The data set was sorted with l varying most
                                  rapidly and h varying least rapidly. Every
                                  10th reflection in this sorted list was
                                  excluded from refinement and included in the
                                  calculation of a 'free' R factor.