the business of one of the particular sciences to deal. Philosophy aims at determining form, without reference to any particular content. This philosophic analysis has an important application in the investigation of those concepts which we ordinarily apply to the object of experience, notably the concepts of physical science. Since all observation, avowedly scientific or otherwise, must start from sense-data, and since all verification of calculation based on such observation must lie in an appeal to sense-data, it follows that if our concepts are valid, they must be capable of being exhibited as logical functions of sense-data. The analysis of the concepts, therefore, finally resolves itself into an attempt to build up such constructions of sense-data as may be considered satisfactorily to represent the concepts involved. Hence the method, at any rate in the last stages of its application, is constructive. Nevertheless, its function is evidently to a considerable extent critical. Its field of application consists in the whole of the objective side of experience, and we may willingly admit the claim of its exponents, that it is the only method of obtaining accurate objective scientific knowledge, provided it is clearly recognized that it is subject to two limitations springing from a common root. In the first place, all individual experiences are essentially particular, and the assumption that they exhibit certain general characteristics of form must therefore be regarded as an approximation which is only justified by the fact that it works satisfactorily in practice so long as we are not concerned with a final complete adequacy, and by the still more cogent fact that we are bound to it by our intellectual limitations. In the second place, the units with which the scientific method works are sense-data, and the sense-datum is a purely artificial and conventional unit. The object of experience is an indivisible unity, and (whatever convention we may be compelled to adopt for the purposes of calculation) cannot be considered to consist in a series of members termed 'sense-data,' compact or otherwise.
Keeping in mind these limitations of the method, its critical and constructive value in its own field is apparent. In any case, however, its results are purely descriptive. Its exponents claim that the determination of such results as their method affords is
