The minor patch below clarifies what the computation is about and
removes the assumption that point-min == 1, while arguably
making the the code ever so slightly more efficient.


        Stefan