This paper is devoted to the late Leibnizian doctrine of vinculum substantiale. In the first section I sketch the old problem of possibility of composite substances. This possibility is refuted on the ground of Monadism (presented in section two). However Leibniz’s correspondence with Des Bosses contains new thoughts concerning composite substances. A vinculum enters the stage as a real unifier, transforming aggregates of monads into genuine substances (section three). In the last section I give a systematic interpretation of a vinculum.I (...) start with the thesis that every composed thing, which is not a pure plurality of objects, must have two structures: the whole-parts structure and the subject-properties structure. In the case of substances the latter is ontologically prior over the former. A vinculum is a subject-of-properties (accidents) determining such a way of composition which makes a compound entity a true substance. Since Leibniz still thought a vinculum unifies independent (existentially and with regard to activity) substances he was inclined to separate a vinculum from the integrated monads and finally conceived it as an additional relatively independent monad-like substance. (shrink)