“…one thus takes into account the actual average of the calculated intensity for the atoms whose positions are supposed known, and the average to be expected on statistical grounds for the atoms whose positions are unknown. The procedure of Parthasarathi & Parthasarathy (1975), following Srinivasan & Ramachandran (1965), goes some way towards this, but is not identical with it. Refining the scaling factor by least squares would seem to be unjustifiable, since the value obtained is known to be biased by 'defects in the model', which missing or misplaced atoms certainly are.…”