports or oral briefings to avoid whatever had to be treated separately.

Not all commanders held Ultra in the same regard, at least when first apprised. General Mark Clark's disdain in 1942, during the planning for Operation TORCH, has been described in books by Winterbotham and Lewin.^[131] General Patton's dislike of the restraints for security reasons on his movement into advanced areas, such as his flight as a passenger (with his dog, Willie) on a bombing raid over Germany, cooled his interest.

The basic agreement covering the respective responsibilities of OP-20-G and GCCS mentioned above had been signed in October 1942. It was a preface to assumption by OP-20-G of a positive role in producing SI. During December 1942 GCCS recovered the ability to read submarine Enigma traffic, now enciphered by the four-wheeled version of the machine. While having SI again applicable to U-boat activities was a great advantage, the German Navy's counterpart to GCCS, known as the B-Dienst, was able in the spring of 1943 to read enough of the Allied communications pertaining to the convoys so that German sinkings were disastrous. The Allies executed a communications change, the submarines then lost that advantage, and in May 1943 so many more of the U-boats were sunk as to induce a transfer of submarine offensive efforts to a different area of the Atlantic.

The German Navy renewed the submarine offensive in August 1943. The new U-boats with fresh crews were expected to score heavily with their recently developed, acoustically guided torpe- does. SI had alerted the Allies to the plans and to the new weapon. Allied ships nullified the torpedo by towing a device that drew a torpedo away from the target ship. On occasion, when more than one U-boat was engaged in an attack, one was probably struck by an acoustic torpedo fired by the other. The Allies at that time were enabled by SI to direct allied aircraft to rendezvous points assigned to German submarines refueling at sea from surface or underwater tankers. Destruction of such supply ships greatly restricted the range of a submarine's voyage or obliged it to deplete the fuel of a sister ship. Aircraft from small escort carriers which protected a convoy often were able to drive attacking submarines underwater, or to damage them with depth charges.

By 1944 OP-20-G was feeding SIGINT to F-21 (Atlantic Section, Combat Intelligence Division, U.S. Fleet) the Operational Intelligence Center, Admiralty, and to Ottawa. Eventually the British would agree that the Americans were bearing the major share of the SI load in the Atlantic and carrying it satisfactorily.

Special intelligence relating to the probable opposition to the Allied invasion of Normandy enabled the planners to know that the enemy had a fair idea of the general region but not of the exact places where the attacks would be made. The Germans had diverse calculations of time as well as place, so that the Allies could expect them to remain deployed to meet a wide range of possibilities. The enemy was generous enough to provide data two weeks before the invasion in messages showing the disposition of German forces in the Cotentin peninsula, including recent strengthening there, and the chain of command under Field Marshal Erwin Rommel as commander-in-chief, Army Group B.

Special intelligence afforded the Allied command a good understanding of what the enemy expected. As early as 10 January 1944, German Army Intelligence was aware of the First U.S. Army Group (FUSAG) and connected it with the expectation of a cross-Channel attack in the vicinity of Calais.

The Allies, partly from special intelligence and partly from other sources, knew that at the time of the invasion OB West controlled Army Group B and Army Group H; that is, four Armies and a Panzer Group West, comprising sixteen operational corps in all.

The Allies knew that Rommel's intention was to defeat the invasion at the beachheads and that his panzer divisions were to be held in readiness to move to the actual beachheads as soon as each was disclosed. Only one panzer division's location was not established; the 2d Panzer Division in Brittany might be near enough to reach the battlefield in Normandy during the critical early days. Allied air plans to "isolate the battlefield" by bombing were intended to keep out of the battle any German armored unit from Rommel's mobile reserve.

The strategic plans of the enemy were supplemented by what was learned on the eve of the invasion, for on 30 May 1944 the main cryptosystem in use by Rommel's Army Group B in an area extending from the Netherlands to southwestern France became readable at GCCS.^[132]

During the week of the invasion, the volume of traffic encrypted on relatively few keys enabled the analysts at GCCS to recover three of the four main army keys for 9 June 1944. On that day, intercepts at British stations alone were about 17,000 in numbers, but the successes gave no promise of reading current German Army traffic; German Air Force traffic was more available. An early illustration of the operational use of special intelligence by the Ninth Tactical Air Command, supporting U.S. First Army (FUSA) in Normandy, can be mentioned here.

A German FW-190 pilot was captured and interrogated at a wing headquarters of the Ninth TAC. He stated that his unit was based near a town which he called Essay. The Wing A-2, who knew of an enemy airfield at Lessay but not of a place called Essay, recommended the bombing of Lessay.

At Headquarters, Ninth TAC, where special intelligence was received, the SSO recalled a German message that had mentioned fighter-group landing grounds at Essay and Lonrai, information not yet included in general intelligence. He also knew from other special intelligence that the field at Lessay had been trenched, ploughed, and abandoned. So he advised that the wing not bomb Lessay and that the area described by the PW be reconnoitered. That mission yielded photographs of three landing strips – Essay, Lonrai, and a previously unknown one at Barville. The next day all three fields were successfully struck.^[108]

As Major Taylor had foreseen during the negotiations early in 1943, U.S. acceptance then of British restrictive conditions governing the production of Ultra was followed by gradual British accommodation to U.S. needs. Before the Nazi surrender ended hostilities in Europe, the Signal Security Agency had sent enough persons for prolonged training in production of SI at GCCS to operate independently elsewhere. When the war shifted to the Pacific and to Japanese communications, where the U.S. had the primary responsibility for SIGINT, U.S. production of relevant special intelligence reflected the prowess of Arlington Hall and OP-20-G and their forward centers. Aware of the situations of Japanese forces on various islands and atolls, Admiral Nimitz and General MacArthur advanced upon the home islands leaving thousands of Japanese soldiers and airmen stranded and far from the battle zone.

In this general chapter concerning British-American collaboration in special intelligence, we have anticipated the conclusion in Europe of a conflict that we now again consider in somewhat greater detail. We have been describing Ultra matters and now return to PEARL and THUMB; that is, we leave special intelligence to resume attention to tactical SIGINT.

^[107]

^[108]

^[109]

^[110]

^[111]

^[112]

^[113]

^[114]

^[115]

^[116]

^[117]

^[118]

^[119]

^[120]

^[121]

^[122]

^[123]

^[124]

^[125]

^[126]

^[127]

^[128]

^[129]

^[130]

^[131]

^[132]

^[133]