The Secret in Building 26

by Jim DeBrosse

  Just as important as the development of new anti-sub technology was the Allied ability to adapt their training and tactics to put those new weapons to best use. In May 1943, the U.S. Navy at last created an organization to pull its fragmented campaign against the U-boats into focus: the Tenth Fleet was the new central command for protecting the convoys and coordinating the attack against the subs. The new fleet, which had no ships, was basically a reshuffling and regrouping of existing units that had sprung up haphazardly during the war in response to specific needs. Under the command of Rear Admiral Francis S. Low, “a tough, conscientious, intelligent and hardworking officer,” the Tenth Fleet cleared the way not only for improved coordination but for uniform training of personnel, faster dispersion of new anti-sub devices, and better instruction in their use.

  The Germans tried to counter the Allied technological advantage with new weapons and new tactics of their own. As early as April 1943, the Nazis had begun equipping many of their subs with powerful new antiaircraft weapons, usually a group of four twenty-millimeter machine guns placed in the conning tower and a thirty-seven-millimeter antiaircraft cannon mounted on a platform (dubbed a “bandstand” by the Allies) behind the tower. The combination proved daunting for many Allied aircraft, including the heavily armed long-range bombers.

  Dönitz eventually made a “fight-back” strategy an edict for his sub crews in September 1943. The subs were to cross heavily patrolled areas in teams, then circle and fire back with antiaircraft guns whenever Allied planes swooped in. Dönitz reasoned that the U-boats were most vulnerable, particularly to air-dropped depth charges and Fido homing torpedoes, during the thirty seconds to a minute that it took them to crash-dive.

  In turn, the Allies countered the “fight-back” strategy by sending out patrol planes in pairs. By staying on the surface to fight, the U-boats only improved the chances of one of the two planes scoring a hit. It is no surprise then that the success rate of Allied aircraft attacks doubled in the last half of 1943, with one in four resulting in sub sinkings and 40 percent causing damage to their targets.

  The Allies also found better ways of detecting U-boats, including more powerful and accurate sonar—the supersonic echo-ranging equipment for finding subs underwater. In addition, small high-frequency direction-finding sets, known as HF/DF or Huff-Duff, were installed on ships to widen the existing land-based net for homing in on radio signals from surfaced U-boats and tracking their general direction. To search for submerged U-boats from the air, pilots could drop sonobuoys—floats equipped with hydrophones and radio transmitters.

  The single most significant improvement for detecting surfaced submarines was the introduction of microwave anti–surface vessel (ASV) radar, which was more accurate and reliable than conventional radar and took up less space and weight so that it could be installed in aircraft. The Germans, who struggled until the end of the war to develop a submarine that could stay submerged indefinitely, considered airborne ASV radar the biggest reason for their defeat in the Atlantic.

  Allied air coverage for the convoys was nearly universal by the spring of 1943. To fill in uncovered regions, such as the notorious “black pit” in the central Atlantic near Greenland, more long-range and very-long-range aircraft were based in Newfoundland and Iceland. Meanwhile, the role of the small escort carriers like the Bogue, known also as CVEs, was expanded so that they could hunt and kill submarines as well as provide continuous air coverage for the convoys. Converted from merchant vessels, the “baby flattops” could transport up to thirty planes and reach speeds of more than eighteen knots—faster than most Nazi U-boats, even while surfaced.

  Admiral Low planned to attack the Nazi U-boat fleet when and where it was most vulnerable: the 1,600-ton milk cows. The big U-tankers allowed the combat U-boats to stay far out at sea. Eliminating the German ability to refuel in mid-ocean would reduce the number of U-boats on patrol. With the promised help of timely breaks into Shark, the Navy’s antisubmarine forces planned to attack the rendezvous points of refuelers—with the chance of surprising multiple subs as they cruised the surface looking for one another or as they were refueling.

  As the British warned, however, the opportunity also brought with it a great risk: too much success in sinking the milk cows would lead the Nazis to question the security of their submarine codes and perhaps put an end to Ultra. But Admiral King insisted that the potential benefits were too great. In an April 27, 1943, telegram to the British Admiralty, he wrote: “While I am equally concerned with you as to security of ‘Z’ information it is my belief that we are not deriving from it fullest value. The refueling submarine is the key to high speed, long range U/boat operations. To deprive the enemy of refuelers would at once decrease the effectiveness and radius of entire U/boat deployment.”

