I am interested in how geologists (and Earth scientists generally) reason about the past when all they can observe are traces of events and processes that occurred long ago. Unlike experimental scientists, historical scientists like geologists cannot intervene to change variables or repeat a trial, so other means have to be sought to make sense of the deep past. Central to the task of geological reasoning, though relatively unacknowledged and underappreciated, is the creation of narratives. Among other things, narratives can serve to organise puzzling or apparently contradictory observations, and they can help to weigh up competing hypotheses by for example, creating counterfactual arguments that connect up causal chains in different ways. Although Earth science narratives play important roles in communication, pedagogy and popularisation, it is in their function as essential elements of the geological reasoning process that I am particularly interested. In this enterprise, I am drawing on my experience in a former career as a petroleum geoscientist.