In any case, it is assumed that causation is some kind of relation between two distinct items of some sort. This assumption has only rarely been questioned. One good reason for this is that given any causal process, it seems easy to come up with defi nitions of R, a, and b, such that the relational statement R(a,b) is true if and only if the causal process occurs. For instance, if one ball pushes another one, there is a relation between the two balls and their movements. This is why D. H. Mellor, for instance, begins his argument against the assumption that causation is a real relation by distinguishing between relations and real relations. According to Mellor, R(a,b) stands for a real relation only if both a and b exist. Thus he agrees with the general assumption that all causal processes may be represented by statements of the form R(a,b), and only objects that some things that may be represented by R(a,b) are not real relations.8 David Lewis does the same in his argument against the claim that causation is a relation. He argues that even though the void is nothing, it has causal properties. Since he also assumes that relations cannot hold among absences, he concludes that causation cannot be a relation.9 I do not fi nd this argument convincing. It seems to me that ‘better than’, for instance, is a real relation, even though there are many things that are better than nothing. So it cannot be true that real relations cannot hold among absences. Nonetheless, I agree with Lewis’s conclusion: Causation is not a relation.