Definitions that are favored by students of violence often focus on the 'inflicting of physical harm.' These definitions are 'product centered' and are advantageous because they highlight behavior that fits the researcher's criteria and circumscribe a specific universe to study. However, unless researchers presuppose homogeneity of process, a product-centered science of violence could encroach on diverse domains involving widely disparate expertise. Unless product and process exploration are combined, researchers risk disjunctures of inquiry or pseudo-solutions. The methodology deployed in the study of family violence, in particular, must be multifaceted and must handle all sorts of data in varying combinations so that the phenomena observed can be meaningfully subdivided, explored in depth and breadth, and described and traced. Typologies that divide violence phenomena (products) into homogeneous syndromes with similar dynamics (processes) facilitate efforts at prevention and resocialization. Researchers' reluctance to explore the violence process directly is unfortunate as persons intimate with violence can tell the researcher about the 'how' and illuminate the 'why.' In the absence of this strategy, a science of violence seems fated to be strong on facts that are weakly buttressed by remote inferences. Five notes and seven references are included.