The Natural Classification of Igneous Rocks. 175 



that differentiation and assimilation are in a measure antithetical 

 processes. 



If the deep-seated magmas of large volume have acquired their 

 various chemical characters in different ways, it appears at once 

 evident that this primary genetic factor cannot be used in classifi- 

 cation, unless the characters of different origin can be distinguished 

 in the rocks. 



Classification by geographical distribution of chemically different 

 rocks is considered, and the groupings proposed by various writers 

 are discussed ; and it is shown that the rocks of the Pacific zone of 

 North America indicate that they possess provincial peculiarities of 

 interest, but that these are not by any means identical with the 

 features emphasized by Becke and others as characterizing the 

 Pacific kindred. 



The factors of magmatic differentiation are then reviewed. 

 The aschistic and diaschistic magmas of Brogger and the ' dyke 

 rocks ' of Rosenbusch are discussed ; and it is contended that 

 certain dyke rocks of Colorado show a notable exception to the 

 rule postulated by Rosenbusch. The conclusion is reached that 

 the sharp distinction between the two ' dyke rock ' groups is a 

 purely arbitrary one, resting on an unproved hypothesis. 



A discussion on the classification by eutectics follows, and the 

 writings of Gr. P. Becker and J. H. L. Vogt on this subject are criti- 

 cized. The view that graphic, spherulitic, and felsitic textures are 

 characteristically eutectic is considered to be incorrect, and it is 

 contended that magmatic classification by eutectics is fundamentally 

 weak— because it rests on hypothesis, because it does not apply to 

 all rocks, and because it does not allow for the entire magma of 

 most rocks. A classification by eutectics may, in the future, be 

 realized ; but it seems inevitable that it must be a classification for a 

 special interest, not for the general science of petrography. 



The author considers that the distinction between felspathic and 

 non-felspathic rocks which has been so prominent in current 

 systems is not only unnatural, but is in the highest degree 

 arbitrary. 



The use of texture is then discussed, and it is shown that classi- 

 fication by occurrence, as determining texture, or by texture, as 

 expressing the broad phases of occurrence, is based on long disproved 

 generalizations made from limited observation. The ' American 

 Quantitative System of Classification ' is then briefly dealt with, and 

 the following general conclusions are formulated: — 



' The scientific logical classification of igneous rocks must apparently be 

 based on the quantitative development of fundamental characters, and the 

 divisions of the scheme must have sharp artificial boundaries, since none exist 

 in Nature. 



'Chemical composition is the fundamental character of igneous rooks, but it 

 may be advantageously expressed for classificatory purposes in terms of simple 

 compounds, which represent either rock-making minerals or molecules entering 

 into isomorpbous mixtures in known minerals. It is probable that the 



