SYSTEMATIC PETROGRAPHY 363 
the oxygen quotient (originally proposed by Bischof) by dividing 
the oxygen of the bases by that of silica. 
Comparing all available analyses in this way Roth recognized 
that no simple chemical relations existed between different 
rocks, and that chemical formule were therefore useless; that 
rocks of different mineral composition fell in the same chemical 
division; that chemical and-mineral arrangements of rocks could 
not coincide; that a purely chemical arrangement would sepa- 
rate things closely related on geological and mineralogical 
grounds; and that a mineralogical arrangement must be con- 
nected with structure and texture. In this dilemma Roth chose 
to arrange igneous rocks primarily by their kinds of feldspar 
and the presence or absence of quartz. He admits that pyrox- 
ene or amphibole might be used in the same way, but considers 
feldspar much preferable because of its greater abundance and 
easier determination. The scheme of classification presented by 
Roth is as follows: 
I. Orthoclase rocks (often contain- 3. Amphibole andesite. 
ing some oligoclase). B. With augite. 
A. With quartz (z. é., more sili- I. Oligoclase-augite- 
ceous than orthoclase). porphyry. 
1. Granite. 2. Melaphyre and spilite. 
2. Gneiss. 3. Pyroxene-andesite. 
3. Felsite porphyry. 4. Nephelinite. 
4. Liparite. 5. Haiiynophyre. 
Re OYenite. III. Labradorite rocks. 
B. Without quartz. 1. Labradorite-porphyry. 
1. Orthoclase porphyry. 2. Gabbro. 
2. Sanidine trachyte. 3. Hypersthenite. 
3. Sanidine-oligoclase-tra- 4. Diabase. 
chyte. 5. Dolerite: 
4. Phonolite. 6. Normal pyroxenic rock (Bun- 
5. Leucitophyre. sen.) 
II. Oligoclase rocks. 7. Basalt. 
A. With hornblende. IV. Anorthite rocks. 
ie Diorites A. With augite— eukrite. 
2. Porphyrite. B. With hornblende. 
This arrangement attempts no expression of a relation 
between chemical and mineral composition; it assumes constant 
