by Clay Blair
for the primary purpose of studying [general] British cryptanalytical research methods but with additional instructions to find out all details possible concerning the British “E” [Enigma] problem. Under the impetus of the presence of these two officers, considerable detailed information was forthcoming concerning the British bombe and wiring diagrams.‡
That is, the British were much more forthcoming than they had been when U.S. Navy officers Prescott H. Currier and Robert H. Weeks visited Bletchley Park in February of 1941.§
Engstrom continued:
As the months passed, it became more and more evident from reports received from our representatives in England [Eachus and Gaschk] that the British would be unable to supply us a machine [bombe] by the promised date.# It became evident, furthermore, that the British were having considerable difficulty in building any workable high-speed machine. Accordingly, early in September after continuous conferences, we reached the conclusion that American methods and design showed sufficient probability of success to inaugurate the bombe program.
After clearing it with superiors up the line, in “early September” Joseph Wenger, head of the U.S. Navy’s codebreaking outfit, OP20G, met with Edward G. G. Hastings, the Bletchley Park representative on the British Joint Staff Mission in Washington. Wenger told Hastings that since the British “had shown no evidence of being able to live up to their promises,” the U.S. Navy would proceed alone with development and production of a high-speed four-rotor bombe to attack four-rotor naval Enigma. Hastings countered that the British had indeed lived up to their promises, but his protests fell on deaf ears. On September 3, 1942, Wenger formally proposed to the Director of Naval Communications, Joseph R. Redman,* that the U.S. Navy proceed with a bombe project. Redman readily approved, as did the Assistant Chief of Naval Operations, Frederick J. Home, who signed off officially on September 4.
From a technical standpoint, the addition of a fourth rotor to naval Enigma “introduced a factor of twenty-six in the time required for a solution,” Engstrom wrote. To find the daily key settings “necessitated either twenty-six times as many [three-rotor] bombes or a bombe which would go twenty-six times as fast.” Based on his theoretical work in 1941 and 1942, Engstrom had concluded that the American effort should be directed “to high-speed approaches to the problem” as he wrote Wenger.
The OP20G proposal did not specify how many bombes should be built. The Americans thought at first that the figure should be 336. In his memo to Wenger, Engstrom explained why:
In the original concept of the problem it appeared that 336 units were desirable since there are 336 possible wheel settings. In forming preliminary notions of the size and power requirements for the equipment, the British 3-wheel bombe was in mind. The British 3-wheel bombes have three levels in each physical piece of equipment. Thus the concept of 336 bombes led to the notion of 112 separate pieces of equipment.
The early decision to build 336 units was predicated upon the belief that we should have to provide for running all 336 wheel orders. As study of the problem progressed, it became evident that there were analytical ways of ruling out many of the wheel orders, thus reducing the bombe requirements. The possibilities of making this reduction were given very serious study for the reason that materials at the time were exceedingly critical. ... Meanwhile, certain changes in German communications which affected cross-cribbing and indications of procedure changes involving the use of the fourth wheel arose. These changes made it impossible to arrive at a definite decision on the number of machines required before it became necessary to proceed with designs for housing the equipment,†
The Americans chose the National Cash Register (NCR) Company in Dayton, Ohio, to manufacture the bombes. A Dartmouth College graduate (1919), inventor, and Navy reservist, Ralph I. Meader, then serving in the Bureau of Ships, went to NCR in Dayton to represent the Navy. On November 11, 1942, the Navy began construction at NCR of the U.S. Naval Computing Machine Laboratory, “for the purpose of assisting the contractor in the production of these bombes and in the training of maintenance and operational personnel.” Meader became officer in charge of this facility, which soon grew to include hundreds of enlisted men and Waves who actually built the bombes.
The National Cash Register Company, which invented and produced the first such “cash register” to bear its name, was the most prominent of numerous Dayton manufacturing facilities, almost all engaged in war production. In addition to the cash register, Daytonians had invented or exploited a variety of mechanical and electromechanical devices: the airplane (Wright Brothers), the automobile starter, bicycles, refrigerators, and radios.
