I doubt these claims. But it is worth considering what they would imply. According to C, each of us should try to have one of the best possible sets of desires and dispositions, in Consequentialist terms. It might make the outcome better if we did not merely have these desires and dispositions, but had corresponding moral emotions and beliefs.

Consider, for example, theft. On some versions of C, it is intrinsically bad if property is stolen. On other versions of C, this is not so. On these versions, theft is bad only when it makes the outcome worse. Avoiding theft is not part of our ultimate moral aim. But it might be true that it would make the outcome better if we were strongly disposed not to steal. And it might make the outcome better if we believed stealing to be intrinsically wrong, and would feel remorse when we do steal. Similar claims might be made about many other kinds of act.

If these claims were true, C would be self-effacing. It would tell us that we should try to believe, not itself, but some other theory. We should try to believe the theory which is such that, if we believed it, the outcome would be best. On the claims made above, this theory might not be C. It might be some version of what Sidgwick called Common-Sense Morality.

If C told us to believe some version of this morality, this would not be Common-Sense Morality as it is now, but an improved version. Common-Sense Morality is not the moral theory belief in which would make the outcome best. Such a theory would, for example, demand much more from the rich. It might make the outcome best if those in the richer nations gave to the poor at least a quarter or even half of their incomes every year. The rich now give, and seem to believe that they are justified in giving, less than one per cent.

Suppose that C told us to believe some other theory. As I have said, it would be hard to change our beliefs, if our reason for doing so is not a reason which casts doubt on our old beliefs, but is merely that it would have good effects if we had different beliefs. But there are various ways in which we might bring about this change. Perhaps we could all be hypnotized, and the next generation brought up differently. We would have to be made to forget how and why we acquired our new beliefs, and the process would have to be hidden from future historians.

It would make a difference here if we accept, not C, but Collective Consequentialism. If we accept C, we might conclude that C ought to be rejected by most people, but should still be believed by a few. Our theory would then be partly self-effacing, and partly esoteric, telling those who believe it not to enlighten the ignorant majority. But as Collective Conse-quentialists we should believe the moral theory which is such that, if we all believed it, the outcome would be best. This theory cannot be esoteric.

Some find it especially objectionable that a moral theory might be esoteric. If we believe that deception is morally wrong, deception about morality may seem especially wrong. Sidgwick wrote: 'it seems expedient that the doctrine that esoteric morality is expedient should itself be kept esoteric. Or, if this concealment be difficult to maintain, it may be desirable that Common Sense should repudiate the doctrines which it is expedient to confine to an enlightened few.' This is what Williams calls 'Government House' Consequentialism, since it treats the majority like the natives in a colony. As Williams claims, we cannot welcome such a conclusion. Sidgwick regretted his conclusions, but he did not think regret a ground for doubt.

I have claimed that it is unlikely that C is wholly self-effacing. It would at most be partly self-effacing and partly esoteric. It might make the outcome better if some people did not believe C; but it is unlikely that it would make the outcome better if C was believed by no one.

Here is another ground for doubting this. Suppose that we all come to believe C. (This will seem less implausible when we remember that C can be a pluralist theory, appealing to many different moral principles.) We then decide that C is wholly self-effacing. We decide that it would make the outcome best if we caused ourselves to believe some improved version of Common-Sense Morality. We might succeed in bringing about this change in our beliefs. Given changes in the world, and in our technology, it might later come to be true that the outcome would be better if we revised our moral beliefs. But if we no longer believed C, because we now believed some version of Common-Sense Morality, we would not be led to make these needed revisions in our morality. Our reason for believing this morality would not be that we now believe it to be the morality belief in which would make the outcome best. This would be why we caused ourselves to believe this morality. But, in order to believe this morality, we must have forgotten that this is what we did. We would now simply believe this morality. We might therefore not be led to revise our morality even if it came to be true that our belief in this morality would increase the chances of some disaster, such as a nuclear war.

These claims should affect our answer to the question whether it would make the outcome better if we all ceased to believe C. We might believe correctly that there is some other moral theory belief in which would, in the short run, make the outcome better. But once Consequentialism has effaced itself, and the cord is cut, the long-term consequences might be much worse.

This suggests that the most that could be true is that C is partly self-effacing. It might be better if most people caused themselves to believe some other theory, by some process of self-deception that, to succeed, must also be forgotten. But, as a precaution, a few people should continue to believe C, and should keep convincing evidence about this self-deception. These people need not live in Government House, or have any other special status. If things went well, the few would do nothing. But if the moral theory believed by most did become disastrous, the few could then produce their evidence. When most people learnt that their moral beliefs were the result of self-deception, this would undermine these beliefs, and prevent the disaster.

Though I have claimed that this is unlikely, suppose that C was wholly self-effacing. Suppose that it told all of us to make ourselves believe, not itself, but some other theory. Williams claims that, if this is so, the theory ceases to deserve its name, since it 'determines nothing of how thought in the world is conducted'. This claim is puzzling since, as Williams also claims, C would be demanding that the way in which we think about morality, and our set of desires and dispositions, 'must be for the best'. This is demanding something fairly specific, and wholly Consequentialist.

Williams makes the third claim that, if C was wholly self-effacing, it would cease to be effective. This need not be so. Suppose that things happen as described above. We all come to believe in some form of C. We then believe truly that, if we all believed some other theory, this would produce the best possible outcome. C tells us all to believe this other theory. In some indirect way, we bring about this change in our beliefs. No one now believes C. But this does not justify the claim that C has ceased to be effective. It has had the effect that we all now believe some other particular theory. And our belief in this other theory will produce the best possible outcome. Though no one : believes C, C is still effective. There are two continuing facts that are the effects of our earlier belief in C: our new moral beliefs, and the fact that, because we have these beliefs, the outcome is as good as it can possibly be.

Williams rightly claims that, if C was wholly self-effacing, it would not be clear what this shows. We would have to decide whether it showed that C 'is unacceptable, or merely that no one ought to accept it.' It is clear that, on our last assumptions, no one ought morally to accept C. If anyone did accept C, it would itself tell him that he ought morally to try to reject C and instead believe some other theory. But, as Williams suggests, there are two questions. It is one question whether some theory is the one that we ought morally to try to believe. It is another question whether this is the theory that we ought intellectually or in truth-seeking terms to believe—whether this theory is the true or best justified theory. I claimed earlier that, if a theory about rationality was self-effacing, this would not show that this theory cannot be the true or the best justified theory. Can we make a similar claim about moral theories?

Our answer to this question will depend in part on our beliefs about the nature of moral reasoning. If a moral theory can be quite straightforwardly true, it is clear that, if it is self-effacing, this does not show that it cannot be true. But we may instead regard morality as a social product, either actually or in some 'ideal constructivist' way. We may then claim that, to be acceptable, a moral theory must fulfil what Rawls calls 'the publicity condition': it must be a theory that everyone ought to accept, and publicly acknowledge to each other. On these meta-ethical views, a moral theory cannot be self-effacing. On other views, it can be. It would take at least a book to decide between these different views. I must therefore, in this book, leave this question open. This does not matter if, as I believe, C would not be self-effacing.