250 CONDILLAC reduce them all to the mathematical form. To reason is to calculate, and to calculate is to reason ; and reasoning, like calculation, comes to be a merely mechanical operation. Condillac rejects the common explanation of reasoning, that it is a comparison of two terms with a common third to find their relation to each other. He sees no need for a middle term. The force of the demonstration, he thinks, lies in the identity of the two extremes, which is made evident by decomposing them. Of the syllogism he says that it makes reasoning consist in the form of expression rather than in the development of the ideas, and that most of its rules have been framed with a view to concluding from the genus to the species, whereas thought sets out from particulars. Regarding the criterion of truth, he attacks the Cartesian test of truth, and proposes instead of it his own criterion of identity. In the Essai sur VOrigine des Connaissances he brings two objections against Descartes : (1) the methodic doubt is insufficient and even useless, because, while calling our ideas in question, it leaves them in all their indeterminateness ; (2) it is impracticable, for we cannot doubt about the relations which exist between familiar and determinate ideas like those of numbers. Of the Cartesian criterion of truth, " all that is contained in the clear and distinct idea of a thing may be affirmed of it with truth," Condillac says that it is both useless and dangerous, and should not have been extended to cases different from the one which gave it birth. For Condillac the sign of truth is identity. The evidence of a proposition is in the identity of the two terms. The evidence of a reasoning is in the identity of the successive propositions. No definition, however, is given of identity. It is said to be recognized when a proposition can be expressed in terms equivalent to these, " the same is the same." But Condillac draws a distinction between identical proposi tions that are frivolous and those which are instructive, and explains the latter to be those in which the terms are identical in thought, but different in expression. Condillac held that three kinds of evidence are needed to arrive at certainty the evidence of fact, the evidence of feeling, and the evidence of reason. The evidence of fact informs us of the relations which bodies have to us ; it can have no other object. The evidence of feeling enables us to distin guish what passes in us, the modes or states of the mind. Some good remarks are made in chapter iii., part i., of L Art de Penser, on the attentive observation of conscious ness. The evidence of reason is discussed in the first three chapters of the first part of L Art de Eaisonner. There Condillac merely formulates the principle of identity, and cites as examples the geometrical theorems which his pupil will require, that he may understand the rest of the work. Condillac was of opinion that one method of analysis is common to all the sciences. Our cognitions ought to form a system in which all is strictly connected together. Every series of facts should be reduced to an initial fact, of which the others are only transformations. Identity is a rule, of method as well as a criterion of certainty ; and analogy completes the primary lessons which are given us by nature. Condillac takes as his model the method of mathematics, and reiterates through his logical writings that we must take nature for our guide. On the relation of analysis to language he held that there is an innate language, although there are no innate ideas. This language produces a kind of analysis, since it is necessary for the communication of our ideas to analyze and express in succession what is simultaneous in thought. Analysis then reacts on language and improves it. Finally, perfec tion of language leads to perfection of analysis, and science ti> only a langue lien faite. The method of invention is discussed chiefly in the Essai sur I Origins des Connaissances and in the Langue des Calculs. In the former, Condillac bids us take the simple ideas furnished by sensation and reflection, form different collections of them, winch in their turn will produce others, and give distinct names to these different collections. In the Langue des Calculs the idea of analogy is .developed. This is not the analogy set forth in the Art de Raisonner, which consists only in forming more or less probable conjectures about the unknown from the known. The analogy of the Logique and the Langue des Calcids is that which creates and regulates languages, which causes us to invent different systems of signs and submit them to uniform rules. Reasoning cannot have the purely subjective character which Condillac s theory assigns to it. It takes its departure from the idea, which is objective, and therefore establishes a real relation between the mind and its object. On the question of the need for a middle term it is not enough to decompose the two ideas. The two decomposi tions must meet at a point, and that is the middle term. Laromiguiere preserves the intermediate ideas, which he thinks are found by analysis of the extremes. So they are. But it has been urged that there is a twofold analysis of the species into its genera and of the genus into its species whereby the middle term is found. Condillac is incon sistent with himself in his criticism of Descartes. His first objection to the methodic doubt is based on the opinion that all our errors proceed from the indeterminate character of language, and that the use of definite signs is the only security against error. But he believes that analysis makes language ; and the methodic doubt is a kind of analysis, for it remounts to the primary truths. His second objection, that we cannot doubt about mathematical rela tions, is invalidated by his own statement that mathe matics are only part of metaphysics. Condillac is right in saying that the Cartesian criterion of truth lacks a theory of ideas and of their origin. But it is not to be condemned as useless because it is incomplete. Condillac was led away by the supposed need for a sign whereby to recognize truth. As Hegel would have put it, he refused to go into the water until he could swim. But it would be as difficult to determine the value of the sign as that of the truth itself. Some such criterion as identity is the only resource of empiricism. But if the notion of identity is derived from experience, it cannot give certainty. If, that it may serve as the basis of logic, it is regarded as necessary, then empiricism cannot reconcile it with its psychology. Laromiguiere tries to get over the difficulty of accounting for progress in a system based on the notion of identity, by drawing a distinction between partial identity and total identity, and saying that the former alone should be admitted. But what is partial identity 1 Condillac himself takes refuge in extreme idealism. Truth, he says, considered in itself and in the divine intelligence, is one and identical. But he had himself laid down the rule to limit our consideration to the condition of human know ledge, and of course he had no idea of developing thought as such from its primal unit by a dialectic process after the manner of Hegel. As to the three kinds of evidence, Condillac in reality reduces the evidence of fact to that of feeling and that of reason. His numerous contradictions are largely due to his attempts to defend the authority of the senses, while he accepts the idealistic theory of external perception. The objections to identity as a criterion of truth apply as well to Condillac s statement of the evidence of reason. And if the three kinds of evidence are inadequate taken separately, they cannot suffice when combined. Condillac rightly insisted that there is one
fundamental method for all the sciences ; but he nowhere