There is another line of argument, more subtle but more instructive, by which we can refute those who say that good means desired, or who propose any other idea, such as pleasure, as the actual meaning of good. This line of argument will not prove that the things that are good are not the same as the things that are desired; but it will prove that, if this were the case, it could not be proved by appealing to the meaning of the word good. So far, it might be thought that such an argument could only have purely logical importance. But in fact this is not so. Many ethical theories have been based upon the contention that good means so-and-so, and people have accepted consequences of this contention, which, if they had relied upon inspection untrammelled by false theory, they would almost certainly have rejected. Whoever believes that good means desired will try to explain away the cases where it seems as if what is desired is bad; but if he no longer holds this theory, he will be able to allow free play to his unbiased ethical perceptions, and will thus escape errors into which he would otherwise have fallen. (§ 7 ¶ 1)

The argument in question is this: If anyone affirms that the good is the desired, we consider what he says, and either assent or dissent; but in any case our assent or dissent is decided by considering what the good and the desired really are. When, on the contrary, someone gives us a definition of the meaning of a word, our state of mind is quite different. If we are told a pentagon is a figure which has five sides, we do not consider what we know about pentagons, and then agree or disagree; we accept this as the meaning of the word, and we know that we are getting information, not about pentagons, but merely about the word pentagon. What we are told is the sort of thing that we expect dictionaries to tell us. But when we are told that the good is the desired, we feel at once that we are being told something of philosophical importance, something which has ethical consequences, something which is quite beyond the scope of a dictionary to tell us. The reason of this is, that we already know what we mean by the good, and what we mean by the desired; and if these two meanings always applied to the same objects, that would not be a verbal definition, but an important truth. The analogue of such a proposition is not the above definition of a pentagon, but rather: A pentagon (defined as above) is a figure which has five angles. Whenever a proposed definition sets us thinking whether it is true in fact, and not whether that is how the word is used, there is reason to suspect that we are not dealing with a definition, but with a significant proposition, in which the word professedly defined has a meaning already known to us, either as simple or as defined in some other way. By applying this test, we shall easily convince ourselves that all hitherto suggested definitions of the good are significant, not merely verbal, propositions; and that therefore, though they may be true in fact, they do not give the meaning of the word good. (§ 7 ¶ 2)

The importance of this result is that so many ethical theories depend upon the denial of it. Some have contended that good means desired, others that good means pleasure, others again that it means conformity to Nature or obedience to the will of God. The mere fact that so many different and incompatible definitions have been proposed is evidence against any of them being really definitions; there have never been two incompatible definitions of the word pentagon. None of the above are really definitions; they are all to be understood as substantial affirmations concerning the things that are good. All of them are, in my opinion, mistaken in fact as well as form, but I shall not here undertake to refute them severally. (§ 7 ¶ 3)