We must, therefore, distinguish a possible action from an action of which it is possible to think. By the former we mean an action which no known cause would prevent, _provided_ the idea of it occurred to us: and that one among such actions, which will produce the greatest total good, is what we mean by duty. Ethics certainly cannot hope to discover what kind of action is always our duty in this sense. It may, however, hope to decide which among one or two such possible actions is the best: and those which it has chosen to consider are, as a matter of fact, the most important of those with regard to which men deliberate whether they shall or shall not do them. A decision with regard to these may therefore be easily confounded with a decision with regard to which is the best possible action. But it is to be noted that even though we limit ourselves to considering which is the better among alternatives likely to be thought of, the fact that these alternatives might be thought of is not included is what we mean by calling them possible alternatives. Even if in any particular case it was impossible that the idea of them should have occurred to a man, the question we are concerned with is, which, if it had occurred, would have been the best alternative? If we say that murder is always a worse alternative, we mean to assert that it is so, even where it was impossible for the murderer to think of doing anything else. The utmost, then, that Practical Ethics can hope to discover is which, among a few alternatives possible under certain circumstances, will, on the whole, produce the best result. It may tell us which is the best, in this sense, of certain alternatives about which we are likely to deliberate; and since we may also know that, even if we choose none of these, what we shall, in that case, do is unlikely to be as good as one of them, it may thus tell us which of the alternatives, among which we _can_ choose, it is best to choose. If it could do this it would be sufficient for practical guidance. =93.= But (3) it is plain that even this is a task of immense difficulty. It is difficult to see how we can establish even a probability that by doing one thing we shall obtain a better total result than by doing another. I shall merely endeavour to point out how much is assumed, when we assume that there is such a probability, and on what lines it seems possible that this assumption may be justified. It will be apparent that it has never yet been justified--that no sufficient reason has ever yet been found for considering one action more right or more wrong than another.