THE CONNECTION BETWEEN THE ETHEK AND ^lATTER. 



By M. Henri Poincahe. 



When M. Abraham came to me and asked that I close this series of 

 sessions of the Societe franpaise de Physique, I was at first incHned 

 to refuse. It seemed as if each subject had been completely dis- 

 cussed and that I could have nothing to add to that which had 

 already been so well said. I could only try to put in words the impres- 

 sion which seemed to emerge as a summary of all the discussions, and 

 that impression was so definite that each of you must have felt it. I 

 did not see how I could make it any clearer by forcing myself to put 

 it into words. But M. Abraham insisted with such good grace that 

 I resigned myself to the inevitable difficulties of which the greatest 

 is to repeat what each one of you has long since felt, and the least is 

 to run through a maze of diverse subjects without the time to dwell 

 on any one of them. 



One thought must at once have struck all those present. The old 

 mechanical and atomic hypotheses have, during recent years, become 

 so plausible that they have ceased to seem like hypotheses; atoms 

 are no longer just a convenient fiction. It seems almost as if we 

 could see them, now that we know how to count them. A theory 

 assimies reality and gains in })robability when it accounts for new 

 facts. Yet this may result in difl'erent ways. Generally it has to 

 be enlarged to include the new data. Sometimes it loses m precision 

 as it becomes broader. Sometimes it becomes necessary to engraft 

 upon it an accessory hypothesis which plausibly fits in with it, but 

 which nevertheless is somewhat foreign to it, and contrived expressly 

 to fit a certain case. Then it can scarcely be said that the new 

 facts confirm the original hypothesis, only that they are not incon- 

 sistent with it. Or, again, there may be between the new facts and 

 the old, for which the hypothesis was originally conceived, such an 

 intimate connection that whatever theory renders account of one must, 

 because of that connection, render account of the other as well. 

 Then the new data which fall in with the old are really only appar- 

 ently new. 



• All address delivered before the Society fran^aise de Physique, April 11, 1912. Jleprinted by permission 

 from Journal de Physique, Paris, 5th series, vol. 2, May, 1912. 



199 



