How does diffraction anisotropy stall improvement of the R-factor?
There are two means by which anisotropy impacts the R-factor. The first has to do with the inclusion of numerous poorly measured reflections in anisotropic data sets (i.e. F/sigma less than 3.0). Because they make up a considerable fraction of the data set, the R-factor in anisotropic refinement tends to be high. Inclusion of these weak reflections is a consequence of the shortcomings of currently available data processing programs such as Denzo and Mosflm. If one wishes to include all the reflections in the well diffracting direction, one is forced to also include the reflections in the weak diffracting direction as well. In other words, the resolution cutoff for a data set processed by Denzo or Mosflm is spherical, whereas the intensity of the difraction pattern is ellipsoidal. The solution to this aspect of the problem is to impose an ellipsoidal resolution boundary on the data, rather than a conventional spherical boundary, so that the weak reflections falling outside the ellipsoid
