Most things have parts. Moreover, very often we can explain the behaviour of compound things in terms of the behaviour of its constituents. Thus it seems to be plausible that compound things behave the way they do in virtue of the behaviour of the parts. That is, at least, what the microphysicalist claims. ‘[E]verything non-microphysical is composed out of microphysical entities and is governed by microphysical laws’, as Philip Pettit put it (1994: 253). The parts or constituents of compound systems are taken to be the ultimate agents. I will argue that this view is wrong.