Note 388 — Chapter V

Which is added to his Theory of Moral Sentiments, edit. 1822, 2 volumes. Compare a letter which Smith wrote in 1763 on the origin of language (in Nichols's Literary Illustrations of the Eighteenth Century, vol. iii. pp. 515, 516), which exhibits, on a small scale, the same treatment, as distinguished from a generalization of the facts which are supplied by a comprehensive comparison of different languages. Dr. Arnold speaks slightingly of such investigations. He says, ‘Attempts to explain the phenomena of language a priori seem to me unwise.’ Arnold's Miscellaneous Works, p. 385. This would lead into a discussion too long for a note, but it appears to me that those a priori inferences are, to the philologist, what hypotheses are to the inductive natural philosopher; and if this be the case, they are extremely important, because no really fruitful experiment ever can be made unless it is preceded by a judicious hypothesis. In the absence of such an hypothesis, men may grope in the dark for centuries, accumulating facts without obtaining knowledge.