I was very interested in Usain Bolt's latest 100m world record of 9.69 seconds in the 2008 Chinese olympics. I especially wondered how much faster he could go, given his apparent slow down and showboating at the end of the race. Here are his time splits every 10m compared to the fastest ever:

distance (m)	bolt (s)	best of the rest (s)
10	1.85	1.77
20	1.02	1.00
30	0.91	0.89
40	0.87	0.86
50	0.85	0.84
60	0.82	0.82
70	0.82	0.83
80	0.83	0.83
90	0.84	0.85
100	0.88	0.85

As you can see he starts slower than most due to his height (like Carl Lewis) but only starts slowing down after around 75m, whereas the rest generally start to slow after 60m. But you can see that Bolt's 0.88s for the last 10m is too long, and that he lost about 0.05s due to his celebrations. I estimate he could have finished in 9.64s at least, which is an amazing improvement over his own previous best of 9.72s. Also there looks to be room for improvement in his start time, and interpolating from the best of the rest here suggests that perhaps 9.58s is achievable? Of course it is more lucrative to break the world record multiple times which may have been in the back of his mind?

I've previously wondered about having a speed as opposed to a distance race. I.E. whoever reached the highest speed over any 1m of the 100m would be the winner. I'm not sure about the logistics of that, but you can see that Bolt would fair well, as he recorded 2 x 10m splits of 0.82s, whereas others were only able to maintain this speed for 10m. A quick conversion with the unix command units -t '10m/0.82s' 'miles/hour' shows that this top speed is around 27.28 mph (43.90 kph), and Bolt averaged around 23 mph (37.15 kph) over his whole record breaking 100m.