Making collections of conversational/interactional phenomena is a cornerstone of CA’s methodology. Our aim in CA research is to identify the practices through which speakers of a natural language conduct action in inter-action, the practices that enable speakers to engage in conduct that is meaningful to one another (the accountability-as-intelligibility of social conduct). The practices for talk-in-interaction are recurrent phenomena; they are to be found in recurrent patterns of talk – in recurrent sequential positions or environments, in recurrent sequence patterns, and in recurrent features of turn design such as linguistic format including morphosyntactic constructions. Whilst recurrence may not by itself be a sufficient condition for determining that an object or pattern constitutes a practice, it is a necessary condition. It is necessary to show that an object, pattern etc. works in a particular way systematically – and ‘systematically’ requires recurrence. Thus, the identification, the discovery or uncovering of practices in talk-in-interaction rests on building collections of cases of a phenomenon in order to find whether it is systematically associated with some recurrent pattern. In this chapter I describe the history of one such collection assembled by Gail Jefferson, a collection of apologies that served as the basis for several analyses and publications.