This talk offers a novel resolution of a familiar tension, that between approaches in which the difference between definites and indefinites is based on uniqueness, and those in which it is based on novelty/familiarity. The claim is that familiarity conditions on definites and indefinites can be derived from a uniqueness-based treatment of the definite article. Specifically, if an ordinary dynamic semantics is defined to allow tracking of discourse referents, and an indexing mechanism is defined to allow identification of descriptions with referents, then both novelty of indefinites and familiarity of definites can be derived without stipulating lexically that the articles have these properties. The derivation relies entirely on principles that are commonly used (e.g. the uniqueness requirement for definites, and Heim’s Maximize Presupposition principle). However, the derivation does not predict familiarity for all definites; in particular, familiarity for definites in case of semantic uniqueness will not be required. Furthermore, the uniqueness requirement will effectively drop away in case of familiarity. The proposal will be explicated by first defining a dynamic system that embodies many of the insights in Heim’s (1982,1983) seminal work on definites, and then showing how basic properties of her treatment can be derived and improved upon.