Clashing cosmologies: Contrasting knowledge in the Greenlandic fishery

Current worldwide discussions about management regimes for natural resources are increasingly focusing on forms of knowledge: for example on the discussion on scientific knowledge versus ‘other’ forms of knowledge (Inglis 1993; Scott 1996; Kalland this issue), be it ‘local’, ‘indigenous’, aboriginal, or users’ knowledge. Many of these discussions have been highly politicised interactions between scientific specialists, administrators and locals. Knowledge has, in other words, become located at the ‘inter-face’ between the different actors (‘faces’) in the discourse (Roepstorff 2000).When ‘knowledge’ becomes more of a battleground than a field of mutual exchange, the pieces of knowledge exchanged may appear as politicised free-floating signifiers (see e.g.Agrawal 1995).This veils the fact that the pieces of knowledge discussed are normally derived from another interface: the interface between certain persons and groups and their environment, and therefore that there are important differences in what counts as knowledge and how this is constructed. I have previously suggested (Roepstorff 2000) that a study of the ‘who, what and how’ of knowledge may be a first step in understanding how conflicts over knowledge might conceal different conceptualisations of the very elements discussed. This chapter attempts to take this argument one step further. Based on a study of the fishery of Greenland halibut in Greenland, it demonstrates that different cosmologies are connected with the different forms of knowledge about the fish. Cosmology is here conceived as the conceptualisation of what is out there (an ontology), a method to validate and examine it (an epistemology) and a prescription for how people should ideally relate to it (an ethics). The argument builds on case material gathered in Greenland during several fieldwork periods between 1996 and 1998, but its structure is not directly derived from the field. It is inspired by an analysis of a Greenlandic version of a common