Japanese military engineers spent more than three years studying captured American B-17 Flying Fortresses, rebuilding several from wrecks, and flying them under the Rising Sun—a painstaking program that produced no Japanese heavy bomber in return but left its mark on smaller technologies. When the Japanese Army swept through the Philippines in late 1941 and early 1942, they found the wreckage of America’s strongest air force outside the United States. The Far East Air Force had concentrated almost all of its heavy bombers at Clark Field and Del Monte, and on the morning of December 8, 1941—hours after Pearl Harbor—Japanese bombers and fighters destroyed or disabled nearly all of them on the ground. At Clark, the mechanics walked the line for days afterward, taking a wing from one wreck, a tail from another, and four undamaged engines off a burned airframe.
Out of 17 ruined bombers, they assembled three flyable aircraft. It bought nothing. The Japanese attacks kept coming, landings followed, and by December 10 the entire 19th Bombardment Group had 18 Fortresses left in the islands, only 12 of which could fly. Survivors were ordered south to Australia, and everything that could not be flown out was destroyed in place.
The field was given up on December 26. The team that came to inspect the wreckage belonged to the Giken, the Japanese Army’s Aviation Technical Research Institute at Tachikawa. Its assignment was to identify the newest American equipment. At Clark, its attention went straight to the Fortress.
Japanese magazines had run photographs of the latest model, the B-17E, the previous August, around the same time American readers first saw them. No E-models were at Clark, but the turbo-superchargers bolted into the engine nacelles of the burned D-models were exactly what they wanted. The Japanese took those, took every spare part that could be unbolted, and then did what the American mechanics had done—built an airplane out of the pile. The rebuilt D-model lifted off Clark’s runway later in 1942 with a flight crew sent down from the home islands.
It still carried its American serial number on the tail, 40-3095, with the stars painted over by red discs. Similar teams moved in behind the assault troops everywhere. The Army sent its people out from Tachikawa, and the Navy sent its own from the First Air Technical Arsenal at Yokosuka. Over the next three years, that arsenal worked through a Hellcat pulled out of Okinawa, an Avenger left behind in the Indies, and a Corsair downed near a flight school.
The largest capture came on Java. Nine Fortresses lifted off Singosari Airfield on February 8, 1942, at 7:35 in the morning, each carrying 700-pound bombs and a fuel tank in the bomb bay, bound for the airfield at Kendari. A dozen Japanese fighters trailed them to the target and then came at them from the front, where a B-17E of that year had the least defensive firepower. Two bombers went down.
A third took a hit in the bomb bay tank, which caught fire, and the crew let it go. The rest aborted the mission and put down at the nearest field that would take them, Yogyakarta. Fighters of the Tainan Air Group found them parked there and shot them up on the ground. One of those aircraft, B-17E serial 41-2471, had been handed over to the Army at McChord Field only seven weeks earlier.
Her own side then stripped the nose section, two engines, and two propellers off her for spare parts and left the rest sitting on the grass. The last American bomber got out of Java in March. The Dutch wrecked what remained of the airfields behind them, and the Allied commanders signed the surrender at Bandung on March 12. On the airfields the Japanese now held lay the remains of 15 B-17Es.
Torao Saito, an aviation editor for the Asahi Press who toured the captured bases, counted four E-models being made ready to fly—one at Malang, one at Cirebon, and two at Bandung. Another repairable airframe at a field nearby carried the name “Putsy. ” Giken technicians had airframes but no airplanes, so captured Dutch and local mechanics were put to work alongside them, unsnarling systems nobody on that side had ever had reason to understand. Saito judged the American maintenance setup at Bandung the best on the island, and it became the place where the Giken ran its first flight testing.
Where serial 41-2471 came from is not settled, and nobody has reconciled the two accounts. Robert Mikesh, later the senior curator at the National Air and Space Museum, puts her at Malang Airfield in good condition apart from a missing nose and missing engines. The serial histories that followed, by William Bartsch, put her at Yogyakarta instead, written off on the field where she landed on February 8 after the Kendari mission, among the airplanes the Tainan pilots had already worked over. Both accounts name the same missing parts.
