Can the realist bundle theory account for the numerical difference between qualitatively non-discernible concrete particulars?

According to the realist bundle theory (RBT), ordinary concrete objects are complex wholes having universals, and only universals, as constituents. One charge usually levelled against RBT is the so-called ‘argument from numerical difference’ (AND). According to AND, RBT fails because it cannot explain the possibility of qualitatively identical yet numerically distinct objects, a possibility that any sound ontology of concrete particulars should be able to account for. This paper (i) presents AND (ii) explains one strategy that may prima facie liberate RBT from AND, and (iii) argues that this strategy comes at the cost of renouncing a methodological feature of ‘constituent ontologies’, the sort of ontology to which RBT belongs.