APPENDIX. 269 



discussion. For chromium, copper, molybdenum, and vana- 

 dium I have no criticisms to offer ; but the remaining ele- 

 ments may be considered individually. 



The value assigned to antimony, 120.231, is the general 

 mean of Cooke's and Schneider's work upon the bromide, 

 iodide, and sulphide. If Ag = 108, Br = 80, and I = 127, 

 Cooke's data for the bromide and iodide give the following 

 values for Sb, all of which fall within a tenth of a unit of 

 the whole number 120 ; 



Early bromide series Sb = 119. 901 



* Late " " =120.009 



Iodide series " = 119.973 



In the case of cerium, the value assigned in the table is 

 the general mean of all reputable determinations. But it is 

 subject to doubt on account of the facts observed by Wolf 

 and by Wing, whose ceroso-ceric oxide was white, while 

 that of all other observers was yellowish. Wolf's and 

 Wing's data, calculated with = 16, give Ce = 138.039. 

 Cerium, then, is not an established exception to Front's law. 



Glucinum and ytterbium have their atomic weights cal- 

 culated from analyses of the sulphates. But if Front's law 

 is true, SO3 = 80. Calculated with this figure, we have 

 Gl = 9.096 and Yb = 173.016. Both elements thus fall 

 within reasonable limits of variation from the hypothetical 

 values. 



Iodine is one of the most important seeming exceptions. 

 If we assume Ag = 108, and calculate the atomic weight of 

 iodine only from the direct ratio between iodine and silver, 

 we have, with Dumas' correction applied, I = 126.966 ; 

 that is, it comes within one-tenth of a unit of the whole 

 number 127. 



The atomic weight of mercur}^ depends upon analyses of 

 the chloride, oxide, and sulphide. Of these three com- 

 pounds the purity of the chloride is most easily assured. 

 Calculated from its composition, with CI = 35.5, lig = 

 199.971. With so high an atomic weight small errors are 

 easily multiplied. 



