In this study, we follow a long line of researchers in asking about the precise role of uniqueness and familiarity in the semantics of the English definite article the. We attempt to answer this question experimentally, by observing how definite descriptions behave in contexts where a speaker potentially uses an incorrect description, as in Donnellan’s classic martini scenario, where a speaker incorrectly believes there is a unique referent for their chosen description. In particular, we investigate how hearers interpret definite descriptions in contexts that are systematically manipulated to vary in whether they do or don't contain a unique referent satisfying the description, and whether the referent has or has not been made familiar via previous linguistic mention. Our experimental results reveal that both uniqueness (construed as uniqueness with respect to the common ground between the interlocutors) and familiarity (construed as strong familiarity or anaphoricity) can act as helpful cues to the hearer during the interpretation of a definite description. However, their effects are graded, with the presence of uniqueness leading to greater referential success than the presence of familiarity. We discuss the implications of these results on several existing standard theories of definiteness, and implement a version of the Rational Speech Acts model to help explain the ways in which the observed behavioral data cannot be fully explained on these theories.