I40 THE GRAMMAR OF SCIENCE 



demonstrable certainty, applies only to the sphere of con- 

 ceptions. What are we, then, to understand when the 

 word proof is applied to natural phenomena ? Shall we 

 say that it is incorrect to use the word prove at all in 

 such relationship ? Yet our leading men of science do 

 use it. Here is a passage from Lord Kelvin's lecture on 

 " The Six Gateways of Knowledge." ^ He is discussing 

 the possibility of our having a " magnetic sense," and he 

 writes : — 



" I cannot think that that quality of matter in space 

 — magnetisation — which produces such a prodigious effect 

 upon a piece of metal, can be absolutely without any — 

 it is certainly not without any — effect whatever on the 

 matter of a living body ; and that it can be absolutely 

 without any peixeptible effect whatever on the matter of a 

 living body placed there, seems to me not proved even 

 yet, although nothing has been found." 



The word prove is here distinctly used of something 

 being demonstrable in the field of perception. There is 

 clearly an inference involved, and this inference is easily 

 seen to be that of the routine of perceptions, namely, 

 that if something has once been perceived, it will under 

 precisely the same circumstances be again perceived. 

 Our conviction of this routine is not a certainty, but, as 

 we have seen, a probability. Hence, when we are speak- 

 ing of the sphere of perceptions we must remember that 

 provable is ultimately the same word as probable. The 

 association of the two words does not therefore seem 

 without profit ; and the etymology may after all serve to 

 remind us of the character of our knowledge in the field 

 of perception. 



The problem before us is the following one : A certain 

 order of perceptions has been experienced in the past, 

 what is the probability that the perceptions will repeat 

 themselves in the same order in the future ? The prob- 

 ability is conditioned by two factors, namely: (i) In 

 most cases the order has previously been very often re- 

 peated, and (2) past experience shows us that sequences 



1 Popular Lectin-es and Addresses, vol. i. p. 261. London, 1889. 



