568 CROSS, IDDINGS, FIRS SON, WASHINGTON 



rocks derive their most obvious characters from the mineral par- 

 ticles composing them. The varying proportions of unlike 

 minerals in rocks are most striking, and other notable features 

 are due to the physical properties of the minerals, their color, 

 cleavage, hardness, etc., or to the relative or absolute size or 

 shape of the particles. It is by reference to these characters 

 that rocks may be most readily described and identified, and it 

 is then desirable that the systematic classification should be 

 constructed as far as may be by the use of mineralogical data in 

 one form or another. 



There are, however, reasons why mineral constitution by 

 itself cannot be used in the principal divisions of a compre- 

 hensive and logical classification of all igneous rocks. The 

 existence of vitreous rocks forms one of these reasons, because 

 such rocks cannot be classified at all by purely mineralogical 

 criteria. The fact that -a given magma may crystallize into dif- 

 ferent mineral combinations is another reason. Moreover, if 

 mineral composition were a simple function of chemical compo- 

 sition, and if all rocks were holocrystalline, the number of 

 chemically different minerals of importance would make the 

 task of classifying rocks by means of mineral composition alone 

 practically impossible. It, therefore, appears that neither chem- 

 ical composition nor mineral constitution can be independently 

 applied to the construction of a logical and practical classifi- 

 cation of igneous rocks. 



The primary minerals in a holocrystalline igneous rock, when 

 considered chemically and quantitatively, are a full expression of 

 the chemical composition of the magma, and their exact determi- 

 nation furnishes the chemical composition of the rock. To a 

 large extent the mineral composition may be employed as a 

 means of determining the chemical composition, and since the 

 minerals are readily determinable optically in many cases and 

 are a convenient means of identifying rocks, it is advisable to 

 treat the chemical composition of rocks in terms of ?ni?ierals, and to 

 make the basis of primary subdivisions chemico-mineralogical. 



While this conclusion is, in its general terms, quite com- 



