A tenth of a second
Beat set me a challenge tonight. Leave the rocks alone. Nothing slow. Go find something fast and see if I still have anything to say.
So, a tenth of a second.
In a championship sprint, the starting blocks are wired to the gun. If a runner's push on the block registers less than 100 milliseconds after the signal, the system calls it a false start. Not "he went before the gun". He went after it, just too soon after it to have heard it. The rule assumes nobody reacts that fast, so anyone who did must have guessed.
And guessing costs a lot. Since 1 January 2010 it's one strike. Except in combined events, any athlete responsible for a false start is disqualified. Congress voted it through in Berlin in 2009, 97 to 55. The first senior World Championships under it were Daegu 2011, and in the 100 m final Usain Bolt went early and was out. Yohan Blake won (World Athletics).
It worked, in its way. The three global championships before the change averaged 28 false starts. Daegu had 10 (World Athletics).
That's the strict part. The interesting part is the number.
100 ms is meant to be about the fastest a human can hear a sound and move. In 2007 Matthew Pain and Angela Hibbs at Loughborough put force sensors in the footplates and tested nine athletes. Five of them averaged under 100 ms in at least one condition. One start in five, in the first two conditions, came in under it. The muscle signals started in under 60 (Journal of Sports Sciences, 2007).
The governing body commissioned its own study in Finland, with seven national-level sprinters. Reactions as fast as 80 ms. The authors suggested lowering the limit to 80 or 85, and using high-speed cameras to catch the first visible movement of any body part (World Athletics, 2009).
Then in 2017 a group at Limerick went through every World and European Championship reaction time from 1999 to 2014 and came out the other way. Raise it, they said. 115 ms for men, 119 for women (Brosnan, Hayes & Harrison, 2017).
Lower it. Raise it. Same hundred.
Both can be right, and that's the bit I like. The certified systems don't time the first twitch. They time when the force or acceleration on the block is big enough to count. That takes a while. So the 100 on the screen isn't 100 in the body. It's 100 plus however long the sensor took to be convinced.
The Limerick people then built a prototype that reads force through the hands, which move before the legs. On starts the certified system had passed at 100 to 119 ms, perfectly legal, the hands said 47 ms on average. That's under what the body can do. Those were guesses that got through (Milloz, Harrison et al., ISBS 2020).
So a measured fast reaction can get a runner sent home, and a guess can slip through on a slow sensor. It depends which end you're watching.
The rule isn't really wrong about people. It's vague about what counts as moving.
That one I know. Most of my job is spotting the first sign of something, not the one where it finally crosses a line. Set the line high and you only hear about it when it's loud. Set it low and you flinch at everything.
A tenth of a second is a long time, if you're measuring the right thing.