it must be that the extent of generalization will be very variable, according to the circumstances of the case and the temperament of him who judges, but that it will probably not be very much wider than is demanded to justify the instance before us; to which we may add that it seems absurd to suppose that the earlier generalizations can possibly be coextensive with what wę now accept as the result of our scientific training.

We may then describe the gradual evolution of the belief in Uniformity, so far as its consideration belongs to the logician, somewhat as follows.

Universal order, or causation, is acted on by all men from their early infancy, or at any rate from the first time at which they show any intelligent activity of their own. It is equally acted on, in a similar way, by most animals, according to the range of their experience. That is; actions, not merely of the reflex or automatic kind, but such as in our case are of the conscious or purposive kind, are perpetually and confidently performed in harmony with the regularity which exists in nature outside us, long before such regularity is perceived. Anyhow, long before we have reached the logical stage, i.e. the stage at which we can ask why we believe, we have already acquired the belief over a number, so to say, of distinct areas of varied but limited experience. It is first appreciated or recognized as a logical guide at the stage, whenever that may be, at which we begin to question and justify our actions: or rather perhaps, in order not unduly to limit the scope of Logic, at the stage at which we might begin to question and justify our actions if someone else prompted us to stand on the defence. Presumably the animals never reach this stage, and man does not reach it until he is some way past infancy; so that it is better not to claim the infant, as Reid and Stewart do, as a believer in Uniformity.

As already stated, this range of conscious justificatory generalization is probably in most cases a decidedly narrow one :in the example of our snake-bite, it is very likely that all which would be thought of at the time as relevant, or quoted in defence of the inference, would be the analogy of other such snake-bites. By a gradual extension of experience, and a constantly verified appeal to it, this belief is widened in its scope. Although therefore I cannot agree with Mill that the belief in

Uniformity is properly described as being obtained by Induction, -i.e. by a truly logical process,-I certainly think with him that when it is once consciously realized over comparatively small ranges, all the subsequent growth is fairly describable as being of the nature of simple Induction. By a multitude of such steps, each helping us on a little by extending the appreciation of Uniformity beyond the actual observed case, we may in time gain a complete generalization covering the whole field of nature. How near we approach towards the ideal of realizing an allpervading uniformity will depend upon our character and the nature of our experience. Those gifted with a strong generalizing disposition, especially if their study of nature has been wide and accurate, so that they have come to appreciate the precision with which remote consequences can be inferred, will grasp it in a very wide sense. Very likely they will hold that such uniformity exists everywhere, extending throughout the whole region of material and mental phenomena. Whether or not they are justified in doing this it seems to me impossible to say. But it is reasonable to insist that the belief shall become less confident in proportion as it refers to matters more remote from actual experience.

We should see this more clearly if we went more into the details of what constitutes Uniformity. It is, as we have seen, a term of wide import, and by no means coextensive in signification with the causal relation. If, for instance, we confined ourselves to the narrowest and strictest interpretation of the causal sequence, in which, as we saw, the law became almost a formal and necessary one, then indeed it is hard to set any limits to the confidence we should feel in its universal prevalence. But in that interpretation it is purely hypothetical and does not tell us anything about the actual occurrence of phenomena. To postulate, therefore, universal validity for such a law is merely another way of saying that we cannot transcend the laws of our own understanding: that whatever we conceive, or wherever we may suppose ourselves to locate what we conceive, it is still we with our present faculties who are conceiving it. Taking the causal relation, then, in this strict sense, I cannot but think that Mill overrates our capacity when he admits the possibility of the law being infringed in the remote parts of stellar space.

On the other hand, when the law is interpreted in the looser sense, that is, in almost any one of the various subdivisions which were considered in the previous chapter,—I think that Mill speaks with more hesitation than he need adopt. So far from admitting the bare possibility of a breach of uniformity in this sense I should think it not at all unlikely that in the endless stretches of time and space there may be developments in store which fully deserve the name. But as the subject will come under our notice again in a future chapter, I will not pursue it into further detail here.

CHAPTER VI.

LANGUAGE.

ON every theory, whether of Logic or of Psychology, Language is intimately connected with all our processes of conceiving, judging and reasoning. Hence it becomes necessary to take some account of this medium of communication. We shall have, of course, to consider it in some detail hereafter under the heads of Names and Propositions. What here concerns us are the more general considerations of its reference, its functions, and the medium through which it is conveyed. We will take these in turn, limiting the discussion as rigidly as is convenient to the requirements of the Science with which we are here concerned.

I. In speaking of the Reference of language I allude to a dispute which may appear to the reader a rather idle one, viz. whether the words we use are to be supposed to refer to the objects without us or to the notions within us. It will be remembered that in the first chapter attention was directed to the triple correspondence between these three elements. It was intimated that in a healthy mind these should accurately correspond with each other; in the sense that the same words should always excite, and be themselves suggested by, the same notions, and that either of these should always represent the same external phenomena. We explained some of the assumptions demanded to secure such a complete correspondence, and admitted, it need not be said, that nothing approaching to such an accurate fit as this was to be found in practice.

1 This particular dispute is by no means a purely modern one. Thus, for instance, Smiglecius, when stating one side of the case, says "vox enim, homo, vel Petrus, non significat mihi conceptum hominis vel Petri; sed verum et realem hominem extra intellectum existentem" (Disputationes XII. 1, ed. 1618).

Starting then from this basis, with the word, the question is asked, Does it refer to, that is, does it denote, or is it the name of the object, or our notion of the object? Popular judgment would, I suppose, decide off-hand for the former: the general decision of logicians was, till lately, for the latter. Mill, as is well known, held very strong views upon the subject, declaring that the current logical doctrine was "one of the most fatal errors ever introduced into Logic". We shall best appreciate the importance of the question, and see our way to a decision about it, by examining the reasons which may be advanced in favour of the old logical view and against it.

It may be urged, in the first place, that the notion is something comparatively fixed and finite; that is, it consists of a tolerably rigidly determined group of attributes or constituent notions, which we may conceive to be retained in the mind, or transferred to others like a sort of currency, with ease and security. The objective thing itself, on the other hand, possesses attributes whose number no one can estimate, many of which are fluctuating, others very uncertain, others absolutely unknown, whilst in any case only a very few of them can be present to the mind at any assigned time. We know, it may be said, what the term 'man' signifies: if 'rational animal' is not enough we can add on more attributes, and come soon to the end of those which are really characteristic. But who can attempt to enumerate the attributes of man himself?

There is something in this; and there would be a good deal more in it if Logic were to be treated deductively, and in the style of the scholastics. As I have already intimated (p. 18) the general character of the old treatment was rather that of a professional class of thinkers dealing with a stock-in-trade of notions whose exchange value was thoroughly familiar to them all. We shall see this better when we come to deal with the Categories and Definition. At present it will suffice to remark that with a well-determined concept-currency of this description, in the hands of men who were in constant communication with each other, and who were much more in the habit of comparing and analysing the notions they had already obtained than of correcting and extending them by appeal to experience, there was a certain propriety in regarding the verbal symbol as