EXPERIMENT IN MINERALOGY. 253 



strengthen the theory, and the same is true of the other 

 numerous groups in which a hypothetical constituent is 

 introduced. 



Even in so well known a group as the amphiboles there 

 is great doubt whether the aluminous varieties are to be 

 explained by the replacement of one atom of Si by the 

 univalent group MgAl, or whether MgSiO a is to be regarded 

 as replaceable by RA1 2 4 , or whether the metasilicate 

 RSi0 3 can form isomorphous mixtures with a silicate of the 

 garnet type R 3 Al 2 (Si0 4 ) 3 . 



Or again take the variations of the optical characters in 

 the pyroxenes — one of the very few groups which have 

 been seriously studied in this respect — it is known that 

 the extinction on a certain face, and the optic axial angle 

 both increase with the percentage of iron, but various 

 views are held on the nature of the variation ; it was at 

 first supposed that the increasing percentage of Fe 2 O s was 

 attended by a regular change in the properties, but Doelter 

 is of opinion that the increase is due to the simultaneous 

 action of no less than three silicates which (with others) 

 may enter into the composition of a pyroxene, namely, 

 CaFeSi 2 O e , MgFe 3 SiO e , and MgAl, SiO e (9). 



In all these cases the production and study of artificial 

 compounds is a great need ; the laws which operate must 

 obviously be more easily established upon pure substances 

 of simple composition than upon the impure and complex 

 compounds which occur as minerals. 



The truth is that in whichever direction the mineralo- 

 gist turns the problems by which he is confronted are pre- 

 cisely those which can only be solved by experiment ; 

 speculations exist in abundance, but they can only be 

 confirmed or rejected by the decisive test of experimental 

 research. 



EXPERIMENTS ON CRYSTALLINE STRUCTURE. 



One or two instances chosen from the crystallographic 

 side of the subject will suffice to indicate how fruitful 

 (though so rare) experiment has been in recent work. 



It was a simple experiment by which Reusch converted 