  In mid-June, the escort-carrier commanders—once given the signal that their convoys were within coverage of land-based aircraft—were free to attack known concentrations of U-boats. Along with his new orders to the escort commanders, Low later wired that he would release special intelligence “that will enable you to more quickly and positively” find U-boat targets.

  Much of that special intelligence was expected to be generated by the U.S. Bombes, but their operational usefulness was still several months away. OP20G’s own history claims that a test run on a Bombe in Dayton found the Shark keys and broke the messages that helped guide the Bogue and McAuslun’s Avenger to U-217 on June 5. But log sheets for the Shark keys that month show a British solution, and the official OP20G war diary says the Dayton Bombe didn’t tap effectively into Shark for the first time until June 22—decrypting a May 31 message.

  Regardless of who gets the credit for the Allied successes in early June, the British and Americans were by then collaborating closely to break into Shark, sharing keys, cribs, and equipment and providing each other with all U-boat-related intelligence. Bletchley solved the toughest parts of the Shark keys, such as the wheel order and steckers, then let the Americans clean up with the easier task of finding the lower-level keys, such as the starting positions of the wheels and ring settings. But even then, the Americans, often relying on hand methods, rarely beat GCCS to the easier keys.

  The British continued to front the attack against Shark all through the summer of 1943, even though it was a bad time for reading U-boat transmissions. In July, the Germans introduced a new reflector on their four-wheel machine, and Dönitz’s withdrawal from the North Atlantic gave the codebreakers fewer messages to work with. In June, the British solved fourteen days of Shark keys, all within a week’s time, while the American Bombe program continued its struggle.

  Yet by summer’s end, America’s role as junior partner in Ultra began to change.

  THE JUNE 18 order from OP20G to Dayton must have seemed like a frightening case of déjà vu. The Navy wanted to scratch the existing Bombe design for a whole new codebreaking machine, with a semiautomatic method of switching rotors and greater machine speeds—all without delaying production. Desch was stunned. In effect, the Navy wanted a machine approaching the complexity of what we now call a modern computer. Leading theorists knew even then that electronic digital processing would become the basis for computers, but more practical types like Desch, who realized the limits of industrial production, also knew it would take years to put such a machine into operation. Desch had told the Navy as much in the fall of 1942, but here they were again with their electronic pipe dream.

  Desch went to Meader and pleaded his case. There simply was no practical, high-speed substitute for the many hardwired rotor wheels that could be rearranged on the Bombe’s drive spindles. Thousands of tubes in very dense circuits would be needed to imitate all the possible rotor wirings of the sixty-four wheels on a Bombe. The energy requirements and the strain on precious war materials would be staggering. And the tubes would generate so much heat that the machines couldn’t be depended on to operate reliably.

  Meader, who normally went along with whatever his naval superiors wanted, wa
s quick this time to back Desch. He dashed off a message to Engstrom that same day, arguing that too much money, manpower, and scarce materials had already been invested in the first Bombe model, which was now so close to production it deserved a chance. The Navy had little choice but to renew its faith in Desch and hope that his hundred machines, when they finally reached Washington, would do the job.

  By the last week of July, fifteen Bombes had been assembled at NCR, but none would work properly for long. “We were plagued with false stops or with no stops at all,” said Gilman McDonald, one of five naval officers assigned to Building 26 in mid-May to study how best to operate the Bombes. Both Desch and Howard “were completely mystified” by the behavior of their production models, he said.

  It was perhaps Desch’s darkest hour. If the Navy scrapped the project, a year of intensive work and millions of dollars would be declared a waste, and he, along with Wenger and Engstrom, would take the brunt of the blame. But Desch relied again on his practical side, and at the very last minute he made a discovery that revived hope: the faces of the rotors still were not flawlessly smooth, causing the sensitive brushes to bounce and trigger false stops. “All rotors were then sanded with fine emery cloth on a flat metal surface” to smooth out the distortions, McDonald said.

