The keywords used for the Double Columnar Transposition (DCT) should have relatively prime lengths. But why?