Our first conclusion as to the subject-matter of Ethics is, then, that there is a simple, indefinable, unanalysable object of thought by reference to which it must be defined. By what name we call this unique object is a matter of indifference, so long as we clearly recognise what it is and that it does differ from other objects. The words which are commonly taken as the signs of ethical judgments all do refer to it; and they are expressions of ethical judgments solely because they do so refer. But they may refer to it in two different ways, which it is very important to distinguish, if we are to have a complete definition of the range of ethical judgments. Before I proceeded to argue that there was such an indefinable notion involved in ethical notions, I stated (§ 4) that it was necessary for Ethics to enumerate all true universal judgments, asserting that such and such a thing was good, whenever it occurred. But, although all such judgments do refer to that unique notion which I have called good, they do not all refer to it in the same way. They may either assert that this unique property does always attach to the thing in question, or else they may assert only that the thing in question is a cause or necessary condition for the existence of other things to which this unique property does attach. The nature of these two species of universal ethical judgments is extremely different; and a great part of the difficulties, which are met with in ordinary ethical speculation, are due to the failure to distinguish them clearly. Their difference has, indeed, received expression in ordinary language by the contrast between the terms good as means and good in itself, value as a means and intrinsic value. But those terms are apt to be applied correctly only in the more obvious instances; and this seems to be due to the fact that the distinction between the conceptions which they denote has not been made a separate object of investigation. This distinction may be briefly pointed out as follows. (§ 15 ¶ 1)

Whenever we judge that a thing is good as a means, we are making a judgment with regard to its causal relations: we judge both that it will have a particular kind of effect, and that that effect will be good in itself. But to find causal judgments that are universally true is notoriously a matter of extreme difficulty. The late date at which most of the physical sciences became exact, and the comparative fewness of the laws which they have succeeded in establishing even now, are sufficient proofs of this difficulty. With regard, then, to what are the most frequent objects of ethical judgments, namely actions, it is obvious that we cannot be satisfied that any of our universal causal judgments are true, even in the sense in which scientific laws are so. We cannot even discover hypothetical laws of the form Exactly this action will always, under these conditions, produce exactly that effect. But for a correct ethical judgment with regard to the effects of certain actions we require more than this in two respects. (1) We require to know that a given action will produce a certain effect, under whatever circumstances it occurs. But this is certainly impossible. It is certain that in different circumstances the same action may produce effects that are utterly different in all respects upon which the value of the effects depends. Hence we can never be entitled to more than a generalisation—to a proposition of the form This result generally follows this kind of action; and even this generalisation will only be true, if the circumstances under which the action occurs are generally the same. This is in fact the case, to a great extent, within any one particular age and state of society. But, when we take other ages into account, in many most important cases the normal circumstances of a given kind of action will be so different, that the generalisation which is true for one will not be true for another. With regard then to ethical judgments which assert that a certain kind of action is good as a means to a certain kind of effect, none will be universally true; and many, though generally true at one period, will be generally false at others. But (2) we require to know not only that one good effect will be produced, but that, among all subsequent events affected by the action in question, the balance of good will be greater than if any other possible action had been performed. In other words, to judge that an action is generally a means to good is to judge not only that it generally does some good, but that it generally does the greatest good of which the circumstances admit. In this respect ethical judgments about the effects of action involve a difficulty and a complication far greater than that involved in the establishment of scientific laws. For the latter we need only consider a single effect; for the former it is essential to consider not only this, but the effects of that effect, and so on as far as our view into the future can reach. It is, indeed, obvious that our view can never reach far enough for us to be certain that any action will produce the best possible effects. We must be content, if the greatest possible balance of good seems to be produced within a limited period. But it is important to notice that the whole series of effects within a period of considerable length is actually taken account of in our common judgments that an action is good as a means; and that hence this additional complication, which makes ethical generalisations so far more difficult to establish than scientific laws, is one which is involved in actual ethical discussions, and is of practical importance. The commonest rules of conduct involve such considerations as the balancing of future bad health against immediate gains; and even if we can never settle with any certainty how we shall secure the greatest possible total of good, we try at least to assure ourselves that probable future evils will not be greater than the immediate good. (§ 16 ¶ 1)

