Mid point vector

A constructed mid-point cannot be used for best-fit vector alignments because the vector is wrong, I normally get around it by constructing a generic point and calculate the average vector of the 2 points...

PNT_MID    =GENERIC/POINT,DEPENDENT,CARTESIAN,$
            NOM/XYZ,<(PNT_1.TX+PNT_2.TX)/2,(PNT_1.TY+PNT_2.TY)/2,(PNT_1.TZ+PNT_2.TZ)/2>,$
            MEAS/XYZ,<(PNT_1.X+PNT_2.X)/2,(PNT_1.Y+PNT_2.Y)/2,(PNT_1.Z+PNT_2.Z)/2>,$
            NOM/IJK,<(PNT_1.TI+PNT_2.TI)/2,(PNT_1.TJ+PNT_2.TJ)/2,(PNT_1.TK+PNT_2.TK)/2>,$
            MEAS/IJK,<(PNT_1.I+PNT_2.I)/2,(PNT_1.J+PNT_2.J)/2,(PNT_1.K+PNT_2.K)/2>

Would be useful if the mid-point worked in the same way as my genetic point.