One of the methods proposed for conversation analysis (CA) referred to in the previous chapter (Chapter 8) is the use of “collections” as both data for, and the verification of, analyses. In this chapter, a deeper appreciation is developed of how turning a practical procedure into a formalised method exemplifies the gravitation of CA towards full-blown constructive analysis. The chapter specifically examines the way in which “collections” of transcribed fragments are used to generate and substantiate “findings”. The discussion refers to a dispute about “epistemics” in conversation and argues that the assembly and analytic use of collections misunderstands one of the major methodological points made by Sacks, Schegloff and others on how investigations of the endogenous organisation of conversation must demonstrate how analytical claims are evidently featured in the detailed organisation of singular instances of naturally occurring conversation.