question are not self-evident, or that the arguments used to support them are superfluous.

It is by the consideration of such fallacies as these that I have been induced to use the word "ultimate," when the expression "à priori" might appear the most natural. A priori" means independent of experience, but "independent of experience" is ambiguous. It may mean either that experience has not produced the judgment in question, or that it furnishes no grounds for believing it. The first meaning is quite beside the purpose; philosophy has no direct concern with the origin of beliefs, which, as before stated, is part of the subject-matter of psychology. The second meaning, on the other hand, while it excludes experience as a ground of belief, and so far expresses the desired idea, does not express the full differentia of ultimate beliefs, viz., that there are no grounds for believing them at all. On the contrary, it sometimes seems to suggest itself directly as a reason for accepting a judgment (as if the fact that experience did not prove anything was a ground for believing it), and sometimes mediately, as showing that the constitution of our mind when in a healthy condition impels us to believe it or that it was implanted in us by the Author of our being; which reasons, whether good or bad, show by the very fact that they are given as reasons, that the judgment called "à priori" is not ultimate.

While, then, it is evidently not the business of philosophy to account for ultimate axioms and modes of inference, it is also clear (though it may be hardly necessary to make the remark) that it is not its business to prove them. To prove any conclusion is to show that it legitimately follows from a true premiss; so that if we were obliged to perform this operation for our axioms and modes of inference before they were to be received as ultimate, we should be driven either to argue in a circle or to an infinite regress. Indeed, this will sufficiently appear if we reflect that all we mean by ultimate is "independent of proof".

But if philosophy is neither to investigate the causes nor to prove the grounds of belief, what, it may be asked, is it to do? Its business, as I apprehend it, is to disengage them, to distinguish them from what simulates to be ultimate, and to exhibit them in systematic order.

What is meant here by disengaging the grounds of belief in contradistinction to proving them, will appear more clearly if we consider what is done by deductive logic. Deductive logic, apart from the practical rules with which it is encumbered, is (according to the terminology here employed) neither an art nor a science, but a systematic account of an ultimate mode of inference by which it may be distinguished from all other modes,

whether legitimate or illegitimate, whether ultimate or derivative: it is therefore by definition a branch of philosophy.

Now when deductive logic says that any three propositions which can be reduced to the form "All A is B, all C is A; all C is B," are legitimately connected as premisses and conclusion, whatever may be their content, it is by no means intended that such pieces of reasoning derive their validity from the fact of their corresponding with the formula. What is meant is simply to distinguish and mark off a certain mode of inference by giving a general description of it; each particular example of such inference being in itself the witness of its own validity.

This example explains the procedure of philosophy with regard to inferences: the axioms of mathematics furnish an illustration of its procedure in the matter of ultimate principles. "240 pence and 20 shillings being each equal to a pound, are equal to one another," is one of an indefinite number of similar self-evident propositions, which are described by saying that "things which are equal to the same thing are equal to one another "; but which do not require to be deduced from such general description in crder to make them certain. Such a deduction is, no doubt, possible. I may, if I please, say: "Things which are equal, &c."; "240 pence and 20 shillings are things which are equal, &c."; "therefore they are equal to each other". But such a syllogism would be as frivolous as Mill supposes all syllogisms to be; and for this reason, viz., that the conclusion is quite as obvious and certain as the premiss which is introduced to prove it.

It is conceivable, of course, that the axioms at the basis of knowledge are incapable of classification; that no two of them have anything in common except the fact that they are ultimate. In such an event the business of philosophy will be to enumerate, instead of describing them. But this can hardly be the case with modes of inference. The philosophy of deduction is already comparatively speaking, complete; and though the same cannot be said of any other mode of inference, it is difficult to believe that the bond connecting premisses and conclusion differs in every case, so as to exclude the possibility of classification. Something very distantly approaching this state of things would exist if each department of knowledge had a mode of reasoning peculiar to itself, as some have supposed (e.g.) theology to have.

To classify inferences is to exhibit what is called their common form. And it is plain that if of two inferences, which by classification have the same form, one is false and the other true, the classification which connects them is philosophically worthless. There would be no use in deductive logic, for instance, if some syllogisms in "Barbara" were trustworthy and others not.

It follows from this very obvious remark that every kind of logic if it is to be philosophical must be formal. The whole object of a philosophy of inference being to distinguish valid and ultimate inferences from those which are invalid or derivative, this can only be done either by exhibiting the common form or forms of such inferences, or (on the violent hypothesis that they have no common forms) by enumerating every concrete instance. To enunciate a form of inference which shall include both valid and invalid examples, can at best only have a psychological interest; philosophically, it is misleading.

The same remark applies mutatis mutandis to any classification of ultimate propositions.

There is no ground "à priori" (i.e., following from the idea of a philosophy) for supposing that ultimate judgments are general, rather than particular. Of course if they are the latter, there must be some legitimate mode of reasoning from particulars without the help of general propositions

Neither would I venture to assert that they must be certain. To say that our ultimate grounds of belief may be merely probable, will appear a paradox to some, and a truism to others. To me it seems to express a bare possibility. For there are these three remarks to be made on it:-1st. That the desire of certainty being the very thing which impels us to seek a philosophy, mere probability can never thoroughly satisfy our inquiries. 2nd. That, as a matter of fact, it will be found, I think, that no merely probable judgment is ever regarded as ultimate; nobody says of any judgment "There are no grounds whatever for believing this, indeed none are required, but I think it probable". 3rd. That since, according to received doctrines, which for the moment I assume to be true, the probability of any conclusion diminishes rapidly with the number of probable premisses required to prove it, if many of our ultimate premisses are merely probable, anything remotely approaching certainty for ordinary knowledge will be out of the question. So that those who aspire to regulate their convictions according to reason, will have to modify considerably their ordinary attitude towards current doctrines.

