The notions of "abstract" and "concrete" are central to the conceptualization of mathematical knowing and learning. Much of the literature takes a dualist approach, leading to the privileging of the former term at the expense of the latter. In this article, we provide a concrete analysis of a scientist interpreting an unfamiliar graph to show how engagement with some object leads to the working out of existing, concrete practical understanding and the articulation of categorical statements ("generalizations"); because the scientist knew something at the end of his interpretive work that he did not prior to it, the event is understood to constitute an episode of learning. The analysis shows that rather than being a movement from concrete to abstract or from abstract to concrete, development occurs in a movement that appears to be simultaneously from concrete to abstract and from abstract to concrete. (Contains 8 figures.)