So far we may regard Kant as having laid down in the most impressive way the principles which must form the basis of every constructive ethical system^[80] But in Kant's own view these positions are associated with two other doctrines which require further examination. In the first place he assumed that out of this bare idea of a categorical imperative, without any appeal to experience, he could extract a moral criterion, i.e. that he could ascertain what is the actual content of the Moral Law, what in detail it is right to do. Secondly, he assumed that, so far as an act is not determined by pure respect for the Moral Law, it possesses no moral value whatever. Let us examine each of these positions in turn.(Bk. 1 Ch. 5 § 2 ¶ 1)

The value of Kant's work consisted very largely in supplying a metaphysical basis for Ethics. So long as it is assumed that all our ordinary knowledge of matters of fact comes from experience of an external world, there is always a sort of suspicion that any kind of knowledge which cannot point to such an origin must be in some sense unsubstantial or delusive. The Critique of Pure Reason demonstrates that in all our knowledge there is an element which is not derived from experience: all knowledge implies forms of perception and forms of understanding which are a priori, part of the constitution of the mind itself, not supplied to it from without. The matter of sensation is from without, but sense by itself is not thought. I cannot judge of the size and distance of particular objects without a matter supplied by sensible perception: but I could not build up these data into the conception of a square table of a certain size unless I had already notions of space, of spacial and causal relations, of substance and accident and the like which do not come from experience^[81]. In all actual knowledge there must be a matter supplied by experience and a formal element which is a priori. But in Morality--in the idea of duty--we are presented with a form which needs no filling up from experience, a form which is (so to speak) its own content, since it is a matter of immediate consciousness that this a priori concept of duty can supply a motive to the will. Now in this position a very important truth is (as is almost universally admitted by the most Kantian of modern Moralists) confused with a very serious error. That no experience can prove an act to be right, that no accumulation of knowledge as to what is can possibly give us an ought, is a truth which can only be denied by asserting that there is no meaning in duty or in Morality. Experience of the past may tell us what has been or what will be: it cannot possibly tell us what ought to be. That which ought to be is ex vi termini something which as yet is not and which may conceivably never be. In that sense our moral judgements are undoubtably a priori or independent of experience. But that without any appeal to experience we can get at the content as well as the form of the moral law, can easily be shown to be a pure delusion. Let us see how Kant made the attempt.(Bk. 1 Ch. 5 § 2 ¶ 2)

The rules of action which the categorical imperative is supposed to give us are the following:--(Bk. 1 Ch. 5 § 2 ¶ 3)

1. So act as if the law of thine action were to become by thy will law universal.(Bk. 1 Ch. 5 § 2 ¶ 4)

2. Regard humanity whether in thine own person or in that of any one else always as an end and never as a means only.(Bk. 1 Ch. 5 § 2 ¶ 5)

3. Act as a member of a kingdom of ends^[82].(Bk. 1 Ch. 5 § 2 ¶ 6)

Let us examine the first of these rules--Act as if the law of thy action were to become by thy will law universal. Now it is quite true that it does follow from the very idea of there being something which is right to do irrespectively of inclination that this ocurse must, in the same circumstances, be binding upon every one else. And therefore in a sense it is true that no action can be really a moral rule the principle of which could not be universalized. It is good practical advice to urge that when we have to pronounce upon the morality of a proposed act we should ask ourselves whether it represents a principle which we should think it rational to will as a universal rule of conduct. But this is by itself a merely negative test. It gives us no definite information until we have made up our minds as to what it is which makes conduct rational or irrational. We can, indeed, with a little ingenuity extract from it the all-important axioms of Benevolence and Equity: for, if there is something which it is intrinsically right to do, what is right for me would be right for any one else in the same circumstances^[83]: hence it must be right for me to treat every other man as it would be right for him to treat me under similar circumstances. If my good is recognized as something which it is intrinsically right for others to promote, the good of each other individual must also be treated as an end the promotion of which I must look upon as incumbent upon me: hence I am bound to promote the greatest good of humanity collectively (the maxim of rational Benevolence), and to treat each individual's good as of equal value with the good of every other (the maxim of Equity). But these rules by themselves will give us no practical guidance till we know what that good is which ought to be promoted by every rational being for every other.(Bk. 1 Ch. 5 § 2 ¶ 7)

