We may, then, attribute the obstinate prejudice that a knowledge of supersensible reality is a necessary step to a knowledge of what is good in itself, partly to a failure to perceive that the subject of the latter judgment is not anything real as such, and partly to a failure to distinguish the cause of our perception of a truth from the reason why it is true. But these two causes will carry us only a very little way in our explanation of why Metaphysics should have been supposed to have a bearing upon Ethics. The first explanation which I have given would only account for the supposition that a thing's reality is a necessary condition for tis goodness. This supposition is, indeed, commonly made; we find it commonly presupposed that unless a thing can be shewn to be involved in the constitution of reality, it cannot be good. And it is, therefore, worthwhile to insist that this is not the case; that Metaphysics is not even necessary to furnish part of the basis of Ethics. But when metaphysicians talk of basing Ethics on Metaphysics they commonly mean much more than this. They commonly mean that Metaphysics is the sole basis of Ethics—that it furnishes not only one necessary condition but all the conditions necessary to prove that certain things are good. And this view may, at first sight, appear to be held in two different forms. It may be asserted that merely to prove a thing supersensibly real is sufficient to prove it good: that the truly real must, for that reason alone, be truly good. But more commonly it appears to be held that the real must be good because it possesses certain characters. And we may, I think, reduce the first kind of assertion to no more than this. When it is asserted that the real must be good, because it is real, it is commonly also held that this is only because, in order to be real, it must be of a certain kind. The reasoning by which it is thought that a metaphysical enquiry can give an ethical conclusion is of the following form. From a consideration of what it is to be real, we can infer that what is real must have certain supersensible properties: but to have those properties is identical with being good—it is the very meaning of the word: it follows therefore that what has these properties is good: and from a consideration of what it is to be real, we can again infer what it is that has these properties. It is plain that, if such reasoning were correct, any answer which could be given to the question What is good in itself? could be arrived at by a purely metaphysical discussion and by that alone. Just as, when Mill supposed that to be good meant to be desired, the question What is good? could be and must be answered solely by an empirical investigation of the question what was desired; so here, if to be good means to have some supersensible property, the ethical question can and must be answered by a metaphysical enquiry into the question, What has this property? What, then, remains to be done in order to destroy the plausibility of Metaphysical Ethics, is to express the chief errors which seem to have led metaphysicians to suppose that to be good means to possess some supersensible property. (§ 72 ¶ 1)

What, then, are the chief reasons which have made it seem plausible to maintain that to be good must mean to possess some supersensible property or to be related to some supersensible reality? (§ 73 ¶ 1)

We may, first of all, notice one, which seems to have had some influence in causing the view that good must be defined by some such property, although it does not suggest any particular property as the one required. This reason lies in the supposition that the proposition This is good or This would be good, if it existed must, in a certain respect, be of the same type as other propositions. The fact is that there is one type of proposition so familiar to everyone, and therefore having such a strong hold upon the imagination, that philosophers have always supposed that all other types of propositions must be reducible to it. This type is that of the objects of experience—of all those truths which occupy our minds for the immensely greater part of our waking lives: truths such as that somebody is in the room, that I am writing or eating or talking. All these truths, however much they may differ, have this in common that in them both the grammatical subject and the grammatical object stand for something which exists. Immensely the commonest type of truth, then, is one which asserts a relation between two existing things. Ethical truths are immediately felt not to conform to this type, and the naturalistic fallacy arises from the attempt to make out that, in some roundabout way, they do conform to it. It is immediately obvious that when we see a thing to be good, its goodness is not a property which we can take up in our hands, or separate from it even by the most delicate scientific instruments, and transfer to something else. It is not, in fact, like most of the predicates which we ascribe to things, a part of the thing to which we ascribe it. But philosophers suppose that the reason why we cannot take goodness up and move it about, is not that it is a different kind of object from any which can be moved about, but only that it necessarily exists together with anything with which it does exist. They explain the type of ethical truths by supposing it identical with the type of scientific laws. And it is only when they have done this that the naturalistic philosophers proper—those who are empiricists—and those whom I have called metaphysical part company. These two classes of philosophers do, indeed, differ with regard to the nature of scientific laws. The former class tend to suppose that they mean only This has accompanied, does now, and will accompany that in these particular instances: they reduce the scientific law quite simply and directly to the familiar type of proposition which I have pointed out. But this does not satisfy the metaphysicians. They see that when you say This would accompany that, if that existed, you don't mean only that this and that have existed and will exist together so many times. But it is beyond even their powers to believe that what you do mean is merely what you say. They still think you must mean, somehow or other, that something does exist, since that is what you generally mean when you say anything. They are as unable as the empiricists to imagine that you can ever mean that 2+2=4. The empiricists say this means that so many couples of couples of things have in each case been four things; and hence that 2 and 2 would not make 4, unless precisely those things had existed. The metaphysicians feel that this is wrong; but they themselves have no better account of its meaning to give than either, with Leibniz, that God's mind is in a certain state, or, with Kant, that your mind is in a certain state, or finally, with Mr Bradley, that something is in a certain state. Here, then, we have the root of the naturalistic fallacy. The metaphysicians have the merit of seeing that when you say This would be good, if it existed, you can't mean merely This has existed and was desired, however many times that may have been the case. They will admit that some good things have not existed in this world, and even that some may not have been desired. But what you can mean, except that something exists, they really cannot see. Precisely the same error which leads them to suppose that there must exist a supersensible Reality, leads them to commit the naturalistic fallacy with regard to the meaning of good. Every truth, they think, must mean somehow that something exists; and since, unlike the empiricists, they recognise some truths which do not mean which do not mean that anything exists here and now, these they think must mean that something exists not here and now. On the same principle, since good is a predicate which neither does nor can exist, they are bound to suppose either that to be good means to be related to some other particular thing which can exist and does exist in reality; or else that it means merely to belong to the real world—that goodness is transcended or absorbed in reality. (§ 73 ¶ 2)

