It is obvious that the difference between science and metaphysics in the matter of necessity is that, whereas science cautiously says, "If certain conditions are fulfilled, certain results will ensue," metaphysics boldly says, "The conditions on which the whole future depends are already absolutely fixed." Once more, this metaphysical theory may be true; but, if so, it is not from science that it derives its truth. The transition from the "if" of science to the "must" of metaphysics is illogical, though not unnatural, and is facilitated by a certain amount of obscurity, which can be thrown over it by drawing illustrations from the past. Thus, if an event has already taken place, we may infer with certainty from the fact of its occurrence that the conditions necessary to produce it were realised. And as each of those conditions must have had a cause, we can infer again that the conditions requisite to produce them were fulfilled. And so we may travel back _ad infinitum_ along a never-ending chain of cause and effect, always moving from one fixed and necessitated event to another event equally necessitated and fixed. Thus the whole past history of the universe may be exhibited as a necessary sequence of events; and the inference may be drawn, and for the purposes of the theory of metaphysical necessity must be drawn, that because the occurrence of an event proves that the conditions required for its production were realised, therefore they and they alone were bound to be realised. Yet this is simply our old familiar _non sequitur_ thrown into the past tense. It is true that if I walk over a precipice I shall fall, according to the law of gravitation. But I am not therefore bound to walk over. It is true that the man who fell over the cliff obeyed the law of gravity. But we cannot infer either from the law of gravitation or from the fact of his falling that he was bound to fall. We can infer that the conditions requisite to produce the fall were present, but we cannot infer from the fall that they were bound to be present. It may be quite true that they were bound to be present, but the effect which followed on them cannot be alleged either as the cause or the proof of such necessity. We must look for the reason of the necessity--if there be any necessity in the case--elsewhere. Shall we, then, say that the conditions of the fall were themselves effects of prior causes, without which they would not have happened? That again is true, but the fact that Z would not have happened had not Y preceded, is not in itself any proof that Y was bound to happen. And so we may travel back _ad infinitum_ along the never-ending chain of cause and effect without ever finding ourselves in a position to infer from the law that everything must have a cause, that this cause was bound to operate rather than that. The occurrence of Z is no proof that Y was bound to happen, nor is the fact that Y really