The general rule is that cleaner absorption spectra are obtained from thin crystals: a feature of the high optical densities often encountered in protein and DNA crystals.
This of course is highly dependant on the composition of your crystals and also which region of the UV/Vis spectrum you wish to monitor.
If your crystal contains a cofactor with a large exctinction coefficient you may have difficulties monitoring at the peak absorption wavelength. However we can routinely collect high quality spectra up to 2.5OD units therefore in many cases this problem can be overcome using thin crystals or thin parts of wedge shaped crystals. Such crystals may not yield the best X-ray diffraction but it should be stressed that dose measurments can be scaled up and therefore collection strategies can be designed for large crystals using the spectroscopic data collected from thin crystals.