RICHARDS AND BAXTER. 



ATOMIC WETGHT OF IRON. 



255 



heated by a Fletcher furnace to about 900°. Constant weight within 

 a few hundredths of a milligram was obtained without difficulty. The 

 weighed oxide was then reduced in a current of electrolytic hydrogen 

 which had been dried by means of fused potassic hydroxide. Ferric 

 oxide is only very slowly reduced at the highest temperature which can 

 be used with glass tubes, so that this reduction also was of necessity 

 carried on in porcelain tubes. Even at about 900°,* long continued 

 heating was needed to complete the reduction, the total period of ignition 

 amounting sometimes to twenty hours. The progress of the reaction was 

 slightly accelerated by alternate oxidation and reduction. 



Two analyses, carried out in the fashion described, are tabulated be- 

 low. Corrections to a vacuum standard have been applied by adding 

 0.00009 gram to every apparent gram of ferric oxide, and 0.00001 

 gram to every apparent gram of metallic iron, the specific gravities of 

 these two substances being assumed to be 5.2 and 7.9 respectively, t 



SERIES I. 



The fact that both of these results are considerably lower than the 

 accepted value of the atomic weight of iron (5G.0) seemed to confirm a 

 suspicion which had already arisen that ferric hydrate might not be 

 completely converted into the oxide by ignition at a high temperature. 

 That the oxide upon ignition easily reached constant weight is, however, 

 evidence in the other direction. Since no direct method of proving the 

 presence or absence of retained water was at hand, it seemed best to 

 make use of another method of preparing ferric oxide. Although, as 

 has been already stated, oxides formed from nitrates occlude nitrogen 

 and oxygen, yet these gases are evolved upon solution of the oxides and 



* At the temperature to which the porcelain tubes were heated pure silver did 

 not melt, while sodic chloride was easily fused. 



t Landolt und Bornstein Tabellen, 118, 133 (1894). 

 vol. xxxv. — 17 



