270 THE ATOMIC WEIGHTS. 



For the atomic weight of platinum Seubert's data give 

 five values, ranging both above and below the round number 

 195. Calculated with integer values for the other elements, 

 three of these figures fall ver}^ close to 195, as follows : 



From per cent. Pt in (NHJ^PtClg Vt = 194.906 



KjPtCle " = 194-933 



From chlorine estimation in K^PtClg " =^ 194-955 



Potassium is the most serious exception of all. But if 

 = 16 and Dumas' correction be ai^plied, the general mean 

 from all the available data becomes K = 39.083. That is, 

 potassium falls within the limit of 0.1 variation. 



The atomic weight assigned to tantalum is the mean of 

 four values. Two of these, recalculated with integers, come 

 out as follows : 



From per cent. KjSO^ in K.^TaF, Ta = 181.912 



Ta^Oj from (NHJ^TaF, " = 182.020 



For tellurium I need only call attention to the discrepan- 

 cies between the several sets of determinations made by- 

 Wills. "A reference to the chapter on tellurium will show 

 that his figures give results ranging from Te = 126.07 to 

 Te = 129.34. The mean value is therefore too much sub- 

 ject to doubt to carry weight as an exception. 



As for thallium, the last case to be considered, I have 

 already shown that Crookes' data, recalculated with integer 

 values for N and 0, give Tl = 204.008. That is, instead of 

 an exception, we have here an admirable instance in sup- 

 port of Front's hypothesis. 



Enough has been said in this brief resume to show that 

 none of the seeming exceptions to Front's law are inexpli- 

 cable. Some of them, indeed, carefully investigated, sup- 

 port it strongly. In short, admitting half multiples as 

 legitimate, it is more probable that the few apparent excep- 

 tions are due to undetected constant errors, than that the 

 great number of close agreements should be merely acci- 

 dental. . I began this recalculation of the atomic weights 



