580 CROSS, IDDINGS, PIRSSON, WASHINGTON 



subdivision here should be into thirds, but, in view of the con- 

 fusion liable to arise from the change in proportions the method 

 adopted seems the most advisable one. 



No subdivision need be made of a factor which is present in 

 negligible amount, the other factor being extreme, since the 

 whole subordinate factor falls within the allowable limit of 

 variation. 



It may be noted in this place that certain prefixes are used 

 in connection with mineralogical and chemical terms to indicate 

 that a factor is extremely abundant, or is dominant. In the first 

 case the prefix is per. In the second case it is do or dom. When 

 comparison is made on a threefold basis, and one factor pre- 



domi?iates over another! > -I, the prefix is pre. 



Since this classification is largely a chemical one, and since 

 in all the calculations of minerals the molecular amounts of each 

 constituent only are used, it must be borne in mind that all 

 chemical comparison is made on the basis of the relative num- 

 ber of molecules, that is, it is purely molecular. For this purpose 

 all percentages in analyses must be reduced to molecular 

 ratios, by dividing each percentage by the proper molecular 

 weight. 



OUTLINE OF THE SYSTEM. 



The subdivisions of igneous rocks proposed by us, based upon 

 the principles discussed in the preceding pages are as follows : 



Class, Subclass. 

 Order, Suborder. 

 Rang, Subrang. 

 Grad, Subgrad. 



The word Rang, which is an obsolete form, equivalent to 

 Rank, has been chosen instead of Rank to avoid confusion, 

 since it is desirable that the technical term should differ from 

 one which is in common use for other purposes. The same is 

 true of Grad which is an old form of Grade. 



The four terms — Class, Order, Rank and Grade — were at first 

 selected because they are of the same category, the first two 



