ABSTRACT
The proposed method of support-ranking, with the conjunction principle that it entails, may be extended to cover causal hypotheses, by embodying trials on controls. People sometimes speak as if correlational generalisations themselves can support a theory. Perhaps a stated value for the acceleration of falling bodies, for instance, may be said to support Newtonian mechanics. A dimension of universal assessment is provided by criteria for accepting a hypothesis at a particular time and for a particular purpose, even if the grades accorded by one support-function are incommensurable with those accorded by another. The paradoxes are only puzzling until the vague and ambiguous term ‘confirmation’, has been analysed out into its various possible meanings of ‘support’, ‘favourable test-results’, and ‘sample-based estimation’, and the different implications of these meanings have been articulated. Much the same is true also of another group of paradoxes, which N. Goodman has formulated in terms of the same verb ‘to confirm’.
