THIB little book, whose value is altogether out of proportion to its size, contains the dearest and most plausible account that Prof. Bosanquet has yet given of his views on logic The author has made a careful study of recent writers whose general position differs considerably from his own, such as Dr. Mercier, Eusserl, and Mr. Leonard Bussell; and much light is thrown on his own system by his discussion of theirs. In particular it is pleasant to see that at least one English philosopher of
... philosopher of eminence recognises the importance of Husserl's work, which has been strangely neglected here, possibly on account of its extreme prolixity and its barbedwire entanglement of new technical terras. Prof. Bosanquet is concerned to maintain that inference is everywhere of the same general type, and that it is not subsumptive or syllogistic The true type is explained under the name of vmplicaturn; subsumption he calls linear inference, and condemns as a ' second-hand' process of argument. The book falls into three closely connected Darts. Chapters I., IV., V., and VIII. explain the nature of implication; and exhibit its connexion with induction, judgment, supposition, and the contrast between the necessary and the contingent. Chapters II. and III. deal with the linear view of inference, and claim to show that most of the oritics of the syllogism have never freed themselves from ita domination. Chapters VI. and VII. deal with points that are somewhat less vital to Prof. Bosanquet's argument, viz., the constant use of sets of three terms or propositions in inference, and the question whether logic has any special connexion with the study of minds and their processes. The essence of this theory of inference seem to be the following. We start with some complex of related terms. This may either be actual or merely supposed. The relating relation that characterises this complex will be such that each term in the complex is relevant to all (or, at any rate, to many of) the other terms. Such complexes are what Prof. Bosanquet means by universals. I may remark, in passing, that this explains, as I had long suspected, why Prof. Bosanquet asserts many propositions about universals which seem to people brought up on a different nomenclature to be patently false. What he says about universals is both true and important when the name is understood in his sense, and false at Frankfurt Univesity Library on May 28, 2015 http://mind.oxfordjournals.org/ Downloaded from 324 CBmcAL NOTICES: when it is understood in the sense of abttractum. The only ground of quarrel that remains is that he seems to deny that universals, in the latter sense, are also real and important; but this is a matter that concerns his large Logic rather N tban the present work. Still, even when it is understood that univeitoals are to mean complexes, it seems to me that Prof. Bosanquet's theory requires universals in the sense of abttracta. For I imagine that what is important is, not some one particular complex, but the characteristic type of structure of all the possible complexes of a class. This, I think, is implied by the fact that what we should commonly call the same complex varies its terms and their relations, within limits, in determinate and interconnected ways. This is assumed by ProL Bosanquet, and seems to imply a contrast between the permanent general type of structure-an universal in the sense of an abttractum -and the determinate complex as it is at a given moment (if it be in time) or distinct instances of it (if it be timeless, as in geometry), which are universals in Prof. Bosanquet's sense. Implication is defined as the relation which subsists between one term or relation in such a complex and the rest, in so far as their respective modifications afford a due to one another. The position then is that if one term or relation in a complex of a certain general structure varies (presumably within the limits required for the complex to remain of the same structure), there will be correlated variations in some or all of the other terms and relations. It appears from the definition that this state of affairs is not itself implication, but is only a precondition for it. For implication it is not enough that modifications in different parts of the complex should in fact be correlated, they must further be so correlated that one ' affords a clue to' the other. Prof. Bosanquet thus agrees BO far with logicians of the Russell-Whitehead type as to regard implication as a relation between terms which subsist whether a mind recognises it or not. He differs in so far as they make implication a very special relation that holds only between proposition*. It is doubtful whether this difference is very important. I take it that the connexion between the two senses of implication is this. The proposition that asserts that such and such a term or relation in a certain complex is modified in a certain way is connected by ' implication,' in the Russell-Whitehead sense, with a proposition asserting that some other term or relation undergoes a correlated variation. The connexions of the actual terms or relations in the complex, in virtue of which the two propositions imply each other in this sense, are ' implications' in Prof. Bosanquet's sense. Thus the oonnexion would seem to be that Prof. Bosanquet's implication is that relation within a factual complex which is the factual correlate of implication, in the Bussell-'Wnitehead sense, between propositions about terms or relations within this complex. We next come to Prof. Bosanquet's use of the word inference. This seems to be bound up with a special theory as to the precise way in which inferences are made. His view is the following, if I at Frankfurt Univesity Library on May 28, 2015 http://mind.oxfordjournals.org/ Downloaded from BOSANQUET, Implication and Linear Inference.