Yet seems other than this. How this appearance is to be explained. 286. This relation which, going beyond our actual experience, we seem to infer as the explanation of invariable contiguity in place or time of certain impressions, and from which again we seem to infer the identity of an object of which the perception has been interrupted, is what we call necessary connection. It is their supposed necessary connection which distinguishes objects related as cause and effect from those related merely in the way of contiguity and succession, [1] and it is a like supposition that leads us to infer what we do not see or remember from what we do. If then the reduction of thought and the intelligible world to feeling was to be made good, this supposition, not being an impression of sense or a copy of such, must be shown to be an ‘impression of reflection,’ according to Hume’s sense of the term, _i.e._ a tendency of the soul, analogous to desire and aversion, hope and fear, derived from impressions of sense but not copied from them; [2] and the inference which it determines must be shown to be the work of imagination, as affected by such impression of reflection. This in brief is the purport of Hume’s doctrine of causation. [1] P. 376. [Book I, part III., sec. II.] [2] Above, par. 195. Inference, resting on supposition of necessary connection, to be explained before that connection. 287. After his manner, however, he will go about with his reader. The supposed ‘objective basis’ of knowledge is to be made to disappear, but in such a way that no one shall miss it. So dexterously, indeed, is this done, that perhaps to this day the ordinary student of Hume is scarcely conscious of the disappearance. Hume merely announces to begin with that he will ‘postpone the direct survey of this question concerning the nature of necessary connection,’ and deal first with these other two questions, viz. (1) ‘For what reason we pronounce it _necessary_ that everything whose existence has a beginning, should also have a cause?’ and (2) ‘Why we conclude that such particular causes must _necessarily_ have such particular effects; and what is the nature of that _inference_ we draw from the one to the other, and of the _belief_ we repose in it?’ That is to say, he will consider the inference from cause or effect, before he considers cause and effect as a relation between objects, on which the inference is supposed to depend. Meanwhile necessary connection, as a relation between objects, is naturally supposed in some sense or other to survive. In _what_ sense, the reader expects to find when these two preliminary questions have been answered. But when they have been answered, necessary connection, as a relation between objects, turns out to have vanished. Account of the inference given by Locke and Clarke rejected.