112 Dr. W. F. Hillebrand on the Specific Heats 



The atomic weights of the cerite metals hitherto adopted have 

 in no case been directly determined, but have been estimated 

 from the capacity of saturation of their salts. The estimation 

 has been based upon the hypothesis, hitherto grounded on no 

 well-established facts, that the amount of oxygen contained in 

 them is one third of that present in the sulphuric acid needed 

 to form the neutral salts. If, however, the composition of 

 these salts were analogous to that of the uranium oxide, the per- 

 centage composition which is generally and unhesitatingly 

 adopted for them would be completely erroneous. In order to 

 test the validity of the hypothesis, it appeared to be of interest 

 to fix definitely the amount of oxygen contained in the didy- 

 mium oxide. For this purpose 0*9071 grm. of the same sample 

 of didymium which had served for the determination of the spe- 

 cific heat of the metal was dissolved in nitric acid with all pre- 

 cautions to render loss by spirting impossible, and the solution 

 was evaporated in a platinum crucible and carefully ignited. 

 The weight of the oxide thus formed amounted to 1*0626 grm., 

 and was unaltered by ignition in an atmosphere of hydrogen. 

 When dissolved in hydrochloric acid and evaporated to dry- 

 ness, it yielded, on re-solution in hydrochloric acid and water, 

 0*0071 grm. of silica. 



The solution was then mixed with ammonia and the preci- 

 pitate (A) filtered off; the filtrate,when evaporated and ignited, 

 gave a residue which weighed 0*0028 grm., and proved to be 

 didymium oxide. Precipitate A, digested for some time whilst 

 moist with oxalic acid, gave an oxalate, from which 1*0640 

 grm. of didymium oxide was obtained by ignition, and a 

 liquid, from which, after evaporation and ignition, 0*0044 

 grm. of alumina was obtained. On converting the oxide of 

 didymium into chloride and testing it by the electric spark 

 before the spectroscope, it showed weak lines of lanthanum, 

 the amount of which, however, was inappreciable. 



All these operations were performed in platinum vessels, 



and with reagents which were perfectly free from all impurity. 



Tares of platinum were employed in all the weighings, in 



order to make the influence of the displaced air as small as 



possible. 



The percentage composition of the oxide obtained from the 

 metal is accordingly: — 



6J I. 



Didymium oxide . . . 98*701 



Silica 0*668 



Ferric oxide 0*414 



Alumina 0*179 



Loss 0-038 



100*000 



