(2) Though the ‘facts’ of the moment fail to include all the facts, they often manage to include too much. The ‘facts’ are not all fact. They include unknown, and often large, amounts of prejudice, illusion, error, superstition, and other remnants of the lurid past and stormy youth of every science. It is useless to repine at this inevitable consequence of past history, and childish to try to purge it away by defining as science only what _ex hypothesi_ is free from such contaminations. To restrict the logical interest to science _qua_ science, which is by definition infallible, is to forbid any logical treatment of the sciences we actually possess. But the logician should surely be encouraged to study the processes by which the sciences correct their initial errors and consolidate their acquisitions. (3) It follows on both these grounds that the ‘facts’ of which a science takes cognizance will be subject to change. As the science grows, ‘new’ facts will come into it, and old facts will be discarded as erroneous. In particular, facts which at first were only inferred on theoretic grounds will be actually observed, even as ‘Neptune’ was the fruit of a theory about the perturbations of Uranus. Hence the antithesis of ‘theory’ and ‘fact’ must not be taken as absolute: they must be expected to play into each other’s hands. It is the business of theories to forecast ‘facts’, and of facts to form points of departure for theories, which again, when verified by the new facts to which they have successfully led, will extend the borders of knowledge. Incidentally, however, this interaction between fact and theory often renders it difficult to decide whether a scientific doctrine is better regarded as a ‘theory’ or as a ‘fact’, and leads to differences of opinion. But it can hardly be wrong to advise the scientific mind to practise hospitality towards new facts, while it is no less fitting to show generosity towards old servants that have done their work and can now advantageously be retired. It is ungrateful to abuse them as ‘errors’, and to despise them with the lofty contempt of the higher knowledge to which they have conducted. And in both cases the truly scientific attitude may be attained if an element of fanaticism is not imported into the conception of truth by attributing to it an absoluteness which no human truth in fact possesses.