Sociologists since Simmel have been interested in social circles as essential features of friendship networks. Although network analysis has been increasingly used to uncover patterns among social relationships, theoretical explanations of these patterns have been inadequate. This paper presents a theory of the social organization of friendship ties. The approach is based upon Homans's concepts of activities, interactions, and sentiments and upon the concept of extra-network foci organizing social activities and interaction. The theory is contrasted with Heider's balance theory. Implications for transitivity, network bridges, and density of personal networks are discussed and presented as propositions. The focus theory is shown to help explain patterns of friendships in the 1965-66 Detroit Area Study. This paper is intended as a step toward the development of integrated theory to explain interrelationships between networks and other aspects of social structure. Implications for data analysis are discussed. Sociologists have long recognized the importance of patterns in networks of relations that connect individuals with each other. Simmel (1955) described modern society as consisting of loosely connected social circles of relationships. Granovetter (1973) has indicated the general significance of these social circles for communication, community organization, and social conflict. Various studies have supported this picture of the essential patterns in social networks, including Moreno's sociometry (1953), Milgram's "small world" experiments (1967), and Kadushin's observations (1966). Unfortunately, the study of social networks has often been carried out without concern for the origins in the larger social context. Most network analysis ends with description and labeling of patterns; and when explanations of patterns are offered, they frequently rely upon inherent tendencies within networks to become consistent, balanced, or transitive. As a consequence of such atheoretical and/or self-contained network theoretical approaches, data are collected and data analysis techniques are devised for