The classification of legal doctrine provides "the familiar landmarks of thought" that order our understanding of the law. The "stark impossibility" of the classification of animals in the Chinese encyclopedia poses for us questions like those Foucault asks: How do we think about doctrinal categories, and what do we hope to achieve? What are the consequences of thinking in categories, and of thinking in the categories that we do? Does the way we classify help or hinder us in achieving our objectives? This article begins to answer these questions about the classification of legal doctrine in general, and about contract and tort law in particular. Only recently has the classification problem received much attention, for the first time since the 1920s. The relative lack of attention is not a sign of the problem's insignificance. On the contrary, the nature and importance of doctrinal classification are implicated in a number of current doctrinal and scholarly debates. The concern about doctrinal classification is a general one, but it has been of particular importance to contract and tort law. The classification issue figures prominently in doctrinal controversies, and contract and tort scholars have devoted considerable attention to it. This article focuses on the boundary between contract and tort and in particular on the issue of liability for non-physical injury caused by defective products. After explaining the significance of the doctrinal classification problem and how it has unfolded along the tort-contract boundary, this article presents two different accounts of how doctrinal classification works and what it means. Each of these two accounts of doctrinal classification is simultaneously second-hand and first-hand, both a way of understanding the process from the outside and a description of the actual process at work itself. This duality can be confusing, but the simultaneity is necessary since "how we think about law" and "the law that we think about" are not really two different things; definition creates reality as much as it orders it. The first description of doctrinal classification is relatively systematic, possessing at least a degree of formal rigor. The second is more of a holistic, interpretive account of the process as based on doctrinal paradigms; it looks beyond announced reasoning to the legal community's cognitive processes. Many persons within the mainstream of legal thought would subscribe to the first account, at least in structure if not in substance. The second account is more removed from traditional views. The article also criticizes the system of doctrinal classification presented by each account. The two accounts are useful because they suggest different kinds of flaws. Finally, the article suggests variations of the paradigmatic approach that address some of the problems with classification as currently practiced.