134 THE ATOMIC WEIGHTS. 



by Nilson and Pettersson, but witb various added precautions and greater 

 elaboration of detail. Their glucina was derived from three sources^ 

 namely, leucophane, beryl, and gadolinite, and the sulphate was repeat- 

 edly recrystallized. The results are subjoined : 



The first two determinations, which give the highest percentage, were 

 made upon sulphate thrice crystallized. The others were made upon a 

 salt four times crystallized, except in one instance, when there were five 

 crystallizations. To the data derived from the four times cr3'-stallized 

 compound Kriiss and Moraht give preference, and so find a slightly 

 lower value for the atomic weight of glucinum. Combining, we have for 

 the mean percentage : 



By Nilson and Pettersson 14. 169, ± .0023 



By Kriiss and Moraht 14. 144, i .0017 



General mean 14. '53, ± .0014 



Taking now all the data for glucinum, we have — 



(I.) GIO : BaSO^ : : 100 : 922.164, ± .985 



(2.) 4CO2 : GIO : : 79.423, zh .0052 : 11.433, ± -oSi 



(3.) Percentage of GIO in G1SO.J.4H20, 14.153, ± 0014 



The antecedent atomic weights are — 



O = 15.S79, =b .0003 C --= 11.920,^.0004 



S ^ 31.828, zb .0015 Ba ;= 136.392, ± .0086 



