276 L. V. Pirsson — Classification of Igneous Rocks. 



All igneous rocks may be expressed in salic and femic 

 minerals, and according to the relative amount of each 

 group as compared with the other, they are divided into 

 five classes, persalane, nearly or entirely composed of 

 salic minerals : 



(sal: fern > 7:1) ; 

 dosalane, mostly salic 



(sal:fem< 7:1 > 5:3) ; 

 salfemane, equal or nearly equal quantities of each 



(sal: fern < 5:3 > 3:5) ; 

 dofemane, mostly femic minerals 



(sal: fern <3:5 > 1:7) ; 

 and lastly perfemane, nearly or. entirely femic 



(sal: fern < 1:7). 

 The classes thus obtained are subdivided into orders on 

 the relations of the salic minerals, quartz, feldspars, and 

 feldspathoids (generally nephelite) to one another in the 

 first three classes and on somewhat similar relations 

 among the femic minerals in the last two. More minute 

 consideration of the mineral oxides divides the orders 

 into rangs, and the rangs into grads. The proportions 

 by which they are thus divided are always the same as 

 those by which classes are made. 



Further details regarding this system will be found in the 

 work referred to. It is the most exact system that has hitherto 

 been proposed, and is based upon the fundamental property of 

 the rock — its chemical composition. Aside from this, its most 

 striking feature is the recognition of the quantitative relations 

 among the component rock-minerals, the bearing of which we- 

 shall presently see. As must be the case in all petrographic sys- 

 tems the divisional lines of classification are arbitrarily drawn, but 

 they are carried out logically on a consistent plan. It has been 

 much used in careful and exact work, but the requirement of the 

 knowledge of the chemical composition of a rock, and the difficulty 

 in manj^ cases, of obtaining this, has doubtless prevented its wider 

 extension. 



General Remarks on Classification. From what has 

 been stated in the foregoing discussions the student will 

 have doubtless perceived that the chief difficulties in the 

 systems described, excepting the last one, have been two ; 

 first, the attempt to introduce simultaneously into one 

 scheme too many of the different properties and affinities 



