On the relation between the velocity of the gradient wind and that of the observed wind