Slide 14 of 27
Mosflm uses the technique of post-refinement to determine accurate cell parameters before the full integration, so the unit cell dimensions should not be refined during integration (as is the case with some other integration programs!). Post-refinement uses the phi centroids of reflections which are spread across adjacent images to refine the cell parameters, crystal mis-setting angles and mosaic spread independently of camera parameters (e.g. crystal to detector distance, yscale and tilt and twist of the detector).
Therefore post-refinement requires at least two adjacent images (if the average mosaicity = twice the oscillation angle for each image) with a sufficiently large number of completely recorded partials for good statistics. If the mosaicity is larger, then more adjacent images are needed.
For crystal systems with orthorhombic or lower symmetry, two segments of data separated by ~90º should be used.