264 THE ATOMIC WEIGHTS. 



to the atomic weight of oxygen. If the latter be 16, instead 

 of 15.9633, this error in some instances becomes muhipHed 

 to a large fraction of a unit, as the subjoined example will 

 show. 



If O =z i6, the atomic weight of uranium = 239.030 

 If O = 15-9633. " " = 238.482 



Difference 0-548 



Other similar errors are repeated continually. The value 

 assigned to an}^ element is necessarily affected by whatever 

 errors may attach to the atomic weights of those other ele- 

 ments through whose medium it is compared with the 

 standard, hydrogen. Thus, the atomic weight of carbon 

 depends upon that of oxygen ; calcium depends upon both 

 carbon and oxygen ; and fluorine, as determined from cal- 

 cium fluoride, involves the foregoing elements, together with 

 sulphur, silver, and chlorine. Since, however, some atomic 

 weights are afi'ected by plus errors and others by minus 

 errors, there is a fortunate tendency to compensation of 

 errors in cases like that of fluorine, and, in reality, better 

 results are obtained than considerations such as these would 

 lead us to look for. 



Another general source of error is to be found in the fact 

 that some of the weighings involved in our discussions had 

 been reduced to absolute standards, while others were merely 

 uncorrected weighings in air. The errors thus introduced 

 into the work are doubtless small, but still they ought not to 

 be absolutely ignored. 



Now, having considered the larger classes of errors, we 

 may properly pass on to a comparison of our atomic Aveights 

 with reference to Front's hypothesis. In order to facilitate 

 work, I have tabulated the figures in two columns, one 

 giving atomic weights referred to hydrogen as unity, the 

 other based upon the standard of oxygen as exactly sixteen. 

 Such imperfectly known elements as decipium, philippium, 

 samarium, terbium, and thulium are not included. 



