A noncrystallographic screw axis parallel to a twin axis can corrupt intensity statistics

A number of methods to detect twinning are based upon the assumption that the statistical properties of diffracted intensities are different for twinned and untwinned specimens. This may not be true for a large portion of the reflections in a twinned specimen if a noncrystallographic screw axis parallel to the twinning axis is present. In this case, up to half of all reflections can obey Wilson statistics, which are typical of untwinned crystals. The distribution corresponding to a whole set of observed intensities is biased towards the Wilson distribution in this case.