  But solving one problem immediately led to another: the machines failed to arrive at the correct stops. The tiny copper slivers from the sanding of the rotor faces were bridging the near-microscopic gaps between the rotor contacts—a mere twenty-thousandth of an inch—and shorting out the legitimate stops.

  To solve both problems, Desch insisted on a rigorous program of preventive maintenance: all rotors and brushes, on each bank of every Bombe, would be sanded and cleaned of copper slivers on a regular basis. The maintenance crew developed tiny tools and blowers for cleaning out the minuscule gaps between the commutator contacts. Desch requested that an entire new maintenance staff and routine be established to make sure the delicate tasks were done right.

  Again, the Navy trusted in Desch’s judgment, and the project was saved.

  August 3, 1943—Central Atlantic

  AFTER NEARLY FOURTEEN weeks of hunting in the Americas, with two kills to their credit off the East Coast, the crew of U-66 was more than halfway home, skimming along the Atlantic’s sun-dazzled surface 457 miles west-southwest of the Azores. Thinking the worst of their mission over, the watch crew was caught off guard when the sub was spotted by a Wildcat-Avenger team from the escort carrier USS Card. The pilots had not found the U-boat by accident. They had been alerted by the Tenth Fleet that U-66 would be in the vicinity to rendezvous with U-117, a milk cow, and refuel for the trip home—information gleaned from Shark intercepts.

  Wildcat pilot Arne S. Paulson bore in on U-66 and strafed the deck and conning tower, instantly killing the deck officer. The sub’s skipper, Kapitänleutnant Friedrich Markworth, ordered battle stations and refused to submerge until Avenger pilot Richard L. Cormier followed in to drop his depth charges and Fido homing torpedo. Both weapons, however, failed to release. Markworth dived and resurfaced to fight again, only to be wounded in the stomach by more of Paulson’s strafing. Cormier made a second run in his Avenger, this time unloading all his bombs. Though badly damaged, U-66 dived a second time and managed to escape once again.

  But inside U-66, the crew was in turmoil. In addition to Markworth, the first watch officer and a midshipman had been seriously wounded. The men waited patiently until the cover of night, when the sub could resurface and radio headquarters for help. Dönitz ordered the sub to a nearer rendezvous point with U-117, but once again the Allies decrypted the time and location.

  Four days later, the two subs met in the early-morning hours. U-117 sent over its doctor to tend to the wounded and its first watch officer to fill in for Markworth. As daylight broke clear and bright, the tanker started refueling U-66.

  Just minutes later, Avenger pilot Asbury H. Sallenger found the two submarines, about five hundred yards apart. Both subs let loose with their flak guns, but Sallenger attacked anyway, bombing U-66 and squeezing off a few machine-gun rounds at the deck of the milk cow. Sallenger hovered out of range for twenty-five minutes until more planes from the Card showed up. When U-66 started to dive, Sallenger went in again to drop a Fido but missed the smaller sub as U-117 opened up with its guns. Two arriving Wildcat pilots strafed U-117 “unmercifully,” according to ship logs, before two more Avengers bore in with depth charges and Fidos, finally sinking U-117 with all hands.

  U-boat headquarters was still unaware of the loss of U-117 when it ordered the tanker to cancel its New York harbor mine-laying mission and to serve as a refueler west of the Canaries. But the only U-tanker in the Atlantic that month was now debris. The submarines that had counted on it for fuel were forced to limp home. All in all, the sinking of U-117 was a persuasive example of King’s offensive strategy.

  As planned, the Allies did indeed nearly wipe out the Nazi U-tankers in the summer of 1943, sinking nine of twelve milk cows in operation at the time. Shark decrypts played an important role for the Americans: two of the three refuelers sunk by U.S. escort carriers were located with the aid of radio intelligence. Still, most of the milk cows fell victim to the intense air campaign waged by the British in the Bay of Biscay.

  The Kriegsmarine may have been on the run, but it was far from giving up. Dönitz ordered the U-boats back into the North Atlantic in September, armed with more powerful antiaircraft weapons, new decoys, and the T5 acoustic torpedo (the Zaunkönig), which could home in on the propeller sounds of escort vessels. But the Zaunkönig failed to give the Germans the upper hand, partly because the Allies were already towing the Foxer noisemaker to divert the torpedoes.

