This paper presents an empirical analysis of simplification processes in the scientific work place. Any scientific task involves complex sets of problems and contingencies. But the conclusions produced do not represent all details of the work performed, nor do consumers take full account of the complexity of results. This paper examines the ways in which chains of inference are simplified at all stages of the research work, from research design and sampling to publication. What is deleted in the `fact-making' process? What constraints operate to make this deletion necessary? This paper examines institutional and intersectional constraints and processes which affect the final work. The analytic approach of the paper is symbolic interactionist/Pragmatist, and field data were collected by participant observation.