ROCK CLASSIFICATION ON THREE CO-ORDINATES 209 
and modified by its author concurrently with the development of 
petrography during the past forty years, has come to have a very 
wide usage, either in its author’s own form, or as modified by 
geologists in some respects to suit local needs. One of its elements 
of strength is the fact that it is based upon the mineral composition, 
and only secondarily defined in terms of chemical composition. It 
is also based upon the geological conditions of formation, so that the 
rocks of batholiths are separated from those of surface flows and 
both are separated from those of dikes and sills. Rosenbusch has 
never put this classification into diagrammatic form, and this fact 
has aided him in adapting it to advancing knowledge and enabled 
him to employ in some groups as minor bases of division factors 
or properties wholly inapplicable to other groups. 
It is surprising that even in the latest form Rosenbusch still 
recognizes geological age as a basis of classification In some groups. 
He also emphasizes the important distinction between alkaline and 
other rocks. 
In spite of its relative inflexibility a diagrammatic or tabular 
scheme of classification has numerous advantages, to which, for 
example, Kemp’s' greatly simplified modification of the Rosenbusch 
classification owes much of its popularity. Such tabular schemes 
have heretofore been arranged on two co-ordinates. The relation- 
ships of rocks are so numerous and complicated that one of the 
chief defects of diagrammatic devices has been that they brought 
out so few of these affinities. By the use of transparent paper 
such a classification may be constructed on three co-ordinates as 
shown on the triple insert, and the advantages of such a scheme 
obtained without the defect mentioned. A classification on three 
co-ordinates does not show all the relationships that exist, but it 
exhibits many more than the ordinary arrangement, and it is 
believed that no more can be shown without sacrificing the advan- 
tages of the tabular arrangement. 
The classification given herewith is based largely upon that of 
Rosenbusch, but it differs from the latter in various important 
respects, so that responsibility for it must lie with the author. 
The first co-ordinate is in the direction normal to the paper, and 
t Handbook of Rocks. 1900. 