The Kantian maxim, properly interpreted, thus occupies in Ethics the same position which the law of contradiction holds in Logic^[84]. The law of contradiction is a negative test of truth: it tells us that two judgements which contradict one another cannot both be true, but as to which judgements in particular are true, it will give us no information: only, when I know that judgement A is true, it will tell me that judgement B, being inconsistent therewith, cannot also be true. In the same way the Kantian rule tells us that a genuine ultimate rule of conduct must not only be logically consistent with itself, but also be such as that all its prescriptions shall be consistent with all other ethical rules. The supreme ethical precept must consist of an harmonious and self-consistent system of precepts. It need hardly be said that this by itself is a most important negative test of ethical truth. It gives us the principle upon which alone inference or reasoning (as distinct from immediate judgements of Reason) is possible in Ethics. The fact that something is a part of the true ethical rule supplies, if we assume this principle to be self-evident, a demonstrative proof that some precept inconsistent with it cannot be a part of it^[85]. But as to what rule of action in particular is reaonsable, it gives us no information whatever. If we interpret the rule of acting on a principle fit for law universal as equivalent to Sidgwick's three ethical axioms--of Benevolence, Equity and Prudence--we shall get rules for the promotion and distribution of the good or ultimate end, but no information as to what particular things are good: and, until we know that, we cannot get any principles from which we can deduce the right course of cnoduct in any one single case. If with Sidgwick (who could quote much in Kant himself to support this interpretation) we made good in this connexion equivalent to pleasure, and interpreted our rule to mean promote universal pleasure and distribute it equally, we should obviously have gone beyond the mere a priori formal rule. We should have appealed to experience--an appeal which our categorical imperative was intended by Kant to exclude. The judgement Happiness ought to be promoted is no doubt in a sense a priori, but not in the sense that no information derived from experience is necessary to its being made. Kant himself admits that the concept of happiness is of empirical origin^[86]. Experience must tell us what happiness is before we can judge happiness to be good. Still more obviously experience is wanted to tell us what particular goods constitute happiness, or what are the means to procure those goods. It might be thought that Kant could get a constant for the Moral Law by holding that the true good of man is simply Morality, a concept which might be said to be of purely a priori origin, and that we should find out what particular actions are right by considering what actions would promote universal Morality. But here again, if the concept of the end is in a sense purely a priori, experience is needed to tell us the means; and Kant has incapacitated himself from adopting this solution of the problem by the exaggerated Libertarianism which made him pronounce an action due to another's influence to be not truly free, nad therefore without moral value^[87]. Consequently, he pronounced that it was impossible for one man to make another's moral good his end. Hence if Virtue is by itself to constitute the end, it must be the man's own virtue that he must treat as his end. To tell a man to make his own virtue an end will not tell him what to do until he knows what acts it is virtuous to perform, and as to this formula that what is right for him is right for others will give him no information whatever. How then did Kant attempt to extract out of the bare form of the Moral Law a knowledge of the particular actions which are right or wrong?(Bk. 1 Ch. 5 § 2 ¶ 8)

