than themselves. What I wish to urge is that the introduction of the idea of correlation is inconsistent with the view of the function of sense-data which the theory maintains, and implies the view which the theory rejects.

Correlation, for Mr. Broad, is not merely connection in time, but means essentially correspondence in shape, or spatial form. If this conception is to be of service, both visual and tactual sensa must literally possess shape in the same sense. It would not avail, of course, to say that shape is literally visual and is only applied to tactual sensa in a Pickwickian sense; for it is not until correlations have been made that Pickwickian senses arise. Whether they are "numerically" identical or not, therefore, tactual space and visual space have an identical nature, in spite of the fact that they are apprehended by modes of sensing so different as seeing and touching. This fact is open to two interpretations. Either spatial form is immediately sensed, or else, though not itself sensed, it is something about which our senses yield us information. The second alternative clearly strikes directly at the foundations of the theory, but the first is not less fatal to it. For it cuts the ground from under the assumption that the object of one sense is ipso facto different from the object of another sense; and this is none the less true even if it be still held that visual space is numerically different from tactual space. But when this is recognized, there seems no longer any ground for holding that visual and tactual space are two and not one. For a spatial form which is the same for two such diverse kinds of material as visual and tactual sensa cannot be merely an attribute of, or an abstraction from, these sensa; it must have a certain independence of either. There seems no

reason why a space so conceived could not be occupied by both visual and tactual sensa. And this is just the commonsense view which the theory is seeking to sub

vert.

It seems, however, inaccurate to speak of shape as being sensed. This is fairly clear in the case of tactual sensa. As Mr. Broad remarks, "passive touch, considered by itself, gives very vague information about shape." (p. 340.) If we learn more from active touch, this is because active touch is not only a tactual experience but an experience of movement. This apprehension of movement, however, as Mr. Broad also tells us, is not in itself a spatial experience, but a series of kinaesthetic and muscular sensations which we learn to interpret as spatial experiences. But this interpretation is itself exhibited as a result of correlation, and the argument thus moves in a circle. On the one hand it is only by correlation that we get beyond sensa; on the other hand we must have got beyond sensa before correlations can be made.

Though we get more direct information about the shape of an object from sight, even the visual shape is something more than a sensum. When we say that a penny looks round, has "round" a merely sensible significance? If anyone says yes, will he maintain the same view if for round we substitute elliptical? The more exclusively geometrical associations of the latter word reveal more clearly the fact that shape is not merely a sensible impression but implies at least an elementary form of measurement; the apprehension of certain relations between the different points on a figure. Shape, in a word, is an object not merely of sense but of the understanding.

The attempt to exhibit the formation of the idea of

a physical world as a result of correlations of shape between independent objects of the different senses thus seems to me to break down at all points. Yet it is obvious that we can and do compare and correlate the evidence about shape given by the different senses. The explanation is to be found in recognizing that seeing is always more than sensing visual sensa and touching always more than sensing tactual sensa.

It is perhaps worth noting that Mr. Broad's account of the way in which we come to form the notion of physical objects is only made plausible by the abstractly intellectual way in which he treats the acquisition of knowledge. If there is originally a conative element in experience, it seems to follow that from the outset we must be in conscious relation to a world of perceptual objects, however vaguely that world may at first be apprehended. One does not see how a conative attitude could exist towards sensa. From this point of view also the process of correlation appears to presuppose the apprehension in principle of the physical world and to have as its function the development and differentiation of this knowledge.

The view of the role played by mind in experience which emerges from the foregoing discussion is very different from the view taken by Mr. Broad. For him, though he does not put it quite so bluntly, the work of the mind consists in constructing elaborate fictions. But if the criticisms offered in the present article are well grounded, the mind in making correlations is not manufacturing fictitious objects but discovering real ones: whilst its activity is already seen in the materials with which it works; they are not supplied to it, ready made, from outside.

It is at this point that I find it specially difficult to

understand Mr. Broad's position. Knowledge, we are told, is developed from sensation, which is a complex consisting of acts of sensing directed on sensa. From such a starting point we might expect that it would be maintained that experience consists of a succession of sensations and nothing more, and that an attempt would be made to analyse mental processes into a succession of acts of sensing. This would seem to me to be the logical development of the position, but it would clearly be a suicidal course for Mr. Broad to take. For a succession of acts of sensing can only yield a succession of sensa; and an activity other than sensing must therefore be introduced in order to account for our belief in a physical world which is more than a number of sensa. But unless this activity is somehow present in even the most primitive form of experience, it is difficult to see how it ever comes into experience at all. However this may be, it is clear that the process of building up our imagined physical world out of sensedata cannot begin until there is something more on the subjective side of experience than acts of sensing. And when this "something more" is present on the subjective side, there seems no logical ground for speaking of the objective correlate of this mental complex as a sensum, i.e. as the object of an act of sensing. To put the point in another way: if the sensum is merely the object of an act of sensing, how can the transforming mental activity become aware of it at all; and how are we to conceive the relation between this activity and the original act of sensing? The only way out of these perplexities, as it seems to me, is to recognise that more than sensing is involved in the so-called "act of sensing." That this is so is strongly suggested by the use of the word "act" in this connection. For the nearer

my state of mind approaches to a sensational level, the less can I find in it anything that can intelligibly be called an act. Mr. Broad's consistent use of the expression "act of sensing" seems to me significant, therefore, as indicating in his primitive form of experience the presence of an unacknowledged element. And if the act of sensing is to be rejected, then, I submit, the sensum must be rejected too.

There remain many interesting aspects of the theory which I have not discussed: for example, the further problems that arise when the perceptual world is considered not merely as my world but as a world that is common to us all. But as this paper is already too long, I shall not attempt to deal with these questions, but shall conclude with two quite general observations. The first is that only a very slight change in the theory would be needed in order to transform it into a one-sided idealism. For the theory holds that what we regard as our real world-the world that we believe ourselves to inhabit and which contains everything that we value and strive for this world, at any rate, is the product of our mental activity. And even the sensa of which this world is actually composed have a very doubtful claim to independent existence. Mr. Broad tells us that sensa are either selected or generated or both. If selection were eliminated-and it is admitted that there are grave difficulties in holding that they are selected the sensum theory would end in a thoroughgoing subjective idealism. For in the last resort it must be the mind that generates sensa; the body is itself only a collection of

sensa.

The other point-an obvious one-I will put in the form of a question. How would Mr. Broad account for the fact that mutually independent classes of sensa