ALUMINA. 305 



leaves 1-121 to be added to the 51-466 given by 

 theory. Now, 52-587 : 19*3 : : 6 : 2-202 = * io ' 

 mic weight of alumina. 



The 12 preceding minerals constitute a suffi- 

 cient number to enable us to approach pretty 

 nearly to the atomic weight of alumina. That 

 weight comes out somewhat differently from each 

 mineral. This may be partly occasioned by 

 small errors in the analyses; but it is owing 

 chiefly to impurities existing in the specimens 

 analyzed, the presence of which we have no 

 means of determining. If a mineral contains 

 mixed with it some silica, or a portion of any 

 stone containing more silica than itself, the ato- 

 mic weight of the alumina will come out too 

 high. The contrary will be the case, when the 

 mineral contains an excess of alumina, or of 

 a stone containing more alumina than the speci- 

 men now under examination. An accurate 

 knowledge of the rock from which the specimen 

 was taken, and of the minerals accompanying 

 it, might throw some light upon the foreign 

 matter most likely to be present. But this has not 

 yet been sufficiently attended to ; meanwhile, 

 it is probable, that the errors in the atomic 

 weight of alumina, resulting from these foreign 

 matters, will be on different sides in different 

 analyses ; consequently, the mean atomic weight 

 from all the twelve analyses given above will 

 come near the truth. Let us therefore see what 



VOL. I. U 



