Over the past decades, education researchers have shifted their understanding of science from "a rhetoric of conclusions" – that is, a fixed canon of content – to a social process of knowledge construction.  While much of the research has investigated individual learners as they engage with scientific ideas, experiments, and methods, increasingly researchers are turning to the social processes of science as it is constructed in a community, with particular interest in scientific argumentation.  This emphasis on argument recasts the role of evidence and data in scientific classrooms: rather than being used to demonstrate the scientific canon or even to guide students to construct correct scientific principles, it is the grounds on which claims – generated by students in the process of argumentation – are warranted.  In this paper, I explore a transcript of scientific discourse, exploring the rules by which participants in the discourse endorse or reject scientific claims.  I appeal for a more nuanced understanding of evidence as one of many criteria by which scientific claims are evaluated, and that evidence, at times, is incommensurable with other, possibly more scientific, criteria for evaluating claims.  This view of argumentation, and the peculiar discourse games associated with argumentation, is particularly relevant for understanding difficulties that diverse student populations may face.