This article reflects on the power and dangers of diagrams as a mode of anthropological exposition, comparing this particular form of non-text to the brief dalliance of mid-century anthropology with algebraic and logical formulae. It has been claimed that diagrams, like formulae, are clearer, simpler, or less deceptive than textual argument. By contrast, this article argues that diagrams are just as slippery and tricky as words, but that images and words slip and slide in different ways. Holding both diagrams and words together when building an argument enables not only a specific kind of rigor, but also moments of unexpected theoretical invention. This technique of holding together contrasting heuristics scales up as a productive epistemic device for anthropology more broadly.