Following the first determinations of protein structures in the late 1950s and the early 1960s (see for example Kendrew et al. 1960; Perutz, 1964), the three-dimensional structures of several hundred proteins have been elucidated by X-ray diffraction on single crystals. By the end of 1991, approximately 150 entries of proteins with substantially different sequences and a well resolved structure (Hobohm et al. 1992) were deposited in the Protein Data Bank (Bernstein et al. 1977; Abola et al. 1987). In addition, many structures of homologous proteins or of mutants have been described, bringing the total number of entries to about 600. While it was soon accepted that almost all of these structures do indeed give a correct picture of the fold of the active protein in spite of the non-physiological environment of single crystals, it is not clear to what extent structural details are reliably described by these structures. In particular the surface of a protein may be modified due to the dense packing of protein molecules in the crystal lattice. A detailed knowledge of the protein surface is, however, essential for the understanding of the function of the protein.