OP ARTS AND SCIENCES. 



263 



Water . 

 Copper 

 SO4 . 



Uncorrected. 



36.007 

 25.449 

 38.435 



99.951 



Corrected. 



36.067 

 25.452 

 38.442 



99.961 



That is to say, an uudoubtedly excessive value for this correction 

 would account for only one fifth of the deficiency for which explana- 

 tion was sought. The correction is at best very doubtful, and it is 

 not applied in the final calculation of the atomic weight. Its appli- 

 cation would make no essential difference in the final result, but 

 would render the individual variations much less marked. 



The only probable interpretation of the deficiency now lay in the 

 assumption that cupric sulphate still held a volatile impurity at 360°. 

 Such an amount of any non-volatile foreign substance would have 

 increased the weight of sodic sulphate found at the conclusion of the 

 analysis by about eight one-hundredths of one per cent, an amount 

 entirely too large to escape detection. Besides, such a source of error 

 had already been shown to be unlikely. 



After it had been experimentally proved that cupric sulphate had 

 no tendency to hold ammonic sulphate at 370°, the hypothesis of the 

 occlusion of a small amount of water became the last resource. At 

 first sight it seems improbable that any material could hold appreciable 

 quantities of water in the presence of sulphuric acid when so strongly 

 heated. But it must be remembered that at this temperature the acid 

 is dissociated, and water is actually present in the vapor. It was 

 hoped that sulphur trioxide might be more efficient as a dehydrating 

 agent, but a single experiment (No. 18) showed that cupric sulphate 

 possessed more affinity for water even at 300° than did the sulphur 

 trioxide. 2.3787 grams (in vacuum) of cupric sulphate lost 0.8554 

 gram in weight on heating to about 300° in a glass tube under a 

 current of dry air. Upon continuing the application of heat in air 

 charged with sulphuric anhydride from Nordhausen acid, no essential 

 change in weight was observed. The apparatus was somewhat com- 

 plicated in order to avoid rubber connections, but a description of it is 

 superfluous. The total loss was 35.962 per cent of the weight taken. 



