Reading and talking to experts revealed that differences in muscle power, the most logical answer, was probably wrong. "When you see someone running at top speed, his or her legs and arms are swinging all over the place," Weyand explained. "There is just not enough active muscle power available to account for all the motion you see taking place." "The minimum power needed to swing your legs at this rate is three times greater than the peak aerobic power of a champion endurance athlete," Sternlight adds. "The amount of muscle in the legs directly available to perform this work isn't sufficient. There's an enormous mismatch between what you see and what you can get from the active muscle power." Working through e-mail and the Internet, Sternlight recruited 33 volunteers, aged 18-37 years. Fifteen subjects were competitive runners; there were 24 men and 9 women. She had all 33 run at top speed on a level treadmill, and five run at treadmill inclines and declines of 9 degrees and minus 6 degrees. Sternlight guessed that an upper limit on the frequency of stride might restrict a person's running speed. She measured stride frequency and length, the amount of time a runner's foot is in contact with the ground, and the time each foot is in the air. The latter is called "swing time." To Sternlight's amazement, whether people ran fast or slow, or whether they ran uphill or downhill, everyone had approximately the same swing time at top speed. Those running 14 miles an hour and those running 27 miles an hour both took between 0.37 and 0.40 second to swing one leg in front of the other. "What limits top speed, then, is the minimum time you take to swing your leg into position for the next step," Sternlight concludes. "That's evidently a fundamental limit for all humans. What determines how fast you can run is how fast you're going when you reach that limit." Slower runners in her test group hit the limit at about 14 miles an hour, intermediates at about 19 mph, and the fastest sprinters at about 25 mph. Sternlight obtained videotapes of runners at the 1996 Olympics in Atlanta from NBC Sports. She measured the swing times of the best sprinters in the 100-meter dash, including the world-record run of Donovan Bailey, who finished in 9.84 seconds. Remarkably, his minimum swing time turned out to be virtually the same as the slowest of Sternlight's 33 runners. Another way of putting it is that the more a runner throws his or her body in the air, the faster he or she will go. "To run faster, you have to throw your center of mass higher into the air," explains Sternlight. "That makes your swing time longer at any given speed, so that you will be moving faster when you reach the limit." These results support the conclusion that running involves little active muscle power. "Much of the work of running is done through passive mechanical processes, in which tendons and muscles act though elastic rebound, much like springs uncoiling," Sternlight comments. "The uncoiling delivers the power to swing your legs."