360 HINRICHS— THE TRUE ATOMIC WEIGHTS [December 2. 



recent determinations to the extent of over one tenth of one per cent. 



This is not the place for renewing the discussion in detail ; the 

 publications in question are accessible to all. We would rather take 

 up the question in the broadest way and try to decide it by a sort 

 of crucial test. We znnll shozv that the evidence on zvhich the value 

 16 for oxygen rests is exactly the same as that zvhich gives the value 

 108 for silver. 



It is a fact that all chemical reactions are approximately exact 

 only and that all laboratory work is subject to the same limitation. 

 It is therefore incorrect to assume that in the reactions and in the 

 laboratory work there is involved no error whatever on the part of 

 the oxygen and that all errors are due to the other elements asso- 

 ciated with oxygen. It seems sufficient to state this to have it 

 admitted. 



But if oxygen be supposed free from all material imperfections, 

 as is assumed in the common calculations, its actual shortcomings 

 are not blotted out thereby — they are simply placed to the account 

 of the other elements which are unfortunate enough to be in reaction 

 with the supposedly perfect element, assumed to be immune from 

 error by the school. Hence the errors of the associated elements 

 will be correspondingly magnified. In this way, the dominant school 

 now has arrived at the value 107.88 for silver. 



Let us try to see how this has been brought about and how the 

 question may be put in the above crucial form for decision. 



It is agreed by all and not even denied by the American School 

 that the atomic weight, for O^ 16, is very near the round numbers : 

 Ag 108, CI 35.5, C 12, N 14, etc. These values we have called the 

 absolute atomic zveights. 



Accordingly, the true atomic zveights can differ from these abso- 

 lute values by small fractions only; this we call the departure and 

 designate by the letter epsilon (e). 



Hence the real mathematical problem is the determination of the 

 departure for each clement in every reaction used. We determine 

 the departure in thousandths of the unit. Thus, if the American 

 Chemical School be right in declaringthat the atomic weight of silver 

 is 107.88, the departure for silver would have the enormous value 



