It turned out that the ontology that is used in materials science is not logically consistent for historical reasons. The class CERAMICS is defined as ``all non--metallic, inorganic materials''. However, if we look at the hierarchy in figure 5 we find the class ORGANIC GLASS as a specific ceramics whose materials definitely consist of organic elements. This situations happens since organic glasses were developed rather late and show the physical behavior of glass rather than that of polymers. In order to be consistent with standard textbooks in this field and to develop a structure that is acceptable for engineers we adopted the classical domain classification. This is uncritical since we do not use a relation that defines chemical composition as in [vdVM91]. Otherwise, either the usage of default definitions or a more elaborated, logical sound classification has to be considered. This leads directly to the question of an appropriate expressive power of a representation language for ontologies.