(d) So much, then, for moral rules or laws, in the ordinary sense—rules which assert that it is generally useful, under more or less common circumstances, for everybody to perform or omit some definite kind of action. It remains to say something with regard to the principles by which the individual should decide what he ought to do, (α) with regard to those actions as to which some general rule is certainly true, and (β) with regard to those where such a certain rule is wanting. (§ 99 ¶ 1)

(α) Since, as I have tried to shew, it is impossible to establish that any kind of action will produce a better total result than its alternative in all cases, it follows that in some cases the neglect of an established rule will probably be the best course of action possible. The question then arises: Can the individual ever be justified in assuming that his is one of these exceptional cases? And it seems that this question may be definitely answered in the negative. For, if it is certain that in a large majority of cases the observance of a certain rule is useful, it follows that there is a large probability that it would be wrong to break the rule in any particular case; and the uncertainty of our knowledge both of effects and of their value, in particular cases, is so great, that it seems doubtful whether the individual’s judgment that the effects will probably be good in his case can ever be sent against the general probability that that kind of action is wrong. Added to this general ignorance is the fact that, if the question arises at all, our judgment will generally be biased by the fact that we strongly desire one of the results which we hope to obtain by breaking the rule. It seems, then, that with regard to any rule which is generally useful, we may assert that it ought always to be observed, not on the ground that in every particular case it will be useful, but on the ground that in any particular case the probability of its being so is greater than that of our being likely to decide rightly that we have before us an instance of its disutility. In short, though we may be sure that there are cases where the rule should be broken, we can never know which those cases are, and ought, therefore, never to break it. It is this fact which seems to justify the stringency with which moral rules are usually enforced and sanctioned, and to give a sense in which we may accept as true the maxims that The end never justifies the means and That we should never do evil that good may come. The means and the evil, intended by these maxims, are, in fact, the breaking of moral rules generally recognised and practised, and which, therefore, we may assume to be generally useful. Thus understood, these maxims merely point out that, in any particular case, although we cannot clearly perceive any balance of good produced by keeping the rule and do seem to see one that would follow from breaking it, nevertheless the rule should be observed. It is hardly necessary to point out that this is so only because it is certain that, in general, the end does justify the means in question, and that therefore there is a probability that in this case it will do so also, although we cannot see that it will. (§ 99 ¶ 2)

But moreover the universal observance of a rule which is generally useful has, in many cases, a special utility, which seems deserving of notice. This arises from the fact that, even if we can clearly discern that our case is one where to break the rule is advantageous, yet, so far as our example has any effect at all in encouraging similar action, it will certainly tend to encourage breaches of the rule which are not advantageous. We may confidently assume that what will impress the imagination of others will not be the circumstances in which our case differs from ordinary cases and which justify our exceptional action, but the points in which it resembles other actions that are really criminal. In cases, then, where example has any influence at all, the effect of an exceptional right action will generally be to encourage wrong ones. And this effect will probably be exercised not only on other persons but on the agent himself. For it is impossible for any one to keep his intellect and sentiments so clear, but that, if he has once approved of a generally wrong action, he will be more likely to approve of it also under other circumstances than those which justified it in the first instance. This inability to discriminate exceptional cases offers, of course, a still stronger reason for the universal enforcement, by legal or social sanctions, of actions generally useful. It is undoubtedly well to punish a man, who has done an action, right in his case but generally wrong, even if his example would not be likely to have a dangerous effect. For sanctions have, in general, much more influence upon conduct than example; so that the effect of relaxing them in an exceptional case will almost certainly be an encouragement of similar action in cases which are not exceptional. (§ 99 ¶ 3)

The individual can therefore be confidently recommended always to conform to rules which are both generally useful and generally practised. In the case of rules of which the general observance would be useful but does not exist, or of rules which are generally practised but which are not useful, no such universal recommendations can be made. In many cases the sanctions attached may be decisive in favor of conformity to the existing custom. But it seems worth pointing out that, even apart from these, the general utility of an action most commonly depends upon the fact that it is generally practised: in a society where certain kinds of theft are the common rule, the utility of abstinence from such theft on the part of a single individual becomes exceedingly doubtful, even though the common rule is a bad one. There is, therefore, a strong probability in favour of adherence to an existing custom, even if it be a bad one. But we cannot, in this case, assert with any confidence that this probability is always greater than that of the individual’s power to judge that an exception will be useful; since we are here supposing certain one relevant fact—namely, that the rule, which he proposes to follow, would be better than that which he proposes to break, if it were generally observed. Consequently the effect of his example, so far as it tends to break down the existing custom, will here be for the good. The cases, where another rule would certainly be better than that generally observed, are, however, according to what was said above, very rare; and cases of doubt, which are those which arise most frequently, carry us into the next division of our subject. (§ 99 ¶ 4)