On the semiclassical interpretation of heavy ion reactions