Splitting and lumping in the classification of social kinds: the case of money

This paper advances a pluralistic theory of social kinds, arguing that successful scientific classification in the social sciences arises from the co-determination of two constitutive axes. The first axis comprises different modes of ontological grounding, including historical lineages, multifunctional clusters, and isomorphic causal mechanisms. The second axis captures a spectrum of scientific interests, such as generalization versus specificity and explanation versus intervention. Projectible kinds are positioned at viable coordinates on these axes, where a robust ontological ground aligns with a specific scientific aim. This framework explains why some folk categories, such as everyday money, can be legitimately refined into multiple theoretical kinds, while others, such as hysteria, cannot, thereby ruling out relativistic “anything goes” conclusions. To illustrate the model, we examine the historical debate between the quantity theory and the classical theory of money. Economists strategically employ lumping and splitting to construct kinds suited to different scientific purposes. Lumping combines diverse monetary forms into a single, broader kind to support general theorization, while splitting differentiates money into narrower kinds to capture historical specificity. The framework yields two main insights. First, classical monetary economists treated kinds such as metallic or convertible money as projectible not through shared lineage, as Millikan’s historical kinds require, but through causal analogies. Second, the model challenges the claim that genuine kinds must support multiple generalizations. In the quantity theory, a ‘thin’, lumped kind of money remains scientifically legitimate because of the strong co-determination between a stable causal regularity and a scientific interest in macroeconomic prediction.