83. It should now be apparent that the familiar problems of reference and modality can be satisfactorily resolved without either abandoning substitutivity of identity in modal contexts or invoking an intensional ontology. It might be argued, however, that the suggested resolution avoids invoking an intensional ontology only because the modal predicates, e.g. Nec and Believes, are left unanalysed. This may well be true but it should be appreciated that the analysis of modal predicates is an issue which is independent of both the problem of reference in modal contexts and the problem of the logical structure of propositions containing such contexts. These problems can be settled, as we have seen, without also settling how the modal predicates are to be analysed. Hence, if an intensional ontology is in the end required, it is required to analyse the modal predicates, not to solve the problems of reference and logical structure. This, we suggest, is an interesting and perhaps unexpected result.