In the literature concerning the semantics of proper names, almost no attention has been paid to methodological questions, such as: What should be the formal structure of theories of reference for the various types of singular terms including names? What should be the subject matter of such theories, and what kinds of facts are relevant to determining their truth-values? In this paper, I wish to suggest some answers to these questions, answers that I believe provide an accurate conception of the form that a theory of names should take. Such a conception is valuable because it constitutes a constraint that any correct theory of names must satisfy, and so it provides a tool for evaluating particular theoretical proposals concerning names. Using this tool, I will argue that no causal theory of names can be correct and that the true theory must be a kind of description theory. The fact that the true theory of names takes the form it does has some surprising consequences regarding the semantic structure of natural languages and the concept of meaningfulness.