certainly have that, namely, of the series of our mental states as they occur in time-is impossible unless the thing to be proved, which is stated (though, as we shall see, incorrectly stated) to be the existence of external objects, be admitted. And the demonstration consists of two steps. First, it is asserted that the experience of a succession of things in time is impossible except in relation to something permanent, or, in other words, that the perception of change is inconceivable unless we at the same time perceive something which does not change. And in the second place, Kant goes on to say that, since that which changes in this case is myself (my phenomenal self), since the "things" which succeed each other in time are my own mental states, the unchanging object to which they are referred must be outside myself; that is, must be the external object whose existence was to be proved. So that if we immediately perceive the one, it can only be on condition that we immediately perceive the other also.

Such is the formal answer which Kant has given to Idealism; but it is not in this way only that he has treated the question, since in his proof of the principle of Substance (which precedes the "Refutation" in the Critique) he has brought forward arguments which, if sound, would seem to render any further refutation superfluous. For the "First Analogy of Experience" asserts this," That in all changes of phenomena substance is permanent; and the quantum thereof in nature is neither increased nor diminished." (Critique, p. 136.) And as by substance Kant means something which, if it is not (as I think it is) exactly equivalent to what is commonly called matter, is at any rate the genus of which matter is one species; clearly this proposition is absolutely inconsistent with Idealism in the sense in which I use the term. If matter is permanent and indestructible, we need not further trouble ourselves as to whether there are or are not in nature other things besides our conscious states.

The proof of this principle of Substance, which I give partly in Kant's words, partly in Mr. Caird's, and partly in my own, runs somewhat in this way:

All phenomena exist in time. Change is only conceivable in an unchanging time. But this time is not, and cannot be, itself an object of perception, but is rather a form given to the relations of perception, which presupposes that they are otherwise related. They must be otherwise related as determinations of a permanent substance. As all times are in one time, so all changes must be in one permanent object. The conception of the permanence of the object is implied in all determination of its changes. Change involves that one mode of existence follows another mode of existence in an object recognised as the same. Therefore a thing which changes, changes only in its states or accidents, not in its substance. An experience of absolute annihilation or creation is impossible, for it would

be an experience of two events so absolutely separated from each other that they could not even be referred to one time. The "First Analogy," therefore, is a deduction from the possibility of experience, and requires no empirical proof. When a philosopher was asked 'What is the weight of smoke?' he answered, 'Subtract from the weight of the burnt wood the weight of the remaining ashes, and you will have the smoke'. Thus, he presumed it to be incontrovertible that even in fire the matter (substance) does not perish, but only the form of it undergoes a change. (Cf. Crit., p. 136; Caird, p. 453.)

The reader will at once perceive that while there is much that is common to the "Refutation" and the "First Analogy," there are some arguments and doctrines peculiar to each, a fact which makes the satisfactory discussion of the question rather difficult; because while it is impossible to treat the two arguments as identical, it is somewhat clumsy and would lead to a good deal of repetition to consider them altogether separately. The most convenient course, perhaps, will be first to consider the points which are to be found in both, and then to proceed with the examination of their mutual relationship and with what is special to each.

The first difficulty which occurs to me, and which perhaps others may feel, refers to that "transcendental necessity" which is the very pith and marrow of the whole demonstration, both in the "Refutation" and in the "First Analogy". Is it really true that change is "nothing to us as thinking beings" except we conceive it in relation to a permanent and unchanging substance? For my part, however much I try to bring the matter into "clear consciousness," I feel myself bound by no such necessity. For though change is, doubtless, unthinkable, except for what Mr. Green calls a "combining" and, therefore, to a certain extent, a "persisting consciousness," and though it may have no meaning out of relation to that which is "notchange," this "not-change" by no means implies permanent substance. On the contrary, the smallest recognisable persisttence through time would seem enough to make change in time intelligible by contrast; and I cannot help thinking that the opposite opinion derives its chief plausibility from the fact that in ordinary language permanence is the antithesis to change; whence it is rashly assumed that they are correlatives which imply each other in the system of nature. It has to be noted also, that Kant, in his proof of the "First Analogy," makes a remark (quoted and approved by Mr. Caird) which almost seems to concede this very point, for he says (Crit., p. 140): " Only the permanent is subject to change: the mutable suffers no change, but rather alternation; that is, when certain determinations cease, others begin". Now there can be no objection, of course, from a philosophical point of view, to an author defining a

word in any sense he pleases: what is not permissible is to make such a definition the basis of an argument as to matters of fact; yet the above passage suggests the idea that Kant's proof of the permanence of substance is not altogether free from this vice. If (by definition) change can only occur in the permanent, the fact that there is change is no doubt a conclusive proof that there is a "permanent ". But the question then arises, Is there change in this sense? How do we know that there is anything more than alternation which (by definition) can take place in the mutable? All transcendentalists convince by threats. "Allow my conclusion," they say, " or I will prove to you that you must surrender one of your own cherished beliefs." But in this case the threat is hardly calculated to frighten the most timid philosopher. There must be a permanent, say the transcendentalists, or there can be no change; but this surely is no very serious calamity if we are allowed to keep alternation, which seems to me, I confess, a very good substitute, and one with which the ordinary man may very well content himself.

