Does it follow, as Feyerabend and Kuhn contend, that there can then be no rational grounds for choosing between them? No. For we can allow that a theory Ta is preferable to a theory Tb, even if they are incommensurable, provided that Ta can explain under its descriptions almost all the phenomena P1…Pn that Tb can explain under its descriptions plus some significant phenomena that Tb cannot explain. This depends of course upon a explicit recognition of the need for a philosophical ontology or intransitive dimension in the philosophy of science. But such an ontology is implicit in the very formulation of the problem. For to say that two theories conflict, clash or are in competition presupposes that there is something – a domain of real objects or relations existing and acting independently of their descriptions – over which they clash. (No one bothers to say that the rules of cricket and football are incommensurable.) Of course, it may be that the two theories are only in competition over a very small domain (as may be the case for example with Marxism and psychoanalysis), so that Lakatosian-type decision rules are of very little help in choosing between them, but this is not then the problem of incommensurability.