Part 1 of the book explores conceptual issues and problems of method, grounding the discussion firmly in ethnography. Chapter 1 introduces the linked problems of the definition, recognition and reporting of emotion, opting for emotion episodes – closely observed sequences of natural behaviour – over artificially evoked emotions in constrained contexts that miss the psychological and social complexity of emotions, their fuzzy boundaries. I review historical shifts in the significance and salience of emotion, and outline commonalities and differences with historical approaches. In tackling emotions, I espouse an engaged perspective from the field, a pragmatic and eclectic approach tempered by scepticism about common-sense psychological categories. An ethnographically sound account of emotion, I argue, will test the value and robustness of social science concepts and methods, permitting new ways of reporting and analysing emotion. The aim is therefore critical and exploratory. What is wrong with the usual ways of researching and writing about emotion? What might work better? In contrast to anthropological accounts of emotion dependent on discourse, hypothetical examples and interviews, an approach through narrative offers fewer pitfalls and greatly enhanced realism. The chapter ends with an overview of current work in psychology relevant to the approach developed in this book.