§ 66 Precedents Constitutive, Not Abrogative
We have already seen the falsity of the theory that all precedents are declaratory. We have seen that they possess a distinct and legally recognised law-creating power. This power, however, is purely constitutive and in no degree abrogative. Judicial decisions may make law, but they cannot alter it, for where there is settled law already on any point the duty of the judges is to apply it without question, and they have no authority to substitute for it law of their own making. Their legislative power is strictly limited to supplying the vacancies of the legal system, to filling up with new law the gaps which exist in the old, to supplementing the imperfectly developed body of legal doctrine.
This statement, however, requires two qualifications. In the first place, it must be read subject to the undoubted power of the courts to overrule or disregard precedents in the manner already described. In its practical effect this is equivalent to the exercise of abrogative power, but in legal theory it is not so. The overruling of a precedent is not the abolition of an established rule of law; it is an authoritative denial that the supposed rule of law has ever existed. The precedent is so treated not because it has made bad law, but because it has never in reality made any law at all. It has not conformed to the requirements of legal efficacy. Hence it is that the overruling of a precedent, unlike the repeal of a statute, has retrospective operation. The decision is pronounced to have been bad ab initio. A repealed statute, on the contrary, remains valid and applicable as to matters arising before the date of its repeal. The overruling of a precedent is analogous not to the repeal of a statute, but to the judicial rejection of a custom as unreasonable or as otherwise failing to conform to the requirements of customary law.
In the second place, the rule that a precedent has no abrogative power must be read subject to the maxim, Quod fieri non debet, factum valet. It is quite true that judges ought to follow the existing law whenever there is any such law to follow. They are appointed to fulfil the law, not to subvert it. But if by inadvertence or otherwise this rule is broken through, and a precedent is established which conflicts with pre-existing law, it does not follow from this alone that this decision is destitute of legal efficacy. For it is a well-known maxim of the law that a thing which ought not to have been done may nevertheless be valid when it is done. If, therefore, a precedent belongs to the class which is absolutely authoritative, it does not lose this authority simply because it is contrary to law and ought not to have been made. No court, for example, will be allowed to disregard a decision of the House of Lords on such a ground; it must be followed without question, whether it is in harmony with prior law or not. So also with those which are merely conditionally authoritative. We have already seen that error is only one of two conditions, both of which are requisite to render allowable the disregard of such a precedent, and in this respect it makes no difference whether the error consists in a conflict with law or in a conflict with reason. It may well be better to adhere to the new law which should not have been made than to recur to the old law which should not have been displaced.