It is impossible to maintain that Kant gives a clear and consistent meaning to his own dictum. Sometimes the irrationality of willing the universal adoption of the immoral course appears to turn simply upon the fact that the social consequences to which the adoption of such a will would lead are consequences which no rational man could regard as good. We cannot will universal promise-breaking because in that case no promises would be made, and at times the irrationality of willing such a consequence seems to turn upon its injurious social effects. Still more clearly when Kant pronounces that we cannot rationally will the non-development of our faculties, the irrationality of such a course is madee to depend simply upon the fact that the rational man actually regards this non-development as bad and their development as good^[88]. Here the appeal to consequences which can only be known by experience is scarcely disguised: the a priori judgement relates simply to the goodness or badness of the end. But Kant was able to conceal from himself the necessity of this appeal to experience, because in certain carefully selected instances he was able to point to the appearance of internal contradiction in the reverse of the accepted rule^[89]. We cannot rationally will that men shall break their promises, because in that case no promises would be made; and we cannot rationally will something to be done which will make it impossible to observe the very rule which we will. In a society in which there were no promises, it would no longer be possible to observe our proposed rule of universal promise-breaking; if no promises are made, none can be broken. Now even here it is evident that Kant falls back upon his experience of human nature to tell him what will be the consequences of his act: but still he might maintain that, given this much experience, the contradiction is self-evident. Yet it is easy to show that absence of contradiction, in this sense, would be a very irrational test of conduct. Kant himself appears to concede that there would be no internal contradiction in willing that all men should leave their faculties undeveloped. Nor would there be any internal contradiction in adopting as our rule of action the promotion of universal misery, or at least the maximum of misery which should be consistent with the continued survival of the human race. That is, indeed, according to some Pessimists, precisely the end which is actually realized in the world as we know it.(Bk. 1 Ch. 5 § 2 ¶ 9)

And, just as we hold many acts to be wrong which involve no internal contradiction, so there are many things which we pronounce right in spite of such contradiction. Kant tells us that we cannot rationally will universal promise-breaking, because the universal adoption of such a rule would lead to a state of things in which the rule Break your promises could no longer be observed. We must not commit suicide, because if every one did so, there would soon be nobody left to practise the virtue of suicide. Then are we, it may be asked, to deny that Philanthropy is a duty because the universal practice of a reasonable Philanthropy would lead to a state of things in which there would be no poor upon whom to practise that virtue? Shall we refuse to bless the peacemaker, because if every one shared his disposition, there would be no quarrels to adjust? And then, again, how unreasonable is the alternative with which we are presented--either to will universal suicide and universal lying, or to forbid these practices in any circumstances whatever! As reasonably might we pronounce Kant's own celibacy a crime because universal celibacy would rapidly extinguish the human race and (consequently) the practice of celibacy.(Bk. 1 Ch. 5 § 2 ¶ 10)

It is true that the emergence of an internal contradiction (in Kant's sense) in any suggested moral rule does show that we have not reached an ultimate principle of conduct. We can, indeed, put such rules as Give to the poor into a universal form by making them hypothetical: So long as there are any poor, relieve them; but so might we say, So long as there are any human beings alive, let them commit suicide. Still, the fact that the rule is only applicable to a particular set of circumstances does show that we have not reached an ultimate principle. The rule, Be charitably disposed, may, indeed, be universally willed: but then Kant's object in applying his test of fitness for law universal is to supply a guide for the details of outward conduct, not for mere dispositions and intentions, and this purpose is not served by such generalities as these. And even in this case there is really a reference to the physical constitution of human beings which is known to us only from experience. We might interpret charity to mean a disposition to promote good, but the absence of internal contradiction will not tell us what good is. Moreover, as has already been pointed out, although an ultimate moral principle must be free from internal contradiction, it is impossible to deny that many immoral principles might very well be universalized without leading to any such contradiction. The structure of the Universe and of human nature is quite as consistent with the non-development as with the development of human faculties. And if the criterion is not of universal application, how are we to know when to apply it, and when not?(Bk. 1 Ch. 5 § 2 ¶ 11)

