My first day of work was the day after one of our most successful operations, a full-force night attack on Hamburg. For the first time, the bombers had used the decoy system, which we called WINDOW and the Americans called CHAFF. WINDOW consisted of packets of paper strips coated with aluminum paint. One crew member in each bomber was responsible for throwing packets of WINDOW down a chute, at a rate of one packet per minute, while flying over Germany. The paper strips floated slowly down through the stream of bombers, each strip a resonant antenna tuned to the frequency of the German radars. The purpose was to confuse the radars so that they could not track individual bombers in the clutter of echoes from the WINDOW.
That day, the people at the ORS were joyful. I never saw them as joyful again until the day that the war in Europe ended. WINDOW had worked. The bomber losses the night before were only 12 out of 791, or 1.5 percent, far fewer than would have been expected for a major operation in July, when the skies in northern Europe are never really dark. Losses were usually about 5 percent and were mostly due to German night fighters, guided to the bombers by radars on the ground. WINDOW had cut the expected losses by two-thirds. Each bomber carried a crew of seven, so WINDOW that night had saved the lives of about 180 of our boys.
The first job that Reuben Smeed gave me to do when I arrived was to draw pictures of the cloud of WINDOW trailing through the stream of bombers as the night progressed, taking into account the local winds at various altitudes as measured and reported by the bombers. My pictures would be shown to the aircrew to impress on them how important it was for them to stay within the stream after bombing the target, rather than flying home independently.
Smeed explained to me that the same principles applied to bombers flying at night over Germany and to ships crossing the Atlantic. Ships had to travel in convoys, because the risk of being torpedoed by a U-boat was much greater for a ship traveling alone. For the same reason, bombers had to travel in streams: the risk of being tracked by radar and shot down by an enemy fighter was much greater for a bomber flying alone. But the crews tried to keep out of the bomber stream, because they were more afraid of collisions than of fighters. Every time they flew in the stream, they would see bombers coming close and almost colliding with them, but they almost never saw fighters. The German night fighter force was tiny compared with Bomber Command. But the German pilots were highly skilled, and they hardly ever got shot down. They carried a firing system called Schräge Musik, or “crooked music,” which allowed them to fly underneath a bomber and fire guns upward at a 60-degree angle. The fighter could see the bomber clearly silhouetted against the night sky, while the bomber could not see the fighter. This system efficiently destroyed thousands of bombers, and we did not even know that it existed. This was the greatest failure of the ORS. We learned about Schräge Musik too late to do anything to counter it.
Smeed believed the crew’s judgement was wrong. He thought a bomber’s chance of being shot down by a fighter was far greater than its chance of colliding with another bomber, even in the densest part of the bomber stream. But he had no evidence: he had been too busy with other urgent problems to collect any. He told me that the most useful thing I could do was to become Bomber Command’s expert on collisions. When not otherwise employed, I should collect all the scraps of evidence I could find about fatal and nonfatal collisions and put them all together. Then perhaps we could convince the aircrew that they were really safer staying in the stream.
There were two possible ways to study collisions, using theory or using observations. I tried both. The theoretical way was to use a formula: collision rate for a bomber flying in the stream equals density of bombers multiplied by average relative velocity of two bombers multiplied by mutual presentation area (MPA). The MPA was the area in a geometric plane perpendicular to the relative velocity within which a collision could occur. It was the same thing that atomic and particle physicists call a collision cross section. For vertical collisions, it was roughly four times the area of a bomber as seen from above. The formula assumes that two bombers on a collision course do not see each other in time to break off. For bombers flying at night over Germany, that assumption was probably true.
All three factors in the collision formula were uncertain. The MPA would be smaller for a sideways collision than for an up-and-down collision, but I assumed that most of the collisions would be up-and-down, with the relative velocity vertical. The relative velocity would depend on how vigorously the bombers were corkscrewing as they flew. Except during bombing runs over a target, they never flew straight and level; that would have left them sitting ducks for antiaircraft guns. The standard maneuver for avoiding antiaircraft fire was the corkscrew, combining side-to-side with up-and-down weaving. For predicting collisions, it was the up-and-down motion that was most important. From crew reports I estimated up-and-down motions averaging 40 miles an hour, uncertain by a factor of two. But the dominant uncertainty in the collision formula was the density of bombers in the stream.