While the work went on, Japanese engineers argued about the airplane in public. Hisanojo Uzawa of Mitsubishi, who had designed the K-21 and the K-67, wrote in Koku Asahi in May 1942 that the newest Fortress amounted to merely modifications made to an old design that first flew in 1935, while admitting in the same piece that the design underneath it was a very good one. Shizuo Kikuhara, who had designed Kawanishi’s big four-engine flying boat, went the other way and wrote with admiration about the cockpit, how few controls it took to fly something that size, and what that freed a pilot to attend to while being shot at. Two Fortresses came out of that patient salvage, rebuilt and flight-tested in Java before the end of 1942.
Getting them home took planning of a kind nobody in that army had needed before. Routes were drawn through Singapore and the Philippines. Fighters flew alongside them, a Japanese-built version of the DC-3 went along as mother ship, and the air defense system was alerted along the whole route so the Empire’s own gunners would not shoot down the prize on its way to Tokyo. They landed at Tachikawa in May or June of 1943, and technicians were over them at once.
The evaluation was headed by Major Kazuyuki Saito, an aeronautical engineer, working with Lieutenant Shuyama, two junior lieutenants named Kurasu and Ohara, and an assistant engineer named Shimamura. Major Uno and Captain Yagi did the flying. An Army engineer named Kato took the engines down component by component. Whole systems went out to specialists elsewhere for study.
One of the E-models became a flying test bench for a captured Norden bombsight, coupled to the Sperry automatic flight control system that flew the airplane during the bombing run. The gunnery fit interested them at least as much, and the automatic computing gunsight most of all. They had the instruction manuals as well. The Japanese public saw the results.
In February 1943, the Asahi Press published a book on the anatomy of enemy aircraft, thick with photographs of captured Allied airplanes and thickest on the Fortress. That same month, a Japanese aviation magazine gave over most of an issue to the captured bombers, with every major part laid out in photographs and drawings. None of that required an airplane that could fly. Dimensions, alloys, rivet patterns, and gun mountings can all be taken off a wreck sitting on jacks.
Behavior cannot. Where a fighter has to sit to stay outside defensive fire, how many seconds a Ki-43 pilot gets in the firing pass before he has to break away, how hard the bomber is climbing away from an interception—all of that has to be watched in the air by somebody who can then tell a squadron about it. That was the other half of the program, and it needed all three airplanes airworthy and a steady supply of parts for machines no factory in the Empire supported. The same work was running in the opposite direction.
Allied technical air intelligence had been filing systematic crashed enemy aircraft reports since April 1943, and by early 1944 had concluded the small manufacturers’ plates riveted inside Japanese airframes were worth collecting on their own account. A unit was stood up for that alone and finished the war holding more than 6,000 of them. The Fortresses were the largest machines in a Japanese collection that also held a Brewster Buffalo, a Hawker Hurricane, a line of captured P-40s wearing the Tachikawa Eagle on their fins, and a Martin Mariner. A few of the Warhawks were flown in combat with the 50th Sentai over Rangoon in 1943.
Nothing in the surviving accounts has the Fortresses flying a combat mission for anyone. In the autumn of 1943, all three of them moved to Fussa, a few miles from Tachikawa, where the Army did most of its flight testing. Two of them flew down to Hamamatsu, the main heavy bomber base, in June or July of 1944. What they were wanted for there is not written down in anything that survived.
The rest of the time they worked as film props. Japanese Army training films used them as the enemy, and at least one of those films showed fighter pilots how to bring a Ki-43 in against a Flying Fortress. The films played in public theaters, and for the cameras, the D-model had its American markings painted back on, with a captured P-40 alongside it in some of them. One Fortress was lost in an accident.
