Note 548 — Chapter XXVIII

Some readers may expect me to plunge into the old debate as to whether the a priori truths are 'analytic' or 'synthetic.' It seems to me that the distinction is one of Kant's most unhappy legacies, for the reason that it is impossible to make it sharp. No one will say that such analytic judgments as "equidistant lines can nowhere meet" are pure tautologies. The predicate is a somewhat new way of conceiving as well as of naming the subject. There is something 'ampliative' in our greatest truisms, our state of mind is richer after than before we have uttered them. This being the case, the question "at what point does the new state of mind cease to be implicit in the old?" is too vague to be answered. The only sharp way of defining synthetic propositions would be to say that they express a relation between two data at least. But it is hard to find any proposition which cannot be construed as doing this. Even verbal definitions do it. Such painstaking attempts as that latest one by Mr. D. G. Thompson to prove all necessary judgments to be analytic (System of Psychology, ii. pp. 232 ff.) seem accordingly but nugæ difficiles, and little better than wastes of ink and paper. All philosophic interest vanishes from the question, the moment one ceases to ascribe to any a priori truths (whether analytic or synthetic) that "legislative character for all possible experience" which Kant believed in. We ourselves have denied such legislative character, and contended that it was for experience itself to prove whether its data can or cannot be assimilated to those ideal terms between which a priori relations obtain. The analytic-synthetic debate is thus for us devoid of all significance. On the whole, the best recent treatment of the question known to me is in one of A. Spir's works, his Denken und Wirklichkeit, I think, but I cannot now find the page.