Constitution of the Natural Silicates. 455 



pite ; and here again we have trihydration, with two molecules 

 of water loosely held and the third more firmly combined. In 

 this case we may compare the analysis, reduced to alumino- 

 magnesian form and 100 per cent, both with the formula given 

 above and with that of the hydrous phlogopite taken sepa- 

 rately. 



This comparison, which is in the main satisfactory, makes it 

 perfectly clear that kerrite is essentially a trihydrated phlogo- 

 pite, with the alkalies replaced by hydrogen. The analogies 

 between kerrite and jefferisite are perfectly clear, and both 

 minerals become monohydrated by exposure over sulphuric 

 acid. It is our intention to examine several other vermiculites 

 in the near future, and we believe that all of them will be easy 

 to interpret with the aid of the evidence already gained. 



Final considerations. — In the foregoing pages we have 

 shown conclusively that gaseous and aqueous hydrochloric 

 acid differ widely in their action upon magnesian silicates. 

 We have also endeavored to show that in this group of 

 minerals, the gaseous acid attacks only that part of the mag- 

 nesium which is present as the univalent group — Mg — OH ; 

 and although the proof is far from complete, the evidence in 

 favor of our view appears to be cumulative. In the first place 

 olivine, which cannot contain hydroxyl, is almost unattacked 

 by the gas in the range of temperatures studied. Secondly, 

 serpentine, which must contain MgOH, is attacked propor- 

 tionally to the excess of oxygen over the ortho-silicate ratio. 

 The results, to be sure, are only approximations to quantitative 

 accuracy, but they are uniform enough to warrant our conclu- 

 sion. Finally, ripidolite behaves like serpentine, and gives an 

 analogous formula ; while the micas, which presumably contain 

 little or no hydroxylated magnesia, are but slightly affected. 

 All the evidence, so far, converges to the one conclusion ; 

 which has at least the status of a legitimate working hypo- 

 thesis. All the hydrous silicates so far examined are not alike 

 in their behavior towards the reagent, but only those are 

 attacked by it in which there are strong reasons for assuming 

 the presence of the basic MgOH. 



In the course of the investigation certain collateral questions 

 have arisen which have been the subject of experiment. For 

 instance, in the action of gaseous hydrochloric acid upon sili- 

 cates may not insoluble oxychlorides be formed, which would 

 escape notice in the analysis of the soluble portion ? To 

 answer this question the residues were in several cases ex- 

 amined, and found to be practically free from chlorine. 

 Traces only of chlorine were retained, insufficient to modify 

 our ratios. Furthermore, precipitated and ignited magnesium 

 oxide, heated at 498°-52 / T in gaseous hydrochloric acid was 