Another reason the Navy chose NCR was because of the presence there of a remarkably inventive thirty-five-year-old electrical engineer named Joseph R. Desch. A graduate of the University of Dayton (1928), Desch came from one of the city’s many large German families. In the darkest of the Depression years, from 1930 to 1938, Desch had worked as a practical inventor/designer for three prominent Dayton firms: General Motors Radio, Telecom Laboratories, and Frigidaire. In 1938, when NCR’s visionary president, Colonel Edward Deeds, established an electrical research laboratory, Desch came aboard to experiment with all kinds of exotic new electronic devices. His brilliant contributions in this field gained him many honors and association with other leading American inventors and scientists, such as Van-nevar Bush at MIT, who rose to direct the nation’s military science projects in World War II.
Dayton was the ideal place to make bombes but it was not a good place to operate them. It was too far from Washington and the “users” of the output of the bombes. Foreseeing this problem, Wenger and Engstrom were already carrying out plans to transfer all the bombes to Washington. To house them, Wenger had acquired a former girls’ school, Mount Vernon Seminary, at 3801 Nebraska Avenue, in the heart of residential northwest Washington at Ward Circle. Work was under way on a strongly built two-story building for the bombes and also quarters for the hundreds of Waves who were to operate and repair the machines, much as British “Wrens”* operated and repaired the British bombes at Bletchley Park. The new Navy installation was merely fifteen minutes by automobile from the Navy Department offices, located downtown on Constitution Avenue.
By coincidence perhaps, at this time the British made some sweeping changes in the Naval Intelligence Division and at Bletchley Park.
• Winston Churchill directed First Sea Lord Dudley Pound to sack the Director of the Admiralty’s Naval Intelligence Division, John H. Godfrey.† Promoted to vice admiral, Godfrey departed England by flying boat on September 24 to carry out a preplanned visit to Washington and Ottawa. He was replaced by Edmund G. N. Rushbrooke.
• At Bletchley Park, deputy director Edward Travis replaced deputy director Alastair Denniston as operating head of that swelling establishment. The era of quaint eccentricity had passed, the codebreaking historian David Kahn wrote. A “table thumper” who had risen through the “codebreaking side,” Travis was “a manager who could bring it into the modern era of cryptanalytic mass production.” Denniston remained as a deputy director of the civil side, a much smaller outfit.
• The brilliant, eccentric British mathematician and coinventor, with Gordon Welchman, of the first British bombe, Alan Turing, proved—not surprisingly—to be a poor administrator. In September, he was eased out of his job as chief of the naval Enigma codebreakers in Hut 8 at Bletchley Park, in due course replaced by a mathematician (and chess champion), Hugh Alexander. After working briefly on Fish and the teletype-decoding machines Robinson and Colossus, Turing left for the United States on the giant troopship Queen Elizabeth, arriving in New York on November 13, 1942. He remained in the States for the next four and a half months.
In Washington on October 2, the Royal Navy’s Captain Carl Holden and the U.S. Navy’s Captain Joseph Wenger entered into a written agreement outlining a complete and open exchange of information and technology regarding four-rotor U-boat Enigma and Italian and Japanese naval codes.
A brief summary of this agreement in Howe’s NSA history was deleted by censors. However, it was presumably a document that finally opened the way for a truly complete exchange of intelligence on U-boats.
The importance of this Anglo-American naval agreement cannot be too strongly stressed. The historian Bradley F. Smith wrote*:
It broke with the age-old tradition of caution and suspicion which admonished even closely allied states not to open their most secret cupboards because today’s comrades in arms could well be tomorrow’s dangerous opponents.
Even though the U.S. Army was shortly to face German armies in Torch (the Anglo-American invasion of French Northwest Africa set for November 8), the British refused to enter into a similar exchange with the Americans of German Army and Luftwaffe Enigma out of fear that the Americans would leak the secrets. Accordingly, Howe wrote, the U.S. Army entered into a contract with Bell Telephone Laboratories on September 30 to produce 144 bombes for a unilateral attack on German Army and Air Force (three-rotor) Enigma. These bombes were known as “Rapid Analytical Machines” (or RAMs) and rather than “rotors” they employed “relay switching” (stepping switches) similar to the Army’s clones of the Japanese “Purple” machines. In a little over one month, Bell Labs demonstrated a successful sample model, Howe wrote, and expected to fulfill half of the order (seventy-two machines) by April 1943.
