oscillation in space, while it really consists in a dynamic progress in which the self and its motives, like real living beings, are in a constant state of becoming. The self, infallible when it affirms its immediate experiences, feels itself free and says so; but, as soon as it tries to explain its freedom to itself, it no longer perceives itself except by a kind of refraction through space. Hence a symbolism of a mechanical kind, equally incapable of proving, disproving, or illustrating free will. [Sidenote: Is prediction of an act possible? Probable and infallible conclusions.] But determinism will not admit itself beaten, and, putting the question in a new form, it will say: "Let us leave aside actions already performed: let us consider only actions that are to come. The question is whether, knowing from now onwards all the future antecedents, some higher intelligence would not be able to predict with absolute certainty the decision which will result."--We gladly agree to the question being put in these terms: it will give us a chance of stating our own theory with greater precision. But we shall first draw a distinction between those who think that the knowledge of antecedents would enable us to state a _probable_ conclusion and those who speak of an _infallible_ foresight. To say that a certain friend, under certain circumstances, will very probably act in a certain way, is not so much to predict the future conduct of our friend as to pass a judgment on his present character, that is to say, on his past. Although our feelings, our ideas, our character, are constantly altering, a sudden change is seldom observed; and it is still more seldom that we cannot say of a person whom we know that certain actions seem to accord fairly well with his nature and that certain others are absolutely inconsistent with it. All philosophers will agree on this point; for to say that a given action is consistent or inconsistent with the present character of a person whom one knows is not to bind the future to the present. But the determinist goes much further: he asserts that our solution is provisional simply because we never know all the conditions of the problem: that our forecast would gain in probability in proportion as we were provided with a larger number of these conditions; that, therefore, complete and perfect knowledge of all the antecedents without any exception would make our forecast infallibly true. Such, then, is the hypothesis which we have to examine. [Sidenote: To know _completely_ the antecedents and conditions of an action is to be actually performing it.]