How Do You Know the Marathon Is Exactly Right? A Bicycle and a Counter

A road is not a line, so the distance depends on where along it a runner travels, and a course measured down the centre is longer than one measured around the inside of every bend. The rule adopted internationally is that the measured line is the shortest route a runner could take without leaving the course, running the tangents through corners and hugging the inside of every curve. That line has to be identified and followed precisely, which is why measurement is done on a bicycle rather than on foot or by satellite, since a bicycle can be steered accurately along a chosen line at a steady speed and carries an instrument that counts wheel revolutions.

The procedure is standardised and exacting:

• •A counter geared to the front wheel records revolutions to a fine resolution
• •The bicycle is calibrated on a steel-taped course of known length before and after
• •Calibration is repeated because tyre pressure and temperature change the wheel
• •The course is ridden at least twice along the shortest legal line
• •The two rides must agree within a small tolerance or the work is repeated
• •Start and finish are marked permanently and the course documented in detail

Measurers add a small margin on purpose, which is the part of the system that surprises people. A factor of one part in a thousand is added to the measured distance, so a marathon course is laid out about forty two metres longer than the nominal distance. The reasoning is that every source of error in practice makes a course shorter than intended, including a rider cutting a corner slightly, a tyre at higher pressure than at calibration, and a course altered on the day by barriers or parked vehicles. Making the course slightly long guarantees that nobody runs less than the stated distance, which protects the validity of every time recorded on it. Running slightly further is accepted as the lesser problem.

Runners routinely finish a certified marathon with a watch reading well over the nominal distance, and the discrepancy has a clear explanation rather than indicating an error in the course. Satellite positioning has an inherent error of several metres in each fix, and a track built from many fixes accumulates a wandering path that is always longer than the true one, since random error adds distance and never removes it. Buildings, trees and tunnels degrade the fixes further. A runner also does not follow the measured line, weaving between other competitors, going wide around corners and crossing to water stations. Those effects add up, and a reading one to two per cent over the stated distance is entirely normal on an accurately measured course.

Accurate measurement is necessary and not sufficient, since courses must also satisfy rules designed to prevent an unfair advantage. The drop in elevation from start to finish is limited to an average of one metre per kilometre, which excludes several fast downhill courses from record eligibility. The separation between start and finish must not exceed half the race distance in a straight line, which prevents a course arranged to run with a prevailing wind. The course must be certified in advance and remeasured after any record performance, and a measurer travels with the race or verifies afterwards. Several apparent records have been refused on these grounds, which causes considerable frustration.

The measured line is the shortest route a runner could legally take, ridden on a bicycle with a revolution counter calibrated on a taped course before and after. A margin of one part in a thousand is added deliberately, so a marathon is about forty two metres long, because every practical error shortens a course. Elevation drop and start-to-finish separation are also limited, and records have been refused for breaching those rules.