Van der Meulen's commentary on our article is not convincing. Some of our statements and points of view are not fully understood; others which we never made o r held are nevertheless disputed zealously. Moreover, Van der Meulen's argument shows internal inconsistencies.Our study fits into the tradition of Ullman, Sorre, Van Paassen, Nystuen, Lukermann, Berry, Papageorgiou, Smirnov and Wirth, who have sought for the central concepts of the discipline. They usually, but not always, got their results in a subjective, philosophical way. We wished to explore the possibilities of a more objective, formal method, namely graph theory, to analyse the conceptual construction of complexes of knowledge. This was the central question, which Van der Meulen did not address. At the same time, we made conclusions about, for example, the position of spatial concepts with regard to temporal concepts. Van der Meulen is right when he states that no normative, valuegiving conclusions were madebut that was not our aim.We d o not share the opinion that our contribution must be classified as linguistic and not as t heoret ical-geograp hical. Theoretical geography does not make statements on the socio-spatial structure, but on the object and method of geography (the philosophy of geography) and also on the assumptions (e.g. images of man) on which geographic research is based. For that purpose studies are made of written texts; they do not thereby become linguistic studies. Van der Meulen advocates putting problems and not concepts into a central position in geography. In our opinion they are two halves of the same whole. Without concepts no problem could ever be experienced or formulated. On the other hand problems lead to new concepts. Despite this argument Van der Meulen is not insensitive to language and concept. Science is for him a language game. Wittgenstein's concept 'language game' points t o a linguistic means (e.g. a concept) by which reality is disclosed. Concepts would then get their meaning in the context in which they are used. This thinking in terms of contexts has its advantages. It bypasses the 370 dangers of essentialism or conceptual realism and nominalism. It has, however, its limitations because of its relativism: a language game is uncircumscribed and has blurred edges (Wittgenstein 1968, pp. 70-7 I). Scientific language would then not be distinguished from natural language. This explains Van der Meulen's criticism of "many social scientists" who begin their analyses with a number of definitions. Van der Meulen's conception of science as a language game is not consistent with his nominalistic point of view that definitions are only agreements in which the meanings of words are determined. Linguistic conventions and social customs have not usually been thought out in order to be carried out, but have developed in a relatively independent way, from the intentions and acts of individuals in communication with each other (Blok 1976, pp. 23,67). Moreover, there is acorrespondence-element present in definitions : definiti...
