6o8 C/WSS, IDDINGS, PIRSSON, WASHINGTON 



In both the above cases a and b it is evident that, given the 

 knowledge of the norm and the mutual interrelations between 

 the rock-making minerals which will be explained in the Part 

 on Calculation, it is only necessary to know the amount of 

 abnormative minerals, whether standard or not, to be able to 

 ascertain the changes from the norm which their presence 

 involves. These distinguishing minerals therefore may be 

 called critical ones. 



The question of how far quantitative distinctions based on 

 the amounts of the critical minerals are to be introduced into the 

 classification is discussed in connection with the nomenclature. 



Having thus explained the various relations possible between 

 norm and mode, it will be well to summarize the correspondence 

 observed in fact among the different Classes. It is obvious that 

 no one systematic method of calculation of mineral composition 

 from the chemical can correspond with all the varying possibili- 

 ties of actual mineral development. The method adopted by us, 

 however, being founded on the most frequently observed min- 

 eral relations in igneous rock and on the most generally appli- 

 cable principles, should yield norms which correspond with the 

 mode in the majority of cases. 



This is a point which can only be tested by appeal to the 

 literature of petrography, and comparison of the calculated 

 norms with the modes as furnished by the descriptions. We 

 have, it may be stated here, tested our proposed method in 

 regard to this point very thoroughly upon many hundreds of 

 rock analyses, and especially by means of a collection made by 

 one of us (H. S. W.) of all analyses of igneous rocks which 

 have been published since 1883, amounting to over three 

 thousand. 1 



This comparison shows that the accord between norm and 



mode is complete or nearly so in the very great majority of 



'It is hoped that this collection, which will show the norm in every reliable case, 

 will appear shortly after the publication of the present paper, and it will serve as a 

 general illustration of the system. Although not as yet available to others, it has been 

 appealed to as a practical check throughout our discussions, so that the system cannot 

 be regarded as merely academic, but as having been tested at every possible point. 



