and we see the vraisemblance of the new rule which is to support the logical priority of the premisses. Without a formal break in the deduction, without an assignable falsity in any premiss, we may have got from a region of self-evidence—say of arithmetical truth—to one of questionable results in any concrete science. And we may express our sense of what has befallen us in the extraordinary statement that the falsity of Conclusions in such a deduction would not involve the falsity of the Premisses—and so the truth of the Premisses does not involve the truth of the Conclusion. We must be testing the Conclusion by the ideal of science, and the later Premisses only by the standard of empirical fact. Bare Conjunction is thus introduced into the very nerve of inference,[1] and not merely made the object of a precaution, as in the traditional
- ↑ The point that a rational nexus is everywhere assumed by science as in principle possible would seem too obvious to be insisted on, were it not that some theorists seem inclined to accept the survival of bare conjunction in the outskirts of knowledge as representing a feature of the universe. This, I take it, is Mr. Spaulding's view (p. 220). It is, is it not? the doctrine of contingent truth. But can it be made consistent with the other part of his view? (see Digression, p. 70). I therefore cite from an authority who shares in many ways our authors' tendencies some evidence of the need which science recognises to postulate rational nexus, in principle, under every empirical observation. "The peculiarity of Geometry is that it became a purely rational science earlier and by more rapid stages than could be the case with other departments of physical
