Designing for a Person Who Does Not Exist

In the late 1940s an air force investigating an unusual number of crashes measured several thousand pilots on ten physical dimensions and calculated the average of each. The cockpits had been built to those averages for two decades. A young researcher then asked how many of the pilots fell within the middle range on all ten measures at once. The answer was none.

[A] The result is less surprising than it sounds. Being close to average on one dimension tells you very little about the next, so the proportion who are average on all of them shrinks with every measure added, and ten is more than enough to reduce it to zero in any real population. [B] The consequence for design, however, was considerable, because a cockpit fitted to the average man fits nobody in particular and fits some people dangerously badly. [C] The solution adopted was not a better average but adjustment: seats, pedals and straps that move. [D] It is worth noticing how ordinary that idea now seems and how strongly it was resisted at the time, on the grounds of cost and of engineering complexity. Both objections were correct, and both were beside the point.