their Determination in general. 277 



chloride of magnesium employed, we cannot repress some sur- 

 prise at the hasty conclusions which M. Dumas has drawn from 

 incomplete investigations. 



On continuing his attempts to deduce the equivalent of mag- 

 nesium from the composition of chloride of magnesium, Dumas 

 has obtained numbers extending from 12'0 to 12"6 ; that is, dif- 

 ferences of 0'6, or 5^jer cent, of the whole equivalent of magnesium. 

 Under these circumstances, there can be but little importance 

 attached to the fact that in most of the experiments the number 

 12-25 was obtained. Finally, Dumas admits that for the pre- 

 sent the number 12 can be retained. It certainly M'ill be re- 

 tained; for the determinations from which it is derived bear the 

 stamp of the greatest accuracy ; the results of the separate ex- 

 periments show no greater variations than one expressed by the 

 numbers 11-98 and 1205. 



If, now, we seek the reason of these great variations in the 

 experiments of Dumas, we can only discover it in the diversity 

 of composition of the chloride of magnesium employed. From 

 the results given, we can at once draw the conclusion that none 

 of the four substances prepared in the four different ways by 

 Dumas had exactly the composition required by the formula, 

 and probably not one had the same composition as another. 

 Dumas in this case has himself admitted that it is extremely 

 difficult to prevent the chloride of magnesium from taking up 

 oxygen, and that, when this has once taken place, even continued 

 treatment with hydrochloric acid is not sufficient to purify it 

 again. It is just this point upon which I rest my doubt of the 

 accuracy of the method employed. And there is every reason 

 to suppose that many other chlorides, especially such as are hy- 

 groscopic, behave similarly to chloride of magnesium. With 

 some, especially with chloride of zinc and chloride of bismuth, I 

 have had ample opportunities of satisfying myself how difficult it 

 is to obtain them free from oxygen. 



If chloride of bismuth which has taken up a trace of moisture 

 be heated, hydrochloric acid is given off, and the residual chlo- 

 ride contains an amount of oxid.e or oxychloride corresponding 

 to the quantity of water absorbed. Former data of Jaquelain 

 perfectly accord with this observation. 



Hence I am not surprised that Dumas, in determining the 

 equivalent of bismuth from chloride of bismuth obtained by frac- 

 tional distillation, obtained various results. It is only surprising 

 that here, too, a number, 214, was considered ripe for publication^ 

 and entitled to be employed in establishing a laiv, which shortly 

 afterward had to be retracted by Dumas himself as inaccurate, 

 because it appeared that the substance which had served for the 

 determinations was impure (coloured brown). Further deter- 