There are, then, judgments which state that certain kinds of things have good effects; and such judgments, for the reasons just given, have the important characteristics (1) that they are unlikely to be true, if they state that the kind of thing in question always has good effects, and (2) that, even if they only state that it generally has good effects, many of them will only be true of certain periods in the world’s history. On the other hand there are judgments which state that certain kinds of things are themselves good; and these differ from the last in that, if true at all, they are all of them universally true. It is, therefore, extremely important to distinguish these two kinds of possible judgments. Both may be expressed in the same language: in both cases we commonly say Such and such a thing is good. But in the one case good will mean good as means, i.e. merely that the thing is a means to good—will have good effects: in the other case it will mean good as end—we shall be judging that the thing itself has the property which, in the first case, we asserted only to belong to its effects. It is plain that these are very different assertions to make about a thing; it is plain that either or both of them may be made, both truly and falsely, about all manner of things; and it is certain that unless we are clear as to which of the two we mean to assert, we shall have a very poor chance of deciding rightly whether our assertion is true or false. It is precisely this clearness as to the meaning of the question asked which has hitherto been almost entirely lacking in ethical speculation. Ethics has always been predominantly concerned with the investigation of a limited class of actions. With regard to these we may ask both how far they are good in themselves and how far they have a general tendency to produce good results. And the arguments brought forward in ethical discussion have always been of both classes—both such as would prove the conduct in question to be good in itself and such as would prove it to be good as a means. But that these are the only questions which any ethical discussion can have to settle, and that to settle the one is not the same thing as to settle the other—these two fundamental facts have in general escaped the notice of ethical philosophers. Ethical questions are commonly asked in an ambiguous form. It is asked What is a man’s duty under these circumstances? or Is it right to act this way? or What ought we to aim at securing? But all these questions are capable of further analysis; a correct answer to any of them involves both judgments of what is good in itself and causal judgments. This is implied even by those who maintain that we have a direct and immediate judgment of absolute rights and duties. Such a judgment can only mean that the course of action in question is the best thing to do; that, by acting so, every good that can be secured will have been secured. Now we are not concerned with the question whether such a judgment will ever be true. The question is: What does it imply, if it is true? And the only possible answer is that, whether true or false, it implies both a proposition as to the degree of goodness of the action in question, as compared with other things, and a number of causal propositions. For it cannot be denied that the action will have consequences: and to deny that the consequences matter is to make a judgment of their intrinsic value, as compared with the action itself. In asserting that the action is the best thing to do, we assert that it together with its consequences presents a greater sum of intrinsic value than any possible alternative. And this condition may be realised by any of the three cases:—(a) If the action itself has greater intrinsic value than any alternative, whereas both its consequences and those of the alternatives are absolutely devoid either of intrinsic merit or intrinsic demerit; or (b) if, though its consequences are intrinsically bad, the balance of intrinsic value is greater than would be produced by any alternative; or (c) if, its consequences being intrinsically good, the degree of value belonging to them and it conjointly is greater than that of any alternative series. In short, to assert that a certain line of conduct is, at a given time, absolutely right or obligatory, is obviously to assert that more good or less evil will exist in the world, if it be adopted than if anything else be done instead. But this implies a judgment as to the value both of its own consequences and of those of any possible alternative. And that an action will have such and such consequences involves a number of causal judgments. (§ 17 ¶ 1)

Similarly, in answering the question What ought we to aim at securing? causal judgments are again involved, but in a somewhat different way. We are liable to forget, because it is so obvious, that this question can never be answered correctly except by naming something which can be secured. Not everything can be secured; and, even if we judge that nothing which cannot be obtained would be of equal value with that which can, the possibility of the latter, as well as its value, is essential to its being a proper end of action. Accordingly neither our judgments as to what actions we ought to perform, nor even our judgments as to the ends which they ought to produce, are pure judgments of intrinsic value. With regard to the former, an action which is absolutely obligatory may have no intrinsic value whatsoever; that it is perfectly virtuous may mean merely that it causes the best possible effects. And with regard to the latter, these best possible results which justify our action can, in any case, have only so much of intrinsic value as the laws of nature allow us to secure; and they in their turn may have no intrinsic value whatsoever, but may merely be a means to the attainment (in a still further future) of something that has such value. Whenever, therefore, we ask What ought we to do? or What ought we to try to get? we are asking questions which involve a correct answer to two others, completely different in kind from one another. We must know both what degree of intrinsic value different things have, and how these different things may be obtained. But the vast majority of questions which have actually been discussed in Ethics—all practical questions, indeed—involve this double knowledge; and they have been discussed without any clear separation of the two distinct questions involved. A great part of the vast disagreements prevalent in Ethics is to be attributed to this failure in analysis. By the use of conceptions which involve both that of intrinsic value and that of causal relation, as if they involved intrinsic value only, two different errors have been rendered almost universal. Either it is assumed that nothing has intrinsic value which is not possible, or else it is assumed that what is necessary must have intrinsic value. Hence the primary and peculiar business of Ethics, the determination of what things have intrinsic value and in what degrees, has received no adequate treatment at all. And on the other hand a thorough discussion of means has been also largely neglected, owing to an obscure perception of the truth that it is perfectly irrelevant to the question of intrinsic values. But however this may be, and however strongly any particular reader may be convinced that some one of the mutually contradictory systems which hold the field has a given correct answer either to the question what has intrinsic value, or to the question what we ought to do, or to both, it must at least be admitted that the questions what is best in itself and what will bring about the best possible, are utterly distinct; that both belong to the actual subject-matter of Ethics; and that the more clearly distinct questions are distinguished, the better is our chance of answering both correctly. (§ 17 ¶ 2)