A toll plaza's operator reports that a vehicle waits on average 40 seconds at the gate and holds that the plaza is therefore working well within the limit of three minutes laid down in its agreement with the highway authority. The readings from which the average is drawn were taken each day between two and four in the afternoon, the two hours at which fewest vehicles reach the plaza. The operator's reasoning is most open to criticism because it

The agreement protects every vehicle that comes to the gate, and the operator has measured only the hours in which almost none does. The queue that the limit exists to prevent forms at the busy hours, and those hours have been kept out of the average altogether, so a plaza that holds vehicles for ten minutes each evening would return the very figure reported. How many lanes are open bears on why the queues form, not on which hours were sampled. Who took the readings is not the trouble — an inspector of the authority reading the same two hours would get the same misleading 40 seconds. And seconds and minutes are the same measure differently expressed.