RICHARDS AND PARKEU. — ATOMIC WEIGHT OP MAGNESIUM. 67 



determiuatious were consecutive, except that one determination met with 

 an accident and was not completed. 



SERIES I. 



A careful consideration of the possible constant errors involved in the 

 foreirointr results leads to the belief that the fiorures found are too high 

 rather than too low, as the presence either of a small amount of water or 

 of oxjchloride in the magnesic chloride would tend in this direction. 



Second Series of Determinations. 



In order to drive all the subliming ammonic chloride to the farther 

 end of the combustfbn tube during the ignition, it had been found neces- 

 sary that the current of gas should be very considerable ; and hence it 

 was desirable to construct a piece of apparatus which should deliver the 

 various gases rapidly, but nevertheless as dry as it is possible to obtain 

 them. It was also desirable to work with larger quantities of materials 

 than could be handled in the former apparatus. For these reasons an- 

 other piece of apparatus was constructed to dry the hydrochloric acid 

 gas ; this apparatus contained several flasks of sulphuric acid, three very 

 efficient towers containing the same acid, which was constantly renewed, 

 and a long tube containing resublimed phosphoric pentoxidg. One of the 

 towers is shown on the following page. The whole apparatus was fused 

 or ground together, thus wholly avoiding rubber or cork connections. 



In the following determinations the boat was allowed to cool in an at- 

 mosphere of dry nitrogen, as a further precaution against a possible 

 partial decomposition of the sensitive magnesic chloride. As soon as the 



