Speaking from this point of view, the decisive consideration as between instrumentalism and analytic realism is whether the operation of experimentation is or is not necessary to knowledge. The instrumental theory holds that it is; analytic realism holds that even though it were essential in getting knowledge (or in learning), it has nothing to do with knowledge itself, and hence nothing to do with the known object: that it makes a change only in the knower, not in what is to be known. And for precisely the same reason, instrumentalism holds that an object as a knowledge-object is never a whole; that it is surrounded with and inclosed by things which are quite other than objects of knowledge, so that knowledge cannot be understood in isolation or when taken as mere beholding or grasping of objects. That is to say, while it is making the sick man better or worse (or leaving him just the same) which determines the knowledge-value of certain findings of fact and certain conceptions as to mode of treatment (so that by the treatment they become definitely knowledge-objects), yet improvement or deterioration of the patient is other than an object of cognitive apprehension. Its knowledge-object phase is a selection in reference to prior reflections. So the laboratory experiment of a chemist which brings to a head a long reflective inquiry and settles the intellectual status of its findings and theorizings (thereby making them into cognitive concerns or terms and propositions) is itself much more than a knowledge of terms and propositions, and only by virtue of this surplusage is it even contemplative knowledge. He knows, say, tin, when he has made tin into an outcome of his investigating procedures, but tin is much more than a term of knowledge.
Putting the matter in a slightly different way, logical (as distinct from naïve) realism confuses means of knowledge with objects of knowledge. The means are twofold: they are (a) the data of a particular inquiry so far as they are significant because of prior experimental inquiries; and (b) they are the meanings which have been settled in consequence of prior intellectual undertakings: on the one hand, particular things or qualities as signs; on the other, general meanings as possibilities of what is signified by given data. Our physician has in advance a technique for telling that certain particular traits, if he finds them, are symptoms, signs; and he has a store of diseases and remedies in mind which may possibly be meant in any given case. From prior reflective experiments he has learned to look for temperature, for rate of heartbeats, for sore spots in certain places; to take specimens of blood, sputum, of membrane, and subject them to cultures, microscopic examination, etc. He has acquired certain habits, in other words, in virtue of which certain physical qualities and events are more than physical, in virtue of which they are signs or indications of something else.
On the other hand, this something else is a somewhat not physically present at the time: it is a series of events still to happen. It is suggested by what is given, but is no part of the given. Now, in the degree in which the physician comes to the examination of what is there with a large and comprehensive stock of such possibilities or meanings in mind, he will be intellectually resourceful in dealing with a particular case. They (the concepts or universals of the situation) are (together with the sign-capacity of the data) the means of knowing the case in hand; they are the agencies of transforming it, through the actions which they call for, into an object—an object of knowledge, a truth to be stated in propositions. But since the professional (as distinct from the human) knower is particularly concerned with the elaboration of these tools, the professional knower—of which the class philosopher presents of course one case—ungenerously drops from sight the situation in its integrity and treats these instrumentalities of knowledge as objects of knowledge. Each of these aspects—signs and things signified—is sufficiently important to deserve a section on its own account.
V
The position taken in the essays is frankly realistic in acknowledging that certain brute existences, detected or laid bare by thinking but in no way constituted out of thought or any mental process, set every problem for reflection and hence serve to test its otherwise merely speculative results. It is simply insisted that as a matter of fact these brute existences are equivalent neither to the objective content of the situations, technological or artistic or social, in which thinking originates, nor to the things to be known—to the objects of knowledge. Let us take the sequence of mineral rock in place, pig iron and the manufactured article, comparing the raw material in its undisturbed place in nature to the original res of experience, compare the manufactured article to the objective and object of knowledge, and the brute datum to the metal undergoing extraction from raw ore for the sake of being wrought into a useful thing. And we should add that just as the manufacturer always has a lot of already extracted ore on hand for use in machine processes as it is wanted, so every person of any maturity, especially if he lives in an environment affected by previous scientific work, has a lot of extracted data—or, what comes to the same thing, of ready-made tools of extraction—for use in inference as they are required. We go about with a disposition to identify certain shapes as tables, certain sounds as words of the French language, certain cries as evidences of distress, certain massed colors as woods in the distance, certain empty spaces as buttonholes, and so on indefinitely. The examples are trivial enough. But if more complicated matters were taken, it would be seen that a large part of the technique of science (all of science which is specifically "inductive" in character) consists of methods of finding out just what qualities are unambiguous, economical, and dependable signs of those other things which cannot be got at as directly as can the sign-bearing elements. And if we started from the more obscure and complex difficulties of identification and diagnosis with which the sciences of physiology, botany, astronomy, chemistry, etc., deal, we should be forced to recognize that the identifications of everyday life—our "perceptions" of chairs, tables, trees, friends—differ only in presenting questions much easier of solution.
In every case, it is a matter of fixing some given physical existence as a sign of some other existences not given in the same way as is that which serves as a sign. These words of Mill might well be made the motto of every logic: "To draw inferences has been said to be the great business of life. Everyone has daily, hourly, and momentary need of ascertaining facts which he has not directly observed.... It is the only occupation in which the mind never ceases to be engaged." Such being the case, the indispensable condition of doing the business well is the careful determination of the sign-force of specific things in experience. And this condition can never be fulfilled as long as a thing is presented to us, so to say, in bulk. The complex organizations which are the subject-matter of our direct activities and enjoyments are grossly unfit to serve as intellectual indications or evidence. Their testimony is almost worthless, they speak so many languages. In their complexity, they point equally in all directions; in their unity, they run in a groove and point to whatever is most customary. To break up the complexity, to resolve it into a number of independent variables each as irreducible as it is possible to make it, is the only way of getting secure pointers as to what is indicated by the occurrence of the situation in question. The "objects" of ordinary life, stones, plants, cats, rocks, moon, etc., are neither the data of science nor the objects at which science arrives.
We are here face to face with a crucial point in analytic realism. Realism argues that we have no alternative except either to regard analysis as falsifying (à la Bergson), and thus commit ourselves to distrust of science as an organ of knowledge, or else to admit that something eulogistically termed Reality (especially as Existence, Being as subject to space and time determinations) is but a complex made up of fixed, mutually independent simples: viz., that Reality is truly conceived only under the caption of whole and parts, where the parts are independent of each other and consequently of the whole. For instrumentalism, however, the alleged dilemma simply does not exist. The results of abstraction and analysis are perfectly real; but they are real, like everything else, where they are real: that is to say, in some particular coexistence