Science researchers have been concerned to understand the ways in which science learning and social relationships are connected. Networks of social interactions have preciously been used to get snapshot pictures of broad aspects of student interactions. Here, I include (1) multiple interaction categories and (2) student self reports of these multiple interaction categories over a period of a semester to create sequences of different types of specific academic and social networks. The cohort is primarily first year physics majors from a Danish university. Once established, a cluster algorithm splits each network into groups of students. The stability across categories and time of groups as well as the attributes of students in each group characterize the cohort as a set of communities of practice. Seen in this way, the networks investigate the dynamics of quantifiable aspects of communities of physics learners in a particular setting.