236 THE AMERICAN NATURALIST. (Vor. XXXV. 
types. The difficulty in all classification schemes has been to select 
such types as will express the essential differences between nearly 
related rock masses and at the same time to show the relationships 
existing between them. The graphic method of representing rock 
analyses by diagrams serves to emphasize the characteristic chemical 
features of the different specimens analyzed, and to enable the student 
to recognize readily their differences and similarities. From these 
diagrams a composite diagram may be constructed, and this repre- 
sents a chemical type. The chemical relationship of various types 
is easily read from their diagrams; and classification of the dia- 
grams is a classification of rocks according to chemical composition. 
The first complete diagrammatic representation of rock composition 
was suggested by Brógger.! His diagrams are polygons drawn from 
the ends of radius vectors whose lengths correspond to the propor- 
tions of the constituents in the rock represented in the diagram. 
The shapes of the polygons show at a glance the relative importance 
of the principal components. They are extremely characteristic of 
the different rock types, and thus may be used to exhibit chemical 
relationships. 
Hobbs? makes a few unessential modifications in the Brogger 
diagrams and combines them to form composite diagrams or dia- 
grams of rock types. This he does, however, not by actually com- 
bining the individual polygons, but by averaging the analyses of the 
rocks which are assumed to belong with the type in question. So far 
as rock classification is concerned this method is not of great value, 
since it does not exhibit the differences in composition between the 
constituent members of the composite and the type produced by them. 
It affords a very convenient method of exhibiting the peculiar chemical 
characters of rock families and a concise method of showing their 
chemical relationships with one another, but the personal element 
enters so largely into the selection of the rocks that comprise the 
groups represented in the diagrams that these are not of great use 
in other respects. If the analyses represented in the diagrams could 
be averaged in such a way as would take into account the abundance 
of the various rocks, the composites would possess a high scientific 
interest. The idea proposed by the author is, however, an excellent 
one, especially for comparative purposes. It is bound to be received 
with favor. 
Harker? suggests a method of using diagrams of rock series 
! Das Gangafolgs des Laurdalits, p. 255. Kristiania, 1898. 
? Journ. of Geol., vol. viii (1900), p. 1. 3 Journ. of Geol., vol. viii (1900), p. 389. 
