91av

WE’VE reached the limit for some sports. No matter how hard we train or how much the design of running shoes improves, we’re just not going to get any better.

A mathematical analysis shows that most of our track and field records are being broken by chance rather than by better equipment or increased skill. The researchers looked at the best annual performances in 22 events in German athletics championships during the 1980s and 90s, including long jump, high jump and running races. Only four events showed any kind of systematic increase over that time, while the rest had hit a brick wall, say Daniel Gembris from the Jülich Research Centre in Germany and his colleagues. The same proved true for most international events.

The record-breaking times fitted the statistical distribution you’d expect if the overall level of performance had stayed the same, they say. If you were to plot all the possible times for a certain race, explains team member John Taylor of King’s College London, you’d get a bell-curve. There would be lots of people capable of finishing it in, say, 10 minutes. Fewer people could finish in under 9 minutes, and maybe only one or two in 8 minutes. But given time, without any increase in average skill, eventually you would stumble upon those people who could cut the tape in 8 minutes flat. Their existence isn’t miraculous—they’re just rare.

Some events did show a leap in performance, like the 10,000 metres. Its record times followed the predictions from 1980 to 1992 and then plummeted over six consecutive races up to 1998, knocking almost a minute off the 27-minute record, instead of the predicted second or two. The researchers have no idea what caused these jumps in performance. “Is it a change in the rules? We don’t know,” says Taylor.