Early in 1944, one of the E-models began its takeoff run at Fussa. A gust came across the runway before the airplane had the speed for its rudder to hold it, and it ran off the paving and put its right wing into a parked Ki-49. The damage finished that airframe, and the program was down to two. The flight testing overlapped the war’s end, and the final fate of the surviving Fortresses is murky.
On October 23, 1945, in Tokyo, a staff officer of the 10th Flying Division sat down opposite an American naval captain named Heting. Major Toga had been on that staff since July 1944, and the division was the whole of the Japanese Army’s fighter defense of the Tokyo area, Tachikawa and Fussa included. Its establishment called for 210 fighters and 12 observation aircraft. It fought along three lines, an outer one anchored on Choshi and Katsura, a second at Kisarazu and Chiba, and the last of them over the capital itself.
In February 1945, Toga had about 200 fighters that could fly. By July, he had 100. There were pilots enough for the airplanes he had, and so little fuel for training that the pilots were getting worse at the work. Heting asked whether the division had kept files on its losses.
Yes, but they have been destroyed, Toga said. The order for that had come from Lieutenant General Kawabe. Late in May 1945, an American photo reconnaissance airplane went over Tachikawa. On the field was a four-engine machine parked there that the interpreters could not name.
It held their attention because Tachikawa was the home of Japan’s Army Aviation Technical Research Institute, and a new four-engine bomber or transport was worth knowing about. They measured the span at about 104 feet and filed the airplane under a code name built out of the measurement: Tachiag Field 104. Weeks went by and nothing further came in. A B-17 spans 103 feet 10 inches.
The measurement was off by two inches. Nobody who saw those prints suspected. The airplane stayed on the books as an unidentified Japanese type until after the surrender. More photographs came in once the fighting stopped.
A Fortress sat at Irumagawa north of Fussa, parked off the flight line at a spot students of the Army’s Air Academy could walk to, and it may have been the machine wrecked at Fussa. A second showed at Tachikawa. Nine days later, the survey flight went over again, and it had gone. The third was never located.
It may have burned in a hangar fire at Tokorozawa, where the Japanese had been gathering airplanes for a museum, a German Stuka among them. Of the thousands of aircraft photographs taken during the occupation, none is known to show any of the three. The Strategic Bombing Survey’s files note that the Japanese had evaluated B-17s. They do not note anyone finding one.
Three years of access changed nothing about how Japan built four-engine bombers. The Nakajima G8N, the Navy’s heavy bomber that reached production at the end of the war, was more advanced in its structure than the Fortress and owed nothing to its airframe engineering. Smaller things did carry a trace. The Navy built a fire control system of the kind the Fortress carried for the G8N.
An automatic computing gunsight for fighters came out of the same study. Taking the Norden apart, both services concluded that their own bombsights stabilized better than the American one, then combined features of the two and ended with an instrument better than either. Then there was the turbo-supercharger, which the Giken had held in its hands in early 1942 off the burned airplanes at Clark, before it had a flying Fortress at all. The Japanese aircraft that most needed a working one—the Tachikawa Ki-74, the Mitsubishi Ki-67 and Ki-83, the Nakajima Ki-87, the Mitsubishi Ki-79, the Navy’s J1N and J2M—all fell short of the performance they had been drawn for, and engine output at altitude is the reason.
Robert Mikesh’s judgment after decades on Japanese aviation is that the engineering knowledge to build an efficient turbocharged engine was there, and the materials were not. The airfields themselves passed to American hands. Tama Army Airfield had been the Japanese Army’s flight test field since the spring of 1940, and the people living around it called it Fussa. On September 4, 1945, a detachment of the United States Army’s 1st Cavalry Division drove onto it.
The buildings were still standing, and there were Japanese aircraft still in the hangars. The Americans called the place Fussa Army Airfield, renamed it Yokota at the end of that month, and it is Yokota Air Base today. Irumagawa became Johnson Air Base, and Tachikawa became an American field as well. American airplanes have flown off that concrete ever since.
No Fortress was on any of the three fields when they got there.