In " two substances ", "two " is not an accident, because an accident is a (replaceable) determination of a genuine substance, or at least a mereo-totality (that is, a tight contiguum). An individual accident thus is a determination of one single being, not of two or more. So if the two substances together form a tight contiguum, it is a mereo-totality. And this mereo-totality consists of two parts. Generally in a tight contiguum the number of parts is not fixed (it is just a [tight] aggregate), and so this number is per accidens with respect to the species represented by this particular aggregate. And in this way we can say that ' two ' is an accident. And it is an accident belonging to discrete quantity and so belonging to the predicament Quantity.
If, on the other hand, the two substances form just a scattered aggregate, then it isn't a mereo-totality at all, let alone of course a (single) substance. And then we cannot say that ' two ' is an accident and the two substances its carrier. And it is this very case that is rightly refuted by Ockham.