186 



Dr. E. J. Mills and J. J. Smith. 



[June 19, 



reagents before and after mixture. The weight of precipitated sodic 

 sulphate varies but little throughout the experiments, its mean value 

 being "0115 grm. In a special determination, we precipitated, with 

 the usual amount of sodic hydrate, l'OOO grm. of nickelous sulphate 

 alone : after three days' washing, it was found to contain '0115 grm. 

 of sodic sulphate. Hence we infer that when the two sulphates are 

 present together, nickelous hydrate carries down sodic sulphate, 

 cobaltous hydrate carrying down cobaltous sulphate. 



IV. Discussion. 



In the following discussion, it will be understood that we refer to 

 the tabular results already given. 



If n represent a weight of nickelous sulphate taken, and v be the 

 hydrate (calculated to sulphate) obtained from it through precipitation, 

 then we consider in the expression n—(pv to represent the precipita- 

 bility of nickelous sulphate : and similarly in c=07, represents the 

 precipitability of cobaltous sulphate. In examining the numbers 

 obtained with the nickelous salt, the best expression we could find for 

 was 0=(a + /3?t). We first calculated the values of a and /3u 

 from all the determinations, and by the method of least squares, thus 

 obtaining 



0=-98891 + -22571W. 



It is, however, clear that a cannot be less than unity ; moreover, the 

 first weight of nickelous precipitate is somewhat higher than is pos- 

 sible, and we have thought well to reject it. With these amendments 

 we finally obtain 



0=1 + -21940ft, 



with a probable error of "02558 for a single determination, or "00904 

 for eight determinations. Hence we infer that the precipitability of 

 nickelous sulphate is directly proportional to its mass. 



In the case of cobaltous sulphate, on the other hand, no such law 

 holds good. After a very careful examination of the numbers, we 

 could not find any satisfactory evidence' of a change in precipitability 

 with its mass, and consequently represent as a constant. The 

 mean value of is in this case 1 - 1845, with a probable error of "02792 

 on a single determination, or "00931 for nine determinations. Thus 

 is about equally well ascertained in both cases. 



Our two equations may now be written 



?i=( + "21940 n)v\ 

 c=l-18457 ) ' 



In order to calculate in what proportions the two sulphates are 

 equally precipitable, we have 



1 + "21940/^=1-1845, 