The Army Chief of Staff, George C. Marshall, agreed with the British that at least for the time being there was no pressing reason why the British should share the fruits of German Army and Luftwaffe codes with the U.S. Army headquarters in Washington. Rather than run the risk of leaks in transmission or in Washington itself, Marshall arranged with the British to send a meticulously selected and screened cadre of U.S. Army officers to England. These officers were to be put in the Ultra picture at Bletchley Park, then head up small Special Liaison Units (SLUs) that were to be attached to U.S. Army headquarters in the Mediterranean and Europe, serving as conduits for British-produced German Army and Air Force Ultra information.
These arrangements between the British and the U.S. Navy in October 1942 would at last lead to a close working relationship between the two nations in naval intelligence matters. The U.S. Navy’s OP20G would work hand in glove with Bletchley Park to break into the Kriegsmarine’s Triton traffic. The U.S. Army codebreakers in Washington would concentrate on breaking Japanese military codes. The British were to provide to the American forces in Europe Ultra material from German Army, Luftwaffe, and other Enigma networks, appropriate to the prosecution of land and air warfare.
Alan Turing’s biographer, Andrew Hodges, wrote that the purpose of his trip to the United States in the fall of 1942 was to help American scientists and engineers at Bell Laboratories (in New York City) develop an absolutely secure high-level telephone link between London and Washington, the technology for which later became known to laymen as a “scrambler.” This very well may have been the case, but much more about this trip needs to be brought to light. Turing spent most of the first two months of his visit in Washington with Howard Engstrom and other U.S. Navy codebreakers, who were in the process of refining the design of the bombes to attack naval Enigma then under construction at NCR in Dayton. He did not report for work at Bell Labs, Hodges wrote, until January 1943. At first he was denied access to classified work, but after a dust-up on the highest levels he was cleared. Thereafter, he was allowed complete access to everything Bell Labs had undertaken, including, of course, its most important project, the RAM (or bombe), of which the U.S. Army had ordered 144.
Turing would no longer play a freewheeling and dominant role at Bletchley Park, which had grown to thousands of rigidly isolated workers in closely administered Ultra production lines. On this trip to America, the U.S. Army and Navy, both recently embarked upon producing an aggregate of about 250 different bombes, milked Turing’s brain profitably.
In his history of the U.S. Navy bombe project, Engstrom wrote that Alan Turing visited OP20G and the naval bombe facility at NCR, in Dayton, in December 1942. Afterward, Turing wrote a report in which he criticized several aspects of the American technical approach to the problem, although he conceded that “starting from scratch on the design of a bombe, this method is about as good as our own.” In his bombe history, Engstrom quoted criticisms from the Turing memo and remarked: “These quotations and other comments in the report indicate the considerable extent to which our design was at variance with British ideas and experience.”
Gordon Welchman, the coinventor of the original British bombe, wrote in his memoir* that the new Bletchley Park chief, Edward Travis, also removed him from his old haunts. Concluding his work in Hut 6, Welchman became Assistant Director for Mechanization at Government Code & Cipher School (GC&CS). Emulating Alan Turing, he sailed to the States on February 17, 1943, aboard the troopship-liner Queen Mary for an extended visit. He wrote that he established himself in the New York office of the senior British spymaster in the States, William Stephenson (the subject of A Man Called Intrepid). He spent Mondays to Fridays in Washington and weekends in New York, “where cipher machines were being built.” In Washington, he worked with U.S. military cryptographic experts, such as the Army’s Frank B. Rowlett and the Navy’s Howard Engstrom. Like Turing, he probably contributed ideas to the Army’s RAM program and the Navy’s bombe project.†
THE ARCTIC: CONVOYS PQ 18 AND QP 14
Unaware of the Allied decision to halt convoys to northern Russia in the months of July and August 1942, the Germans wasted many U-boat patrols and aircraft sorties in a futile search for convoy PQ 18 to Murmansk and its reverse, convoy QP 14 from Murmansk. Finally, on September 5, the Germans learned of the postponement and new scheduled sailing, from documents recovered in a British Hampden aircraft that crashed near Vardø, Norway, and from decrypts of some Russian Air Force radio transmissions.
