Thank you all for your informative responses!
While examining the effects of unusual beam profiles on data
collection due to capillary optics, I had collected a wedge of data
on a large, high-quality lysozyme crystal at 8 different sample to
detector distances. I restricted the analysis of all data to the
resolution range seen at the largest distance, so each data set
contained nearly the same reflections with the same redundancy. <I/
sig> does indeed improve with distance: starting at 7.89 at 135 mm
and reaching its maximum of 10.35 at 614 mm. The total linear R
factor for that range also falls from 0.09 down to 0.059. So, even
with a 1.8 mrad divergent beam (about as divergent as practical at a
beamline), the effect is significant.
Surely someone has done this experiment before long ago.
I like Jim Plufgrath's way of looking at the problem: reflections and
background scatter have different effective source sizes and
distances. For a reflection, the source is the same as the beamline
source (some place way upstream). For solvent the source is the
sample itself, for air, the source is spread out over a range of
distances ... thus background is very divergent.
Richard
