This works applies the economic theory of communities of practice to the area of science and scientific education. A document-centered view on COPs is taken, i.e. it is assumed that scientific practice is inscribed in the content, structure, and appearance of documents. However, scientific practice is also encoded in the meta view on these documents, i.e. the community's rating, discussions, and reviews of document. It is claimed that explicating and comparing both types of practice allows for identifying various scientific communities and sub-communities. In order to test these intuitions, the collaborative and interactive reader panta rhei is implemented. panta rhei supports its users to easily find, discuss, and rate information. The system infers the users' virtual practice from their ratings, annotations, and browsing behavior in the system. Moreover, panta rhei analysis the users' collection of imported scientific document in order to extract their inscribed practice. Based on the comparison of practices, panta rhei makes predictions on the users' COP membership, i.e. their social networks and implicit relations among each other. In addition, panta rhei utilizes community information to analyse the relevance as well as the user's preferred content, structure, and appearance of scientific documents. This paper focuses on the discussion of scientific COP from a document-centered perspective. It further introduces the current features of panta rhei's educational release and the system's extension towards a community tool.