The long delayed PQ 18, consisting of forty heavily laden cargo ships, some towing tethered blimps for protection from low-flying aircraft, and others streaming Admiralty Net Defense,‡ sailed from Scotland on September 2. The close escort and support force was massive: fifty-two British vessels, consisting of:
1 “jeep” carrier, Avenger
2 antiaircraft light cruisers
1 light cruiser, Scylla
21 destroyers
4 corvettes
7 minesweepers
9 submarines
4 ASW trawlers
2 tankers
1 rescue ship
Another twenty-five British warships and support vessels provided distant cover or carried out special missions in conjunction with the voyage of PQ 18. These included three battleships and six cruisers with appropriate destroyer screens and two tankers. By Churchill’s reckoning, PQ 18 required altogether “seventy-seven escorts” (as he told Roosevelt), a heavy strain on the Royal Navy, which also had to carry out a large assignment in Torch on November 8.
The “jeep” carrier Avenger was included not to defend the convoy from torpedo planes and dive-bombers, as many supposed, but rather to destroy or drive off Luftwaffe shadowers and U-boats. For that purpose she embarked a dozen old Sea Hurricanes (and another half dozen in storage) and three old radar-equipped Swordfish biplanes. Neither type of aircraft was suitable for the task. The Sea Hurricanes were too lightly armed. Owing to the slow speed of Avenger in convoy, the Swordfish could not take off with a load of depth charges. They could only reconnoiter, drop flares, and hope to guide destroyers to the U-boats.
The opposite-sailing convoy, QP 14, assembled at the northern Russian port of Archangel, which was more distant than Murmansk from the Norway-based Luftwaffe aircraft and is ice-free from July to September. Composed of fifteen cargo ships, the convoy was ordered to sail on September 13. It was guarded by twelve other British warships: two destroyers, four corvettes, three minesweepers, and three ASW trawlers.
The new German admiral commanding the Arctic area, Otto Klüber, planned to bring all available German sea power to bear against PQ 1
8 and QP 14. This time, the U-boats were to attack both convoys. The “pocket” battleship Admiral Scheer, the heavy cruiser Hipper, the light cruiser Köln, and five destroyers were to attack QP 14.* While Admiral Scheer was shifting from Narvik to Altenfiord to prepare for this mission, the British submarine Tigris spotted and attacked her but the four torpedoes missed. However, Scheer developed serious engine problems and was forced to abort. Later, when Hitler again cautioned Admiral Raeder not to unduly risk the big ships, the OKM canceled the entire surface-ship mission. The attacks on PQ 18 and QP 14 were thus to be restricted to U-boats and aircraft.
German reconnaissance aircraft found and shadowed PQ 18 on September 6 and 8. In response, Klüber brought to bear fifteen Norway-based U-boats. Initially, eleven boats that were ready (or soon to be ready) were to operate against PQ 18, and four boats in (or soon to be in) the eastern area were to operate against QP 14. On the OKM’s order, Dönitz temporarily diverted seven new Type VIIs sailing from Kiel to the Atlantic to positions near Iceland to intercept QP 14 and its heavy escorts at the end of the voyage. One of these, U-606, aborted to Bergen with an ailing skipper.†
Counting the U-606, Klüber had twenty-two U-boats ready or near ready for battle. Seven rushed at maximum speed to intercept PQ 18. Since Klüber had declared the primary target to be the “jeep” carrier Avenger, the boats were dubbed group Traegertod (Carrier Killer). Rolf-Heinrich Hopmann in U-405 was the first to make contact with the carrier on September 13. His beacon signals brought in Reinhard von Hymmen in U-408 and Hans-Joachim Horrer in U-589. All three skippers claimed they hit and sank 7,000-ton freighters, and Horrer claimed two hits on Avenger as well. Postwar analysis credited von Hymmen with sinking the 7,209-ton American freighter Oliver Ellsworth and Horrer with sinking the 3,600-ton Russian freighter Stalingrad, but no boat hit Avenger.