I think the claim is that surviving texts about training archers describe shooting an area is a certain distance and loosing two more before the first hits the ground (with some extra requirement that the first not be aimed at some high angle). You can then go back from the kinds of bows they had and turn that into an estimate.