In order to ask what work the elusive concept of ‘character’ might do for anthropology, this article first ask what work the concept does for Euroamerican epistemology more broadly. It examines two invocations of ‘character’ in relation to animals at a scientific research site in South Africa. The first is the commonplace use of the term to denote the way the research subjects have been made into ‘characters’ on the tv show ‘Meerkat Manor’. The second is the technical term ‘biological character’ – the basic unit of contemporary evolutionary biology, and the main object of study at the site. These two characters are more than mere homonyms – they hark back to related concerns about purposive action, they populate conflicting moral narratives, and they operate on the threshold between self-conscious fiction and essential truth. Building on this case, I argue that the distinctive value of the concept of character for anthropology resides in its ambivalence – the way it can point both to a contrived mask (a character in an account) and to the very essence of the entity in question (its true character). Such ambivalence maps a particular social form, which echoes across the anthropology of institutions, of ethics and of knowledge.