If you have ever run a marathon then you will know that the best finish is not achieved by running as fast as possible at all times until the distance is complete. In fact, it is quite the opposite. The human body is at its most efficient when it is using its resources evenly over time.

So, it is a well understood edict of distance running that the fastest times are set when the pace is even. The problem over a two-hour marathon is that this is incredibly difficult to do. It’s hard enough when the only goal is a good performance, but...

If human runners can’t maintain the speed to the required accuracy, then getting a car to do it feels like the obvious solution — every modern car has cruise control after all. Unfortunately, the versions in production cars are nowhere near accurate enough.

Very few cars have an accelerator resolution that can give you better than 0.1KPH accuracy. With that resolution as the limit of the hardware, if you extrapolate that over the course of a 42KM race that ends up being seconds of time that is left unaccounted for.

If the pace car had run 0.1KPH too slow for the two hours, it would have taken two hours and 34.3secs to run the marathon distance. It was a big enough error to have derailed the entire Challenge. A one second error over the two hours is just 0.0139% of the final time. This translates to running the pace car at a speed of 21.0975kph plus or minus 0.00293kph.

That meant that the team had to build a cruise control with better than 30 times the precision of the best that was currently available.

It might seem that this should be enough to deliver the +/-1sec result that the team required, but RML were not taking any chances.

“The next level of complication was that all of that distance was based around the distance measurement from the vehicle”, explained Peter Vint, “which wasn't necessarily the distance along the course. We were measuring how far the car had travelled, but if the course that the car was travelling wasn't perfectly straight — because the car meandered slightly along the course or it took a slightly wider radius around some of the bends — then the distance the car had travelled would be different to the distance along the course.”

It might seem that these multiple levels of control would be enough, but Chris Francis and his RML colleagues still were not finished. They also wanted to build in sufficient levels of redundancy to guarantee that it all worked on the day.

There were two speed sensors and a lot of cross checking to make sure everything was working okay. We also had two identical cars equipped with the same software, just in case there was a problem with one.

“We were running both of the cars on the event. One was further up the road than the other, so a few tens of meters ahead. In the event of a problem with the primary car we could have swapped the cars, or swapped the spare car into position... It would have been done gradually over a minute or two to back into position and once in position we would then have turned the lasers on to signify we were ready.”

The result was an impressive level of accuracy — five times better than that demanded by the INEOS 1:59 Performance Team. A remarkable piece of technology that worked perfectly on the day, enabling split times of incredible consistency as Kipchoge churned out one 2min 50sec kilometre after another, until — confident of his place in history — Eliud Kipchoge waved the car and the pace-makers aside to run the final 500 metres alone.