After decades of research, opinions are still split as to whether geminates should be represented as long or as heavy. In this paper, we attempt to resolve this issue by entertaining a model that rests on the assumption that all underlying geminates are moraic consonants but they might not emerge as such on the surface. We argue that this intuition—due to Davis (2011)—can be formalized in a theoretical model where 1) different degrees of prosodic integration are possible and 2) phonetic interpretation can refer to the difference between epenthetic and underlying elements. This model predicts that the questions of whether a mora has an effect for segmental length and whether it has an effect for syllabic weight are independent from each other. We show that this representational consequence correctly predicts the typology of attested combinations of geminates in different positions and singleton coda consonants in the languages of the world.