250 



PROCEEDINGS OP THE AMERICAN ACADEMY 



several samples from different sources was much larger than that 

 demanded by the old atomic weight, and closely approached the 

 theoretical quantity based upon the new value. 



In the third experiment the weight of water lost by heating the 

 salt to constant weight in the manner described by Hampe * was also 

 determined, and this amount proved to be nearly three tenths of a per 

 cent less than it should have been according to any hypothesis. Since 

 larger crystals of cupric sulphate exposed to air above the desiccating 

 mixture already mentioned lost none of their lustre, and since the salt 

 under these conditions comes to perfectly constant weight, it seemed 

 highly probable that the apparent deficiency of crystal water was due 

 to occlusion in the so-called anhydrous salt, and not to any previous 

 efflorescence of the crystals. 



The sulphuric acid produced during the electrolysis f was preserved 

 in each case, and approximately determined by means of the first prep- 

 aration of sodic carbonate ; also afterwards by the evaporation of the 

 solution of sodic sulphate. Since the manipulation was more or less 

 imperfect, and the materials and water not quite pure, it is thought 

 unnecessary to include these somewhat bulky data in full here, but 

 the evidence which they furnished was conclusive upon one point. 

 The percentage of acid was not equal to the complement of the other 

 percentages. In other words, over one tenth of a per cent of material 

 liad not been determined. 



It is seen that we are dealing, not with infinitesimals, but with grave 

 error evident to the crudest analysis. The reason for the deficiency 

 has already been suggested, but the proof is yet wanting. 



* Loc. cit. 



t In tliis ronnection it may he noted tliat Hart and Croasdale liavo quite 

 independently used tiie sulphuric acid tims formed as a basis of alicalinietric 

 analysis. (Cliem. News, LXIII 93, 1891.) The idea appears to be a good one 



