A family resemblance model of emotions is proposed which uses Darwin's discussion of emotions and Eleanor Rosch's and the philosopher Ludwig Wittgenstein's work on family resemblances. In Darwin's discussion of emotions, certain core features are thought to be widely shared by the members of the respective families of emotions, and more marginal features are shared by only a few members of each family. This approach to emotions is very similar to Rosch's conception of physical object concepts as entities having a high degree of internal structure and boundaries which are not well-defined. It is suggested here that the adoption of the family resemblances perspective renders emotions no more difficult to define than physical objects. Emotions sharing many of the features of an emotional family are more prototypical of the respective family (and therefore recognized and labeled more easily), than are weaker features and expressions of the same emotions. Additional variations in labeling would depend on the intensity of emotional expressions and of the stimulus conditions, on qualitative differences in stimuli, and on the complexity of the emotional expressions. It is surmised that children's early nonverbal emotional development and their learning of emotional language may occur in contexts in which highly prototypical members of the various emotional families are expressed. (RH)