OF ARTS AND SCIENCES. 249 



tilled in a platinnra still, it appeared to be very pure. The first por- 

 tions of the distilhite wei-e rejected, although no ammonia was found 

 in them. The water left absolutely no trace of residue upon evapora- 

 tion, even after long standing. Such doubly distilled water was used 

 in the second series of experiments. 



For the third series all the water used, even for minor operations, 

 was distilled four times : first alone, then over alkaline potassic per- 

 manganate, next over acid potassic sulphate, and finally once more 

 alone in the platinum still. Since this water was not different in 

 any of its properties from the second preparation, the last refinements 

 were probably unnecessary ; and in later work water which had been 

 distilled only twice was used. 



The value of the present work has been largely increased by the 

 unlimited supply of platinum ware placed at the disposal of the writer 

 through the kindness of Professor Cooke. Because of the large size 

 of some of the retorts and bottles, it was impossible to free these vessels 

 from iron after the method of Stas. Accordingly they were digested 

 alternately with strong hydrochloric and nitric acids until pure acid 

 which had remained in them for days gave no test for iron after 

 evaporation upon tiie steam bath. 



Smaller vessels were freed from iron in the usual manner, by treat- 

 ment with the vapors of aramonic chloride at a red heat, as well as 

 with fused acid potassic sulphate. They were usually protected dur- 

 ing ignition by an outer crucible, from which they were separated by 

 a coil of platinum wire. Crucibles weighing twenty grams rarely 

 showed a variation of more than one twentieth of a milligram between 

 the weights taken before and after any ordinary operations, except- 

 ing when ignited with cupric oxide under conditions which will be 

 explained later. 



Description of Analyses. 



First Series. 



It is evident that the most probable constant error in Hampe's 

 work lay in the great hygroscopic power of cupric sulphate dried at 

 250°. A possible means of detecting such error existed in the use of 

 the crystallized salt as a starting point. This salt was therefore 

 coarsely powdered, dried to constant weight over a mixture of crys- 

 tallized and partially dehydrated cupric sulphate, and analyzed by 

 electrolysis in the usual fashion. Although the crystalline powder 

 must have contained occluded water, the quantity of copper found in 