II.

Before proceeding to extend and apply these remarks on the idea of a Philosophy in general to the philosophy of Ethics in particular, it is necessary to correct an error which, in these days, when science and the knowable are supposed to be co-extensive, is natural though not the less mischievous:-the error I mean by which Ethics is degraded to a mere section or department of Science. At first sight, and from some points of view, the opinion seems plausible enough. That mankind have passed through

many ethical phases (for example) is a fact in history, and history belongs to science: that I hold certain moral laws to be binding is a fact of my mental being; and, like all other such facts, is dealt with by Psychology-also a branch of science. Physiology, Ethnology, and other sciences, all have something to say concerning the origin and development of moral ideas in the individual and in the race; it is not unnatural, therefore, that some men of science, impressed by these facts, have claimed, or seemed to claim, Ethics for their own.

To hold such a view would be a most unfortunate error; not to hold clearly and definitely its contrary may lead to much confusion. For though, as will appear, scientific laws form necessary steps in the deduction of subordinate ethical laws, and though the two provinces of knowledge cannot with advantage be separated in practice, still the truth remains that scientific judgments and ethical judgments deal with essentially different subject-matters.

Every scientific proposition asserts either the nature of the relation of space or time between phenomena which have existed, do exist, or will exist; or defines the relations of space or time which would exist if certain changes and simplifications were made in the phenomena (as in ideal geometry), or in the law governing the phenomena (as in ideal physics). Roughly speaking, it may be said to state facts or events, real or hypothetical.

An ethical proposition, on the other hand, though, like every other proposition, it states a relation, does not state a relation of space or time. "I ought to speak the truth," for instance, does not imply that I have spoken, do speak, or shall speak the truth; it asserts no bond of causation between subject and predicate, nor any co-existence nor any sequence. It does not announce an event; and if some people would say that it stated a fact, it is not certainly a fact either of the "external" or of the "internal" world.

One cause, perhaps, of the constant confusion between Ethics and Science is the tendency there appears to be to regard the psychology of the individual holding the moral law as the subjectmatter of ethics, rather than the moral law itself; to define the position which the belief in such a proposition as "I ought to speak the truth" holds in the history of the race and of the individual, its cause and its accompaniments, rather than its truth or its evidence; to substitute, in short, psychology or anthropology for ethics. The danger of such confusion will partly be shown by the few remarks which follow on the "Idea of a Philosophy of Ethics": that is, on the form which any satisfactory system of Ethics must assume, or be able to assume, whatever be its contents.

The obvious truth that all knowledge is either certain in itself,

or is derived by legitimate methods from that which is so, was sufficiently dwelt on before; and this, which is true of knowledge in general, is of course also true of ethical knowledge in particular. A little consideration will enable us to go on, and state this further fact, which is peculiar to ethics: The general propositions which really lie at the root of any ethical system must themselves be ethical, and can never be either scientific or ontological. In other words, if a proposition announcing obligation require proof at all, one term of that proof must always be a proposition announcing obligation, which itself requires no proof. This truth must not be confounded with that which I have just dwelt upon, namely, that Science and Ethics have essentially different subjectmatters. This might be so, and yet Ethics might be indebted for all its first principles to Science.

A concrete case will make this second statement clearer. A man (let us say) is not satisfied that he ought to speak the truth. He demands a reason, and is told that truth-telling conduces to the welfare of society. He accepts this ground, and apparently, therefore, rests his ethics on what is a purely scientific assertion. But this is not in reality the fact. There is a suppressed premiss required to justify his conclusion, which would run somewhat in this way-"I ought to do that which conduces to the welfare of society". And this proposition, of course, is ethical. This example is not merely an illustration, it is a typical case. There is no artifice by which an ethical statement can be evolved from a scientific or metaphysical proposition, or any combination of such; and whenever the reverse appears to be the case, it will always be found that the assertion, which seems to be the basis of the ethical superstructure, is in reality merely the minor of a syllogism, of which the major is the desired ethical axiom.

If this principle be as true as it seems to me to be obvious, at one blow it alters our attitude towards a vast mass of controversy which has encumbered the progress of moral philosophy. So far as the proof of a basis of morals is concerned, it makes irrelevant all discussion on the origin of moral ideas, or on the nature of moral sentiments; and it relegates to their proper sphere in psychology or anthropology all discussion on such subjects as association of ideas, inherited instincts, and evolution, in so far, at least, as these are supposed to refer to ultimate moral laws. For it is an obvious corollary from our principle, that the origin of an ultimate ethical belief never can affect its validity; since the origin of this belief, as of any other mental phenomenon, is a matter to be dealt with by Science; and my thesis is, that (negatively speaking) scientific truth alone cannot serve as a foundation for a moral system; or (to put it positively), if we have a moral system at all, there must be contained in it,