Rietveld refinement of a wrong crystal structure

Rietveld refinements are generally used to confirm crystal structures solved from powder diffraction data. If the Rietveld refinement converges with low R values and with a smooth difference curve, and the structure looks chemically sensible, the resulting structure is generally considered to be close to the correct crystal structure. Here we present a counter example: The Rietveld refinement of the X-ray powder pattern of gamma-quinacridone with the crystal structure of beta-quinacridone gives quite a smooth difference curve; the resulting crystal structure looks reasonable in terms of molecular conformation, molecular packing and intermolecular hydrogen bonds. However, neither the lattice parameters, the molecular packing nor the conformation of the molecules show any similarity with the actual structure, which was determined from single-crystal data. This example shows that a successful Rietveld refinement is not always final proof of the correctness of a crystal structure; in special cases the resulting crystal structure may still be wrong.