The scaling function is constructed so that it can compensate for anisotropic decay in the scattering and the lack of a solvent continuum in the model. (The solvent continuum correction was suggested in Moews and Kretsinger (J. Mol. Biol. (1975). 91, 201-228)). The function is
The parameter B is an isotropic temperature factor correction. is a purely anisotropic temperature factor correction. is the ratio of the average solvent density and the average protein density. is an additional temperature factor used to blur the solvent relative to the protein. In addition there is a scale factor (K), independent of resolution, by which Fo is multiplied, which brings the observations onto an absolute scale. You have the option of either allowing Rfactor to choose a value for each of these parameters, or of setting any or all of the values yourself.
The anisotropic temperature factor ( ) is a symmetric matrix containing 6 unique elements. They are named B11, B22, B33, B12, B13, and B23. Many space groups will restrict the values of these elements. Rfactor takes care of enforcing these restrictions. The matrix is required to be purely anisotropic and this is ensured by the requirement that B11, B22, and B33 sum to zero.