MAGNESIUM. 107 



(l.) MgSOt : BaSO^ : : lOO : 193.700, ±: .0331 

 (2.) MgO : MgSO^ : : 100 : 296.806, dz .0475 

 (3.) Per cent, of MgO in oxalate, 27.3665, ±: .0023 

 (4.) Per cent, of MgO in carbonate, 47.627, ± .0018 

 (5.) Ag : MgClj : : 100 : 44.261, ± .020 



From these we find three values for the molecular weight 

 of MgO : 



From (2) MgO = 40.587, rb .0126 



From (3) " r= 40.603, ± .0069 



From (4) " =39.922,^.0030 



Genera! mean " = 40.054, ± .0027 



We have also three values for the atomic weight of mag- 

 nesium : 



From molecular weight of MgO Mg = 24.091, i .0044 



From ratio (i,) corrected " = 24.283, zfc .033 



From ratio (5,) Dumas " = 24.576, ±z .032 



General mean " = 24.103, it .0043 



Or, if == 16, Mg becomes = 24.159. 



In this general mean all the determinations are included, 

 good or bad. Dumas' result is unquestionably wrong ; the 

 error, probably, being due to the presence of oxychloride in 

 the MgCl, which was used. It is doubtful whether any 

 precautions could have eliminated that error. If we take 

 only Marchand and Scheerer's work on magnesium carbon- 

 ate as having positive value, we shall get from their analy- 

 ses the following result, viz : Mg = 23.959, ± .0040. Or, 

 if = 16, this becomes 24.014. The atomic weight of mag- 

 nesium, therefore, varies from the whole number 24, only 

 within the ordinary limits of experimental error. 



