CLASSIFICA TION OF IGNEOUS ROCKS 85 



good of all concerned, that the systematic classification of rocks, 

 according to which their specific names are applied, must be 

 based on their properties as objects, together with only such 

 geological criteria as may be found adaptable, to the end that 

 the system may be uniform, stable, and as natural as possible. 



At this point I wish to digress for a moment and compare 

 the task of the petrographer with that of the zoologist, the sys- 

 tematic botanist or the mineralogist. From the beginnings of 

 natural history, all natural objects have been subject to classifi- 

 cation, at first on the most evident properties, and subsequently 

 according to relationships. The modern zoologist finds it pos- 

 sible to adopt nearly all of the general groups of animals early 

 set up by the naturalist. Fishes, reptiles, birds, and other 

 groups, needed only to be defined in scientific terms to bring 

 general and scientific usage into harmony. The botanist has 

 not been able to make his system correspond so closely to that 

 of the naturalist. He has found that many natural groups of 

 plants cannot be brought into his system, and he has wisely 

 refrained from redefining the old names for those groups in such 

 a way as to destroy their old and legitimate meaning. Thus 

 trees (silvae), shrubs, bushes, vines, evergreens, deciduous 

 plants, and others, are not divisions of systematic botany, 

 though recognized as useful and natural groups in the broader 

 science of the vegetable kingdom. The properties and relation- 

 ships of minerals may be nearly expressed in one system, but, 

 as has been shown, rocks are of such manifold relationships that 

 they defy a single system of classification to a much greater 

 degree than plants. 



If we now examine the schemes for the classification of rocks 

 which have been current in the past few decades it appears that 

 geological criteria have frequently been applied to produce the 

 first divisions. It has been plain to all that rocks may be divided 

 primarily into a few great classes on grounds either of geological 

 occurrence or relationship, or of material properties. Each classi- 

 fication has its own justification, but the criteria to be applied in 

 constructing asystematic scheme should plainly be those caus- 



