Observations of systematic variation in the shapes of protein crystals have unique potential to report chemical effects on protein-protein interactions, because diffraction can be used to image the detailed structure that links an easily controlled "cause" (change in chemical conditions) with an easily observed "effect" (change in crystal shape). Crystal shape correlates directly with the relative growth rates of the various faces. By studying cases of known structures in which crystal shape varies systematically with some solution parameter (pH, ionic strength, concentration of a specific solute, etc.), it is sometimes possible to correlate that parameter with its effect on a particular crystal contact. Such correlations provide the basis for models of how solution parameters affect protein-protein interactions. We have applied this technique in a model system involving a subtilisin BPN' variant. Crystal shape varies systematically with pH and ionic strength, an effect that has been traced to a specific interaction in one crystal contact. Mutation of an Asp sidechain in this contact (to Asn) eliminates the morphology variation effect, confirming the key role of this residue in the observed shape effect, and enabling the mechanism to be placed in a mathematical framework.