DIDYMIUM. 237 



To A I have added a single result of Erk's, to be described 

 further along. It will be seen that although A is theoreti- 

 cally defective, its figures are much more concordant than 

 those in B. In fact, the latter would almost vanish for the 

 final general mean for the atomic weight of didymium : > 



From A Di = 143.929 



" B " =150.436 



In a later paper* Marignac adopts two other methods for 

 establishing the atomic weight of didymium. The carefulh^ 

 dehydrated sulphate was taken, the didymium was precipi- 

 tated as oxalate, and the latter, ignited, yielded oxide. The 

 following percentages of oxide were found : 



58.22 

 58.24 

 58.29 



58.31 

 58.29 



Mean, 58.27, dr .0115 



The chloride of didymium was also studied. As the anhy- 

 drous salt could not be obtained in an absolutely definite 

 state, Marignac prepared neutral solutions of it and deter- 

 mined the ratio between didymium oxide and silver chlo- 

 ride. The latter compound was first precipitated in the 

 usual way, and filtered off; the excess of silver in the filtrate 

 was removed by hydrochloric acid, and after that the didy- 

 mium was thrown down as oxalate and weighed as oxide. 

 The subjoined weights of AgCl and Di^Og were found. In 

 a third column 1 give the ratio between the two compounds, 

 putting AgCl at 100 : . 



AgCl. DLfiy Ratio. 



10.058 grm. 3.946 grm. 39-232 



5.029 " 1.960 " 38.974 



5.844 " 2.276 " 38-946 



Mean, 39.051, rb .061 



Hence Di = 143.637, ± .263. 



* Ann. d. Chim. et d. Phys., (3,) 38, 148. 1853. 



