Page:Philosophical Review Volume 31.djvu/497

No. 5.] labours, yet will no great progress ever be made in science by means of anticipations; because radical errors in the first concoction of the mind are not to be cured by the excellence of functions and remedies subsequent." This estimate of intellect is in sharp contrast to Aristotle's νοῦς which is represented as the guarantee of certainty. But Bacon was not a sceptic. Like Descartes, he doubted not for its own sake but so as to reach truth. And, most of all, he doubted, in opposition to Scholasticism, the ipse dixit of Aristotelian authority. Bacon's distrust sometimes tends to overreach itself. If "radical errors in the first concoction of the mind are not to be cured," does not Bacon veto the possibility of reaching truth, not only by hypothesis but, so long as activity of intellect in some mode is necessary to knowledge, by any means whatever? Strictly, is not his own philosophy of discovery already refuted?

Bacon's neglect of hypothesis is further seen in his view of the relation of fact and theory. For the speedy advancement of scientific knowledge, Bacon proposed a division of labour. Let observers collect facts. Let theorists evolve the laws governing the facts. Now modern science has taught that fact and theory cannot be severed in this way. A 'working hypothesis' is necessary. With it we select the facts necessary for its proof or disproof. This selection of facts affords a greater economy of effort than Bacon, who supposed natural phenomena limited, instead of infinite, in number, could have held.

Referring to Bacon's method of procedure in connection with the investigation of heat, Craik has said that that method "sets out simply with a blind accumulation of instances, no more collected under the guidance of any kind of anticipation or hypothesis than are the fishes, great and small, that the net brings up when cast into the sea." This is untrue. Bacon's method involves Three Tables of Instances. These Tables are formed under the guidance of three corresponding principles. And these principles are, as it were, hypotheses as to the best general mode of