The Complete Prose of T. S. Eliot: The Critical Edition
Vol. 1: Apprentice Years, 1905-1918

followed by a note on the subject-matter of psychology, in illustration ¹

On 5 May 1914, Eliot read “Suggestions toward a Theory of Objects,” his fourth and final paper for Royce’s seminar. A record of the day’s proceedings, including a summary of Eliot’s presentation, is included in Costello’s notebook ( JRS173-77). ²

I consider the classification of the types of object entirely an empirical affair. The object as such is quite independent of the subject–that is, the object as object, not the object as “object.” We have simply to look about and discover what objects there are.

Distinction between classification of objects and the Kantian problem of knowledge. ³ Kant’s categories cover only things, and the categories are themselves objects (classes). ⁴ Kant’s problem is no more closely related to mine than the analysis of things into their material constituents is to his. According to the point of view of the theory of objects, we do not explain, we only describe: an explanation, that is, is always for purposes of practice. Much is explanation in epistemology which is not so in the theory of objects. For instance, the epistemologist may explain things as bundles of sense-data; but for us, this is simply the substitution of one group of objects for another. We incline to think, in such a substitution, that we have found something more real than was first given us. But whether this alteration of objects is a discovery about the same thing is a matter of degree, and depends upon the practical use that we can make of the discovery. If a coat is made up of threads, or if it is made up of sense-data, the coat and the threads and the sense-data are equally real. Yet we may ask whether the reals are all real in the same way, and whether we can handle all types of reals upon the same level. This question is the starting point for this theory.

We may begin by thinking of the world as quite wild and chaotic and contradictory, full of all sorts of object both positive and negative. We may take thingsas the primitive type of object. ⁵ An unreal thing, of course, is still a real object. All the higher objects, I believe, may be founded upon things–even the objects of the epistemologist, out of which he constructs his things, for the knowledge is subsequent to the knowledge of things: things are the first objects known to be such. This priority, however, must be understood to be only relative, inasmuch as the points of view of knowing and of being constantly shift before our attention, like the staircase in the illusion: ⁶ from one point of view, that is, things are prior in reality, and from another, they are prior only in knowledge–according as we consider knowing as knowing or knowing as an activity. The point of view here taken will be (provisionally of course) that of knowing as knowing, and the higher objects will be held to be “founded” upon things.

I repeat, then, that we do not begin in knowledge with sense-data, but that sense-data are derivative from things. The ultimate elements of knowledge, in this way, would be rather the individual constituents of matter. In any case the sense-datum is an object of a very different sort from a sense-datum [sic]. Consequently it is meaningless to say that this or that is what is “immediately” given without stating the field of objectivity with which one is concerned. Putting data together to make a chair involves an interpretation, a translationfrom one type to another, and is, from the point of view here taken, illegitimate.

We have here an indication of what I mean by a type of object. Things form one type, sense-data another. Universals and relations may form another type. I think that higher objects, such as these mentioned, are understood only by reference to things: i.e. we “explain...