That such a reduction of all propositions to the type of those which assert either that something exists or that something which exists has a certain attribute (which means, that both exist in a certain relation to one another), is erroneous, may easily be seen by reference to the particular class of ethical propositions. For whatever we may have proved to exist, and whatever two existents we may have proved to be necessarily connected with one another, it still remains a distinct and different question whether what thus exists is good; whether either or both of the two existents is so; and whether it is good that they should exist together. To assert the one is plainly and obviously not the same thing as to assert the other. We understand what we mean by asking: Is this, which exists, or necessarily exists, after all, good? and we perceive that we are asking a question which has not been answered. In the face of this direct perception that the two questions are distinct, no proof that they must be identical can have the slightest value. That the proposition This is good is thus distinct from every other proposition was proved in Chapter I; and I may now illustrate this fact by pointing out how it is distinguished from two particular propositions with which it has commonly been identified. That so and so ought to be done is commonly called a moral law; and this phrase naturally suggests that the proposition is in some way analogous either to a natural law, or to a law in the legal sense, or to both. All three are, in fact, really analogous in one respect, and in one respect only: that they include a proposition which is universal. A moral law asserts This is good in all cases; a natural law asserts This happens in all cases; and a law, in the legal sense, It is commanded that this be done, or left undone, in all cases. But since it is very natural to suppose that the analogy extends further, and that the assertion This is good in all cases is equivalent to the assertion This happens in all cases or to the assertion It is commanded htat this be done in all cases, it may be useful briefly to point out that they are not equivalent. (§ 74 ¶ 1)

The fallacy of supposing moral law to be analogous to natural law in respect of asserting that some action is one which is always necessarily done is contained in one of the most famous doctrines of Kant. Kant identifies what ought to be with the law according to which a Free or Pure Will must act—with the only kind of action which is possible for it. And by this identification he does not mean merely to assert that Free Will is also under the necessity of doing what it ought; he means that what it ought to do means nothing but its own law—the law according to which it must act. It differs from the human will just in that, what we ought to do, is what it necessarily does. It is autonomous; and by this is meant (among other things) that there is no separate standard by which it can be judged: that the question Is the law by which this Will acts a good one? is, in its case, meaningless. It follows that what is necessarily willed by this Pure Will is good, not because that Will is good, nor for any other reason; but merely because it is what is necessarily willed by a Pure Will. (§ 75 ¶ 1)

Kant's assertion of the Autonomy of the Practical Reason thus has the very opposite effect to that which he desired; it makes his Ethics ultimately and hopelessly heteronomous. His Moral Law is independent of Metaphysics only in the sense that according to him we can knnow it independently; he holds that we can only infer that there is Freedom, from the fact that the Moral Law is true. And so far as he keeps strictly to this view, he does avoid the error, into which most metaphysical writers fall, of allowing his opinions as to what is real to influence his judgments of what is good. But he fails to see that on his view the Moral Law is dependent upon Freedom in a far more important sense than that in which Freedom depends on the Moral Law. He admits that Freedom is the ratio essendi of the Moral Law, whereas the latter is only the ratio cognoscendi of Freedom. And this means that, unless Reality be such as he says, no assertion that This is good can possibly be true: it can indeed have no meaning. He has, therefore, furnished his opponents with a conclusive method of attacking the validity of the Moral Law. If they can only shew by some other means (which he denies to be possible but leaves theoretically open) that the nature of Reality is not such as he says, he cannot deny that they will have proved his ethical principle to be false. If what This ought be done means This is willed by a Free Will, then, if it can be shewn that there is no Free Will which wills anything, it will follow that nothing ought to be done. (§ 75 ¶ 2)

And Kant also commits the fallacy of supposing that This ought to be means This is commanded. He conceives the Moral Law to be an Imperative. And this is a very common mistake. This ought to be, it is assumed, must mean This is commanded; nothing, therefore, would be good unless it were commanded; and since commands in this world are liable to be erroneous, what ought to be in its ultimate sense means what is commanded by some real supersensible authority. With regard to this authority it is, then, no longer possible to ask Is it righteous? Its commands cannot fail to be right, because to be right means to be what it commands. Here, therefore, law, in the moral sense, is supposed to be analogous to law, in the legal sense, rather than, as in the last instance, to law in the natural sense. It is supposed that moral obligation is analogous to legal obligation, with this difference only that whereas the source of legal obligation is earthly, that of moral obligation is heavenly. Yet it is obvious that if by a source of obligation is meant only a power which binds you or compels you to do a thing, it is not because it does do this that you ought to obey it. It is only if it be itself so good, that it commands and enforces only what is good, that it can be a source of moral obligation. And in that case what it commands and enforces would be good, whether commanded and enforced or not. Just that which makes an obligation legal, namely the fact that it is commanded by a certain kind of authority, is entirely irrelevant to moral obligation. However an authority be defined, its commands will be morally binding only if they are—morally binding; only if they tell us what ought to be or what is a means to that which ought to be. (§ 76 ¶ 1)