We find that a certain class of geometric models of crystal growth predicts the formation of an expanding facet with a discontinuous edge adjacent to rough orientations-a shocking facet. However, we show that although such structures are observed experimentally, aspects of their long-time evolution are not entirely captured by the same class of geometric models that predict their formation. We suggest a simplified description of interfacial evolution that reproduces the experimentally observed behavior. (C) 2002 Elsevier Science B.V. All rights reserved.