The aim of this paper is to motivate a view on dispositions according to which dispositions and their manifestations are partially identical, the DM identity theory. It sets out by extrapolating the desiderata of a dispositionalist account of properties. It then shows that the previous theories are burdened with different problems, whose common cause, so the argument goes, is the separation assumption, which almost all share. It states that dispositions and their manifestations are numerically distinct. The paper then explores whether (...) the separation assumption can be abandoned and shows that there are precursors of a DM identity theory. The DM identity theory is then outlined in its central features and it is outlined how they can fulfil the desiderata of dispositionalism. (shrink)