established lines of his thought to call for serious consideration.

But if new relationships and new properties appear in the course of evolutionary progress, where is the opportunity for that unification of scientific knowledge which, according to Spencer, is the goal of philosophy? To be frank, I am by no means sure that this question can be answered in a manner that is other than tentative. Perhaps we have not yet reached the stage at which more than provisional unification is possible. Such provisional unification as is suggested by a survey of the facts is that of seemingly uniform correlation in a hierarchy of logical implication. There are certain modes of relatedness which belong to the cognitive type. It would seem that whenever these obtain they may be correlated with other modes of relatedness which are of the vital or physiological type; and that these, in turn, may be correlated with those that are physico-chemical. Thus C implies B, and B implies A. The order cannot be reversed. Physico-chemical relations, as a class, do not imply those that are physiological.[57] The implication is not symmetrical. Spencer was within sight of this when he spoke[58] of the abstract-concrete sciences as 'instrumental' with respect to the concrete sciences, though the latter are not 'instrumental' in the same sense with respect to the former. But unfortunately he regarded the 'chasm' between the two groups as 'absolute'. And for him the proper home of properties—of all properties it would appear—is the abstract-concrete group—mechanics, physics, and chemistry. This seemingly leaves no place for a specific type of properties connected with vital relatedness as such. In fact Spencer's method of treatment reduces all modes of relatedness to the A type, the laws of which are, for him, the primary 'causes' of all kinds of differentiation and integration. Hence the laws of biology and psychology can ultimately be expressed and explained, he thinks, in mechanical or mechanistic terms. But in the doctrine of implication they are just the laws of B and C respectively, though laws of A may underlie them in a logical sense. And as we ascend the evolutionary plane from A to AB and thence to ABC—from the physico-chemical to the vital and thence to the cognitive—we find new modes of relatedness, new forms of more complex integration and synthesis, new properties successively appearing in serial order. This seems to me simply to express, in outline, the net result of interpretation based on empirical observation—though much, very much, requires to be filled in by future research. And the new properties are not merely additive of preceding properties; they are constitutive, and characterize the higher evolutionary products as such. Why they are thus constitutive, science is unable to say. Spencer, of course, calls in the Unknowable to supply the required nexus.[59] Otherwise, in each case, he confesses that 'we can learn nothing more than that here is one of the uniformities in the order of phenomena'.[60] None the less we may be able some day to establish an ordinal correlation[61] of cognitive processes with physiological processes, and an ordinal correlation of these physiological processes with those of the physico-chemical type. That I conceive to be the ideal of strictly scientific interpretation if it is to be raised progressively to a level approaching that of the exact sciences. It certainly is not yet attained. But I see no reason why we should not regard it as attainable. It will involve the very difficult determination of many correlation coefficients and constants—and for some of these our data are, it must be confessed, both scanty and unreliable.

We must here note a much-discussed departure on Spencer's part from his earlier position. On the first page of the Biology in the earlier editions, and in the last, we are told:

'The properties of substances, though destroyed to sense by combination are not destroyed in reality. It follows from the persistence of force that the properties of a compound are resultants of the properties of its components, resultants in which the properties of the components are severally in full action, though mutually obscured.'

There is no hint here of Mill's heteropathic laws nor of Lewes's emergents. But in the last edition a special chapter is inserted on the Dynamic Element in Life. We here find a tardy recognition of the presence of specific vital characters.

'The processes which go on in living things are incomprehensible as the results of any physical actions known to us.... In brief, then, we are obliged to confess that Life in its essence cannot be conceived in physico-chemical terms.'[62]

I speak of this as a tardy recognition; but it is one that does honour to the man; it is a frank admission that his previous treatment was in some measure inadequate, which a smaller man would not have had the honesty or the strength of character to make. Of course it is traced down to the Unknowable. 'Life as a principle of activity is unknown and unknowable; while phenomena are accessible to thought the implied noumenon is inaccessible.'[63] Still, certain specific characteristics of living organisms are explicitly recognized as among the accessible phenomena; and these cannot be conceived in physico-chemical terms. But did Spencer fully realize how big a hole this knocks in the bottom of the purely mechanical interpretation of nature he had for so long championed?

There remains for consideration the place of the cognitive relation in Spencer's philosophy of science. We need not here discuss his transfigured realism. Apart from the customary references to the Unknowable, of which what is knowable is said to be symbolic, it comes to little more than laying special emphasis on the truism that what is known in the so-called objective world involves the process of knowing; from which it follows that, apart from knowing it the objective world cannot be known. From this Spencer draws the conclusion that terms in cognitive relatedness have their very nature determined in and through that relatedness, and cannot in themselves be what they are, and as they are, in the field of cognitive symbolism. This may or may not be true. I am one of those who question the validity of the arguments in favour of this conclusion. Since, however, the philosophy of science deals only with the knowable—of which the so-called appearances with which we have direct acquaintance are the primary