68 GEOLOGY OF THE HIGH PLATEAUS. 



Now, let us make the following arrangement. Place at the head of 

 the series hornblendic propylite. Select from the list in the order given 

 those rocks which are more acid than propylite. Take next those which 

 are more basic than propylite, and write them also in the order in which 

 they occur. We shall then obtain the following grouping: 



1. Hornblendic jiropylite. 

 3. Hornblendic trachyte. 2. Hornblendic andesite. 

 5. Sanidin trachyte. 4. Augitic andesite. 



6. Liparite. 7. Dolerite. 



8. Ehyolite. 9. Basalt. 



This resolves the lithologic series into two semi-series, each of which 

 displays a distinct and unmistakable progression of chemical and physical 

 properties. The first includes the acid and sub-acid groups, which increase 

 in acidity with the process of the volcanic cycle. The second includes the 

 basic and sub-basic groups, which correlatively decrease in acidity. The 

 law may be thus expressed in terms of chemical properties to which the 

 phj^sical properties stand in a relation of dependence: At the commencement 

 of the volcanic cycle the rocks first erupted are those which belong to the 

 middle of the lithological scale. As the cycle advances, the rocks resolve 

 themselves into two semi-series, growing more and more divergent in char- 

 acter, and when the end of the cycle is neared they become extreme in 

 their contrast. 



Taking Richthofen's five orders (major groups) and arranging them on 

 the same plan, we may express the same correlation as follows: 



1. Propylite. 

 3. Trachyte. 2. Andesite. 



4. Ehyolite. 5. Basalt. 



Possibly it might be thought that this mode of finding a sequence and 

 a correlation bears a resemblance to some problems in the properties of 

 numbers, in which, any fortuitous collection of numbers being taken and 

 treated to certain manipulations, a law of arrangement appears ; the real 

 explanation being a latent petitio principii. But this is not so. Even if we 

 took Richthofen's five orders only, the probabilities against a merely fortu- 

 itous coincidence of orders of eruption with the above double sequence of 

 physical properties would be as 3 to 1 . But if we apply the same treat- 