To those who agree with the preceding account of our intellectual necessities, who can either conceive change without permanence, or are content to get along with the help of "alternation," it will seem absolutely fatal to the whole Kantian. argument, both in the "First Analogy" and the "Refutation". To those who do not agree, it will only be a difficulty in so far as the existence of any mind unconscious of transcendental necessities is inconsistent with the transcendental theory-a point I have already discussed. But let us pass over this, and grant, for the sake of argument, that change in general, or the succession of our mental states in particular, can only be perceived in relation to a permanent something; then I ask (and this is the next most obvious objection) why, in order to obtain the permanent something, should we go to external matter? As the reader is aware, the "pure Ego of apperception' supplies, on the Kantian system, the unity in reference to which alone the unorganised multiplicity of perception becomes a possible experience; and it seems hard to understand why that which supplies unity to multiplicity, may not also supply permanence to succession. Kant has, indeed, anticipated this objection and replied to it; but as I understand the objection much better than I do the reply, I will content myself with giving the latter, without comment, in Kant's own words:

"We find," he says, "that we possess nothing permanent that can correspond and be submitted to the conception of a substance as intuition, except matter. In the representation I, the consciousness of myself is not an intuition, but a merely intellectual representation produced by the spontaneous activity of a thinking subject. It follows that this I

has not any predicate of intuition, which, in its character of permanence, could serve as correlate to the determination of time in the internal sense-in the same way as impenetrability is the correlate of matter as an empirical intuition." (Critique, p. 168.)

Though I do not profess altogether to understand the reasoning, it is, at all events, clear from it, that "the permanent whose existence in demonstrated, must be an object of perception, a fact which is also evident from various passages in the proof of the "First Analogy," as, for instance, this: "Time itself cannot be an object of perception. It follows that in objects of perception, that is, in phenomena, there must be found a substratum," &c. (Critique, p. 137.) It is difficult to see indeed how that which is a quantity, incapable of either increase or diminution, can be other than an object of perception-it cannot at all events be a concept-and we may, I think, assume from the whole tenor of Kant's argument, as well as from his categorical assertions, that the substance of which he speaks is a phenomenal thing. But if it be perceived,

if it be a phenomenon, where is it to be found? In the perpetual flux of nature, where objects do indeed persist for a time, but where (to all appearance) nothing is eternal, who has had experience of this unchanging existence? By a dialectical process, probably familiar to the reader, we may with much. plausibility reduce what we perceive in an object to a collection of related attributes, not one of which is the object itself, but all of which are the changing attributes or accidents of the object. But if this process be legitimate, the "substratum" of these accidents is either never perceived at all, or at all events is only known as a relation. In neither case can it be the permanent of which Kant speaks, since in the first case it is not an object of immediate perception; in the second it can hardly be regarded as an object at all.

"But (it may perhaps be replied), by a remarkable coincidence, science has established by a wide induction the very truth which Kant attempts to prove à priori. When men of science tell us that matter is indestructible, it is to be presumed that they attach some meaning to the phrase, and are referring neither to a metaphysical substance nor to an evanescent appearance. When Kant uses the same phrase, it may be supposed that he refers to the same object." For my own part, I confess to a rooted distrust of those remarkable coincidences between the results of scientific experiment and à priori speculation; nor does a closer examination of this particular case tend to allay the feeling. It is true, no doubt, that science asserts matter to be indestructible; but what is the exact meaning of the phrase, and what is its evidence? Can we perceive any thread of

identity running through all the various changes which (what we describe as) one substance may undergo? To a certain extent science assures us that we can. There are two, though, so far as I know, only two attributes of matter, namely, its relation to a moving force and its power of attracting and being attracted by other matter, which never alter; or-to put it more strictly-if we take a certain "area of observation" (say a closed vessel) out of which matter cannot pass and into which it cannot enter, then, whatever changes occur within this, the matter there, whether always the same or not, never varies in respect of these two properties. But it has to be observed, that though we can directly per eive both velocity and weight, the fact that there are unchanging relations between a given portion of matter and a given force, or between two portions of given matter, can only be established by an elaborate process of inference involving a large number of assumptions. It might therefore be plausibly contended that, though they are perceived, their permanence is not, so that they cannot properly be said to form any permanent element in perception. Passing over this possible objection, however, and granting, for the sake of argument, that we directly perceive the permanence of these two properties of matter, it is still clear that, since these are the only two properties of which we can say as much, either they must constitute matter, or matter, in so far as it is permanent, cannot be an object of perception. The first alternative is inadmissible, because these properties are merely relations between certain portions of matter and something else. The second would seem to be inconsistent with the Kantian proof.

The reader will understand that I am not here contending that Kant's conclusion is inconsistent with science, or that the scientific inference is wrong, either in its method or its results. My point is rather this:-Though Kant does not, of course, conclude to the necessary permanence of matter merely from its permanence in perception, nevertheless its permanence in perception would seem to be involved in his proof. Now I assert that what we perceive, in so far as it is perceived, is either not matter or is not permanent; and I maintain that an examination of that part of the ordinary scientific or empirical proof which bears on the question really confirms this view.

It may perhaps be thought (and some of Kant's expressions countenance the view) that he means to say no more than that we perceive the permanent substance by means of certain of its accidents. But this seems to raise new difficulties. First, how is the phenomenal substance, thus mediately known, to be distinguished from the noümenal substance, which, if it be known at all, is known precisely in the same way? Why