In the new edition of his Principles (1922) Bradley repudiates his previous assumption of ideas divorced from judgments, and of judgments divorced from inferences. The concrete conceptual fact is inference, of which mere ideas and simple judgment are abstractions. Inference is the ideal self-development of a given object taken as real. The given object is an ideal content before us, and it is taken to be real as being in one with reality, or the real universe. The possibility of inference rests upon the fact that the object is not only itself, but is also contained as an element in a whole; in fact, it is itself only as being so contained. From the nature of the case inference must always be incomplete, and subject to unknown conditions. But this ultimate problem does not concern Logic.
is, perhaps, more able than Bradley has shown himself, to use material from alien sources and to penetrate to what is of value in the thought of writers from whom he dis agrees. He treats the book-tradition, however, with a judicious exercise of freedom in adaptation, i.e., constructively as datum, never eclectically. In his fundamental theory of judgment his ob ligation is to Bradley. It is to Lotze, however, that he owes most in the characteristic feature of his logic, viz., the systematic de velopment of the types of judgment and inference from less ade quate to more adequate forms. His fundamental continuity with Bradley may be illustrated by his definition of inference. "In ference is the indirect reference to reality of differences within a universal, by means of the exhibition of this universal in differ ences directly referred to reality." In the new edition of his Logic (191 1) Bosanquet defends at length his metaphysical absolutism and the coherence theory of Truth. No finite individual is self-dependent or self-contained, that is, a real substance. Even the self finds its reality in some thing beyond itself. Hence absolutism. But an individual not self-complete can be predicated of the whole of which it is a part. With regard to Truth he maintains that Truth is its own criterion, and can only be tested by more of itself. Any system can be tested further only by being made more complete. The coherence theory of Truth is further elaborated by Bosanquet in his Implication and Linear Inference (192o). All inference, it is maintained therein, is of the same type, namely, implication. The essence of implication is interrelation between the parts of a system (or concrete universal), which makes modification in some parts clues to the modification of others. Syllogistic inference, he says, is linear, not systematic, and therefore not true infer ence. In true inference we survey a system of facts, see it in its relation to the whole of reality, and read off the implications. The inference is immediate if we can read off the implications directly; it is mediate if we must construct this system first. The
starting point of all inference is the realization that to deny the truth of all propositions would involve self-contradiction, as it would deny this very denial. We must consequently believe some propositions, even if we question the truth of this or that propo sition. In true inference we transfer our certainty that some propositions are true to the truth of the proposition inferred. In the last resort, all inferences may be reduced to the alternatives. "Either this proposition is true, or no proposition is." Within the system from which we start we read off conclusions the denial of which would shatter the whole world of our experience.
Of epistemological logic in one sense of the phrase, Lotze is still to be regarded as a typical exponent. Of another type Chr. Sigwart (q.v.) may be named as representative. Sig wart's aim was to "reconstruct logic from the point of view of methodology." His problem was the claim to arrive at proposi tions universally valid, and so true of the object, whosoever the individual thinker. His solution within the Kantian circle of ideas, was that such principles as the Kantian principle of causality were justified as "postulates of the endeavour after complete knowledge." What Kant has shown is not that irregular fleeting changes can never be the object of consciousness, but only that the ideal consciousness of complete science would be impossible without the knowledge of the necessity of all events. "The uni versal presuppositions which form the outline of our ideal of knowledge are not so much laws, which the understanding pre scribes to nature . . . as laws which the understanding lays down for its own regulation in its investigation and consideration of nature. They are a priori because no experience is sufficient to reveal or confirm them in unconditional universality; but they are a priori . . . only in the sense of presuppositions without which we should work with no hope of success and merely at random and which therefore we must believe." Finally they are akin to our ethical principles. With this coheres his dictum, with its far-reaching consequences for the philosophy of induction, that "the logical justification of the inductive process rests upon the fact that it is an inevitable postulate of our effort after knowl edge, that the given is necessary and can be known as proceeding from its grounds according to universal laws." It is character istic of Sigwart's point of view that he acknowledges obligation to Mill as well as to Ueberweg. The transmutation of Mill's in ductions into a postulate is an advance of which the psychological school of logicians have not been slow to make use. The compar ison of Sigwart with Lotze is instructive, in regard both to their agreement and their divergence as showing the range of the epis temological formula.