less than the word 'percipi': that the two words are precise synonyms: that they are merely different names for one and the same thing: that what is meant by esse is absolutely identical with what is meant by percipi. I think I need not prove that the principle esse is percipi is not thus intended merely to define a word; nor yet that, if it were, it would be an extremely bad definition. But if it does not mean this, only two alternatives remain. The second is (2) that what is meant by esse, though not absolutely identical with what is meant by percipi, yet includes the latter as a part of its meaning. If this were the meaning of 'esse is percipi,' then to say that a thing was real would not be the same thing as to say that it was experienced. That it was real would mean that it was experienced and something else besides: 'being experienced' would be analytically essential to reality, but would not be the whole meaning of the term. From the fact that a thing was real we should be able to infer, by the law of contradiction, that it was experienced; since the latter would be part of what is meant by the former. But, on the other hand, from the fact a thing was experienced we should not be able to infer that it was real; since it would not follow from the fact that it had one of the attributes essential to reality, that it also had the other or others. Now, if we understand esse is percipi in this second sense, we must distinguish three different things which it asserts. First of all, it gives a definition of the word 'reality,' asserting that word stands for a complex whole, of which what is meant by 'percipi' forms a part. And secondly it asserts that 'being experienced' forms a part of a certain whole. Both these propositions may be true, and at all events I do not wish to dispute them. I do not, indeed, think that the word 'reality' is commonly used to include 'percipi': but I do not wish to argue about the meaning of words. And that many things which are experienced are also something else—that to be experienced forms part of certain wholes, is, of course, indisputable. But what I wish to point out is, that neither of these propositions is of any importance, unless we add to them a third. That 'real' is a convenient name for a union of attributes which sometimes occurs, it could not be worth any one's while to assert: no inferences of any importance could be drawn from such an assertion. Our principle could only mean that when a thing happens to have percipi as well as the other qualities included under esse, it has percipi: and we should never be able to infer that it was experienced, except from a proposition which already asserted that it was both experienced and something else. Accordingly, if the assertion that percipi forms part of the whole meant by reality is to have any importance, it must mean that the whole is organic, at least in this sense, that the other constituent or constituents of it cannot occur without percipi, even if percipi can occur without them. Let us call these other constituents x. The proposition that esse includes percipi, and that therefore from esse percipi can be inferred, can only be important if it is meant to assert that percipi can be inferred from x. The only importance of the question whether the whole esse includes the part percipi rests therefore on the question whether the part x is necessarily connected with the part percipi. And this is (3) the third possible meaning of the assertion esse is percipi: and, as we now see, the only important one. Esse is percipi asserts that wherever you have x you also have percipi that whatever has the property x also has the property that it is experienced. And this being so, it will be convenient if, for the future, I may be allowed to use the term 'esse' to denote x alone. I do not wish thereby to beg the question whether what we commonly mean by the word 'real' does or does not include percipi as well as x. I am quite content that my definition of 'esse' to denote x, should be regarded merely as an arbitrary verbal definition. Whether it is so or not, the only question of interest is whether from x percipi can be inferred, and I should prefer to be able to express this in the form: can percipi be inferred from esse? Only let it be understood that when I say esse, that term will not for the future include percipi: it denotes only that x, which Idealists, perhaps rightly, include along with percipi under their term esse. That there is such an x they must admit on pain of making the proposition an absolute tautology; and that from this x percipi can be inferred they must admit, on pain of making it a perfectly barren analytic proposition. Whether x done should or should not be called esse is not worth a dispute: what is worth dispute is whether percipi is necessarily connected with x.

We have therefore discovered the ambiguity of the copula in esse is percipi, so far as to see that this principle asserts two distinct terms to be so related, that whatever has the one, which I call esse, has also the property that it is experienced. It asserts a necessary connexion between esse on the one hand and percipi on the other; these two words denoting each a distinct term, and esse denoting a term in which that denoted by percipi is not included. We have, then in esse is percipi, a necessary synthetic proposition which I have undertaken to refute. And I may say at once that, understood as such, it cannot be refuted. If the Idealist chooses to assert that it is merely a self-evident truth, I have only to say that it does not appear to me to be so. But I believe that no Idealist ever has maintained it to be so. Although this—that two distinct terms are necessarily related—is the only sense which 'esse is percipi' can have if it is to be true and important, it can have another sense, if it is to be an important falsehood. I believe that Idealists all hold this important falsehood. They do not perceive that Esse is percipi must, if true, be merely a self-evident synthetic truth: they either identify with it or give as a reason for it another proposition which must be false because it is self-contradictory. Unless they did so, they would have to admit that it was a perfectly unfounded assumption; and if they recognised that it was unfounded, I do not think they would maintain its truth to be evident. Esse is percipi, in the sense I have found for it, may indeed be true; I cannot, refute it: but if this sense were clearly apprehended, no one, I think, would believe that it was true.

Idealists, we have seen, must assert that whatever is experienced, is necessarily so. And this doctrine they commonly express by saying that 'the object of experience is inconceivable apart from the subject.' I have hitherto been concerned with pointing out what meaning this assertion must have, if it is to be an important truth. I now propose to show that it may have an important meaning, which must be false, because it is self-contradictory.

It is a well-known fact in the history of philosophy that necessary truths in general, but especially those of which it is said that the opposite is inconceivable, have been commonly supposed to be analytic, in the sense that the proposition denying them was self-contradictory. It