840 Ervest W. Cheats: 



The volcanic hypothesis was once commonly held until the ex- 

 amination of recent volcanic rocks in Austialasia and the absence 

 of similar forms from any known active volcanoes made the 

 explanation of their composition and distribution difficult. 



Dr. Summers has clearly pointed out that though these bodies 

 have been called obsidianites, the material is not obsidian, and 

 differs from it in certain notable chemical characters, especially 

 the low alkali percentage, and the relatively higli percentage of the 

 alkaline earths, the latter being normally in excess of the former^ 

 whereas the reverse relation is true of obsidian. 



Comparison of the Taradale so-called obsidian with 

 the Uralla austr-allte. 



Mr. Dunn has made a comparison of the chemical composition 

 of the Taradale rock with that of the Uralla australite, and comes 

 to the conclusion that the two rocks are practically identical and 

 both are obsidian. I have shown that the Taradale rock is not 

 obsidian, neither for that matter is the Uralla australite. Neither 

 are the tw^o rocks identical. It is true that the percentages of silica 

 alumina and alkaline earths are fairly comparable, but the percent- 

 ages of soda and of oxide of iron, especially ferric oxide, are widely 

 different. 



Two ways exist of comparing analyses, each of which is more 

 illuminating than a simple comparison of oxides. The one method 

 is by a comparison of the norms determined on the American 

 classification. These are shown above (p. 338), and show wide 

 differences. 



The other method is by the drawing and examination of " varia- 

 tion curves,'' as utilised by Dr. Summers in his recent paper. l By 

 this method variations of composition within the limits of a rock 

 species can be shown to lie along a curve, and not only so, but 

 where a number of different rocks in a district have a genetic 

 relationship with one another this is brought out clearly b}^ the fact 

 that all the related types conform to the curves. On the other 

 hand if two or more rocks are not so related this difference is 

 indicated by a lack of conformity between the luives representing 

 the different rocks. 



I have made such a comparison of " variation curves " between 

 the curves of the Uralla australite and that of the Taradale roc'; 



1. Op. cit. 



