138 THE ATOMIC WEIGHTS. 



roasting, and the ratio between the weights of the two sub- 

 stances was determined. 



By the first method it was found that 100 parts of M0O3 

 will expel the following quantities of CO 2 : 



31-4954 

 31-3749 

 31-4705 



Mean, 31.4469, zfc .0248 



The carbon dioxide was determined simply from the loss 

 of weight when the weighed quantities of trioxide and car- 

 bonate were fused together. It is plain that if, under these 

 circumstances, a little of the trioxide should be volatilized, 

 the total loss of weight would be slightly increased. A con- 

 stant error of this kind would tend to bring out the atomic 

 weight of molybdenum too low. 



By the second method, the conversion by roasting of MoS, 

 into M0O3, Svanberg and Struve obtained these results. 

 Two samples of artificial disulphide were taken, A and B, 

 and yielded for each hundred parts the following of trioxide: 



89-7919 1^ 

 89.7291 i 



89.6436 ^ 

 89.7082 I 

 89.7660 

 89.7640 

 89.8635 



-B. 



Mean, 89.7523, ± .0176 



Three other experiments in series B gave divergent re- 

 sults, and, although published, are rejected by the authors 

 themselves. Hence it is not necessary to cite them in this 

 discussion. We again encounter in these figures the same 

 source of constant error which apparently vitiates the pre- 

 ceding series, namel}^ the possible volatilization of the 

 trioxide. Here, also, such an error would tend to reduce 

 the atomic weight of molybdenum. 



Upon discussing the data given in the foregoing para- 