  The American Bombes were too late to help turn the tide in the Atlantic that summer, but the battle there and the war itself were far from over. Despite their successes, the Allies were still projecting a net increase in the number of U-boats by year’s end. What’s more, they still rightly feared that Dönitz would try to renew the war against Allied shipping with newer, faster U-boats and more deadly weapons.

  Enigma decrypts would be vital to helping the Allies keep track of updates in the U-boat arsenal, including the development of engine snorkels that allowed the subs to stay submerged longer, new microwave equipment for detecting Allied radar, and still more powerful antiaircraft guns. Dönitz especially clung to the hope that a new generation of faster, longer-submerging U-boats would be able to attack Allied ships and speed away without detection. He believed the new U-boats could turn the tide back in favor of the Nazis—if only his embattled crews could maintain their fighting spirit to the end. Even late in the war, he was rallying his U-boat commanders to carry on the war with renewed intensity. In a report transmitted and intercepted on October 22, 1944, Dönitz told his sub fleet that “a new U-boat war must and shall be our most important goal in the war against the western powers. The Navy and its equipment services must work for that end with fanatical energy.”

  Through the summer of 1943, the engineers and technicians at NCR continued their fight as well, their long hours unabated. In July, the Navy’s faith and insistence that Joe Desch could work out the glitches in his Bombe paid off, and the early production models began to show that they could do the job. By August, the machines scored their first useful break into Shark, within a week of message transmission.

  IN AUGUST 1943, Bletchley’s worst fears about American security seemed to be confirmed: British codebreakers decrypted a stunning message to the Abwehr, Germany’s intelligence department, that pointed to a traitor in the highest civilian ranks of the U.S. Navy Department. The decrypt read: “On August 10, the following message came from KO Switzerland: ‘For several months coded German orders to operating U-boats have been read. All orders are being read. Addendum: Source: a Swiss-American in a high position in the U.S. Naval Department.”

  Fortunately for Ultra, German security officials once again pooh-poohed the notion, as they had so often
in the past, that the Allies were breaking into the Enigma with any kind of regularity. The message went on:

  Chief MND [Military Intelligence Service] excludes the possibility that radio traffic is being read on a continuous basis by the opponents. Naturally, the possibility is always open that a complete set of code equipment with an adjusted key fell into the hands of the enemy from a lost boat. Protection should be given against this possibility by a change of the key word “Andromeda.”

  There are no records that the Navy ever accused or attempted to prosecute anyone for leaking the Ultra secret to the Nazis via a Swiss agent. But the list of possible civilian suspects within the Navy was very short, leaving open the question of whether the suspect was in a high enough position of status and power to elude prosecution. Regardless of the identity of the man who may have sought to betray the Ultra secret or whose loose lips may have done the same, he couldn’t stop the U.S. Bombe from entering production in the fall of 1943 and ultimately shortening the war.

  STANDING GUARD ON the night of September 11 by a railroad siding in an empty field behind NCR’s Building 26, Navy Midshipman Raymond Torchon kept watch as huge wooden crates were rolled on dollies from a truck bed and into a waiting baggage car. In all, six sealed crates, each several feet taller than a man and longer than a pickup, were brought from a warehouse at the rear of Building 26. Torchon had no idea what was inside the giant boxes, and he had sense enough not to ask. “All I knew, it was NCR, and I thought they were cash registers,” he said in an interview in the fall of 2001. “What did we know? Nobody knew—not for fifty years.”

  Torchon was one of four Navy sailors assigned to guard the baggage car and its top secret cargo until a train engine arrived at 5:00 A.M. the next morning. The guard detail was then to escort the shipment to Washington, D.C., delivering the first six of some 120 U.S. Bombes to OP20G’s codebreaking annex near Tenley Circle. Every week, the same size load left from the same railroad siding, with Torchon and his fellow guards sometimes catching up on their sleep atop the giant boxes. Hidden inside each was a mammoth N-530 production Bombe.