ch. v.] TEE TWO METHODS. 131 



insisted more strenuously than Comte upon the necessity of 

 distinguishing between legitimate and illegitimate hypotheses, 

 or who have more clearly prescribed the conditions under 

 which alone can any given hypothesis be regarded as legiti- 

 mate. Unfortunately, by a strange and ironical fate, the 

 writer who contributed so much toward the establishment of 

 sound methods of philosophizing, lived to become a proficient 

 in the subjective method, a pitiless scorner of crucial experi- 

 ments, and a weaver of vagaries which might well be matched 

 with those above cited from Plato and Hegel. The historical 

 importance of this phenomenon is great enough to justify us 

 in treating it at some length. 



Though in Comte's earlier works a somewhat obtuse sense 

 of the requirements of verification is now and then to be 

 noticed ; and though there is a tendency, which visibly in- 

 creases toward the end of the " Philosophic Positive," to sub- 

 stitute intensely dogmatic ex cathedra dicta in the place of 

 arguments ; yet the necessity for strict obedience to the 

 objective method is nowhere explicitly denied. It is in- 

 sisted, with entire justice, that every hypothesis which does 

 not admit of verification should be remorselessly discarded 

 from philosophy; and that even a verifiable hypothesis 

 should never be incorporated as a part of philosophy or 

 science until it has been actually verified. Far different is 

 the attitude taken by Comte in his later works, when he is 

 attempting to reconstruct society. In the " Politique Posi- 

 tive " he begins by endeavouring to reinstate the subjective 

 method ; deluding himself, by a play upon words, into the 

 belief that that method can be so reformed as to become 

 available in the search for positive truths. " The subjective 

 method," he tells us, " possesses striking advantages which 

 can alone compensate for the inconveniences of the objective 

 method." This unhappy sentence is of itself enough to show 

 how far the writer had strayed from positive grounds. Here 

 we see the necessity for constant verification characterized 



K 2 



