This chapter presents a case study that draws together many of the critical issues discussed in previous chapters of this volume. It reviews a group of studies that address a familiar discursive object – apologies – in a demonstration of how constructive analysis has become intimately embedded in contemporary conversation analysis (CA). Although the studies examined present themselves as contributions to ethnomethodology and CA, they use Goffman's notion of “virtual offence” to set up the analysis of a collection of transcribed fragments. The studies in the collection make eclectic use of taxonomies, statistical correlations, coding, indexing, and operational definitions. Of particular interest is a pair of studies, the first of which sets out to test a social science hypothesis that the design of apology is systematically correlated with the (imagined) severity of the “virtual offence” in question. The second adopts ethnomethodological auspices to critically examine what the first study fails to explain, and advances an unusual thesis that treats incorrigible interpretation as both a topic and resource.