416 Prof. B. Sillimari) Jr., on Gibbsite and Allophane. 



4th Analysis. — -90S gramme of alumina obtained in analysis 

 No. 1, treated by the above process, gave 



- • • 



0-0126 Mg P = -677 per cent, of phosphoric acid. 



This quantity of phosphoric acid is equivalent to 



•753 per cent, of phosphate of alumina. 



The remainder was only alumina. 



From the foregoing results, it is evident that the Gibbsite of 

 Richmond, Massachusetts, corresponds to the formula first given, 



• » • 



Al H 3 , which requires 



Ai 65156 



H 34-440 



100000 and the mean results of these analy- 

 ses correspond very closely to the calculated per-centages. The 

 phosphoric acid is therefore like the magnesia, iron and silica con- 

 tained in Gibbsite, only a contingent impurity. 



Allophane. 



specimens 



coated by a hyaline crust, cleaving in laminated scales, sometimes 

 opake white and pulverulent, and again perfectly limpid and col- 

 orless. It is extremely brittle, easily reduced to powder, and 

 yields easily to the knife. Before the blowpipe it loses water and 

 falls to a white powder, or loosely coherent slag. In salt of 

 phosphorus leaves a siliceous skeleton. Gives off water abund- 

 antly in the close tube ; this water is neutral. With nitrate of 

 cobalt on charcoal it yields a blue slag, and it dissolves with ef- 

 fervescence in carbonate of soda. 



A qualitative examination gave evidence of the presence of 

 water, alumina, silica, and magnesia. The whole character of 

 the mineral left no doubt that it was allophane, and the chemical 

 analysis confirmed this conclusion. It was not easy to find a por- 

 tion of the mineral which was perfectly homogeneous — or to 

 separate entirely all those portions which had evidently undergone 

 more or less decomposition. 



After several analyses and repeated determinations of the quan- 

 tity of water, I arrived at the following result, which if we re- 

 gard magnesia as an accidental constituent, and distribute it and 

 the loss between the silica and alumina, will give us almost pre- 



per-centages required by the formula Si 2 A1 3 + 1^ B 



viz: 



