Lczvi?ison-Lcssi/ig s Classification of Rocks. 357 
more or less decomposed. Is it certain that other analyses are 
not liable to the same defect? And again, what are the chemi- 
cal types named olivine-basalt, quartz basalt, boralonyte. camp- 
tonyte, lamprophyre, missouryte, ijolyte, nordmarkyte, etc., 
of which the formulas are established on the basis of a single 
analysis? It is possible, and even probable, that there exist 
definite chemical types of eruptive rocks, and that they are not 
mixtures in accidental proportions, but there are no positive 
proofs of this, for there have not been made a sufficient num- 
ber of good analyses of rocks frgm different regions, i. e., of 
rocks that are absolutely fresh and susceptible of comparison 
from a mineralogical standpoint. So long as this is not done 
chemical classifications will be built upon the sand, and will 
not have any actual value, being simply premature tentatives. 
This is as much applicable to that of Lewinson-Lessing as to . 
that of all others. It seems to be necessary to insist on this, 
on account of the exorbitant pretensions of chemical classifi- 
cations. MM. Michel-Levy and Fouciue. and Lacroix (En- 
claves des Roches) have several times expressed the idea, in 
connection with their mineralogical classification of rocks, that 
this classification, based upon facts, (structure, mineralogical 
composition) has to do with the establishment and grouping 
of the types of rocks. These types, established by means of 
the microscope, the only instrumentas yet giving exact results, 
can be grouped in a different manner, according to theories of 
age, of genesis, of chemical composition, etc. The mineralog- 
ical classification therefore makes very modest pretensions. 
On the contrary Lewinson-Lessing (and according to his his- 
torical sketch all others who have advanced chemical classifi- 
cations) affirms that chemical composition only can indicate the 
intimate and actual relations between different types of rocks, 
that the chemical composition is the independent variable, and 
consequently on it is to be based all classification. The author 
well states that a rational classification ought to take account 
of every consideration, but in practice he forgets that, and 
takes account only of chemical composition. Therefore, while 
for inorganic bodies chemical composition is the principal 
property for classification, it is not the only one, and it is neces- 
sary to take into account physical properties. If rocks are not 
arbitrary (oraccidental) mixtures, but represent certain fixed 
