Constitution of the Natural Silicates. 309 



In the last determination the soluble magnesia was accident- 

 ally lost; and the value given was estimated by difference. A 

 duplicate determination on the Newburyport mineral, at the 

 same temperature, with 39 hours' heating gave only 14'43 per 

 cent of magnesia removed as chloride. The different times re- 

 quired for obtaining constant weight and so fixing the limit of 

 the reaction vary considerably. 



Similar experiments upon serpentines A and B at the tem- 

 perature 498°-527° gave similar results. 



A B 



Hours heated 18 18 



MgO removed 10-83 14-28 



R„0 3 removed -10 -16 



In duplicate determinations at the same temperature, A 

 yielded 14*17 and B gave 1736 per cent of magnesia converted 

 into chloride. 



Now these data, although not satisfactorily concordant, have 

 nevertheless some significance. They show first that the action 

 of the gas is much the same at both of the temperature inter- 

 vals, except that the limit of change is reached more quickly 

 in the hotter series. They show, also, that the magnesia of 

 serpentine is probably combined in two ways ; one part being 

 affected, the other unattached by the acid. That part which 

 is converted into chloride, and so rendered soluble, we may re- 

 gard provisionally as represented by the group — Mg — OH ; 

 even though our estimate of its amount may not be so sharply 

 accurate as might be desirable. JSTo other hypothesis as to the 

 nature of the dissolved magnesia seems to be practicable, or to 

 account in any way for the results of the experiments. Roughly 

 speaking, a maximum amount of about one-third of the mag- 

 nesia in serpentine is extracted by dry HC1 under the condi- 

 tions of our experiments, the other two-thirds being more 

 stable. 



By aqueous hydrochloric acid all of the five serpentines were 

 easily and completely decomposed. In three instances this fact 

 was determined quantitatively ; by evaporating the acid to dry- 

 ness with the mineral, extracting with weak acid, and weigh- 

 ing the residue. In each case this residue agreed in percen- 

 tage with the silica found by actual analysis. The data are as 

 follows : 



A c D 



Insoluble in HC1 42'32 42-21 42-55 



Silica found 42-05 42-42 42-94 



By very weak hydrochloric acid, however, these three ser- 

 pentines are but partially decomposed ; the picrolite being the 



