Social indicators have not fulfilled their promise, or at least have not lived up to the expectations held of them in the late 1950s and 1960s. Despite the continued growth of social statistics, produced both by governments and other organisations, the aim of producing precise, concise and evaluatively neutral measures of the state of society and of change in society has apparently eluded some of the best minds of the social science and governmental statistics communities. Whereas a wide range of economic indicators and data are readily available, if not without their problems (cf. Johnson, 1988), and integrated into the concepts of economic theory, standard measures of crime, health, well-being, education and many other social characteristics have proven much more difficult to construct and establish as standard yardsticks of social conditions. This note considers some of the reasons for these difficulties. It relates specifically to the aspiration to construct social indicators, not to social statistics more generally (as reviewed in, for example, Carley, 1981).