The fact is that Kant appears to have confused two distinct senses of the term categorical. When he sets forth that it is of the essence of every moral law to be categorical, he means that it must admit of no exception due to the subjective disinclination of the individual for the course of action which it prescribes. We must not say, I admit Temperance or Veracity to be right in a general way: only I personally happen to have such a rooted antipathy to Temperance or Veracity, or whatever it be, that I must regard myself as an exception to the general rule. To talk in that way no doubt destroys the very nature of a Moral Law. It is an essential characteristic of the Moral Law that whatever is right for me must be right for every man in precisely the same circumstances^[90]. But when Kant tries to make out this mere unconditionality of a rule an absolute test of its reasonableness, he has to assume that the categorical character of an imperative excludes the possibility of an exception based not on the mere subjective disinclination of the individual, but on the nature of the case. He does not see that the rule Do this except in such and such circumstances is just as categorical and just as little hypothetical as the rule Do this under all circumstances whatever, so long as the exceptions are recognized as no less universal in their application, no less based upon the reason and nature of things, than the original rule. Kant in fact confuses the inclusion of an exception in a moral rule with the admission of an exception to a moral rule. He does not recognize that the difference between a rule with an exception and a grammatically categorical rule is often a purely verbal one. The precept Do no murder admits of no exceptions, because murder means killing except in such and such circumstances. The rule Thou shalt not kill has exceptions. So the rule Lie not could be represented as equally categorical if there were as clear a usage in favour of the proposition that a legitimate untruth is no lie, as there is in favour of the proposition that in certain circumstances killing is no murder. We are obliged sometimes to express a moral rule in the form of a general command with an exception simply because the enumeration of the circumstances to which the rule is inapplicable is shorter and more convenient than an exhaustive enumeration of all the cases to which it is applicable. And it is clear that every rule, however general, implies some set of circumstances in which alone it is capable of being applied. The duty of not committing adultery is only applicable t othe relations between two persons of whom one at least has a lawful spouse, and it is obvious that this term lawful postulates a larger number of highly complicated social arrangements, about which there is by no means a universal consensus, and which the most enthusiastic Kantian could hardly attempt to determine on any a priori principle. Either, then, we must say that every possible rule really involves a hypothesis under which alone it is applicable; or we may say that every moral law excludes all exception if only you put it into a sufficiently general, and a sufficiently internal, form. Kill not has exceptions: Thou shalt love thy neighbour as thyself (properly understood) has none^[91]. But, in whichever way it is put, it is plain that we can get no criterion of Morality out of the presence or absence of exceptions. Kill not has exceptions, and yet (subject to the exceptions) is accounted a good moral principle. On the other hand, Thou shalt love thy friend and hate thine enemy does not appeal to us as the highest morality, in spite of its being as categorical as the Christian precept.(Bk. 1 Ch. 5 § 2 ¶ 12)

Kant's attempt to extract an ethical criterion out of the bare form of the Moral Law is the more remarkable, because he did not hold (as he is sometimes supposed to do) that there is no other rational end of action except the bare performance of duty. Had he held that view, it would have become fairly impossible for him even to have persuaded himself that he had discovered in the bare form of the law any content for the idea of duty^[92]. If a man is to perform his duty, he must know what that duty is; and the mere knowledge that, when he has discovered what his duty is, it is a thing categorically commanded does not help to find out what it is. It is impossible, in short, to show the rationality of one course of action rather than another until we have admitted that something else besides the performance of duty--some objective good other than the state of the will--is a rational end of action or possesses value^[93]. And Kant did admit that there is such another rational end of action--which possesses worth, not indeed absolutely and unconditionally, but on one condition--that it does not interfere with Virtue. And that other end is Happiness. From this position it would seem logically to follow that the true criterion would be the tendency of an action to promote for all mankind Happiness in so far as is compatible with Virtue. This would supply us with a quite intelligible and workable view of the moral criterion, and it would correspond roughly with the actual deliverances of the moral consciousness. That it is an inadequate view of the ultimate end of human life, I have already attempted to show; and its deficiencies will be further illustrated when we pass on to the other mistaken assumption, from which I am anxious to dissociate Kant's fundamental doctrine of a categorical imperative.(Bk. 1 Ch. 5 § 2 ¶ 13)