108 
PETROLOGY: CLARKE AND WASHINGTON Proc. N. A. S. 
fair size, but the formulae should not be used for very small absolute 
numbers of births and deaths. 
^ Papers from the Department of Biometry and Vital vStatistics, School of Hygiene 
and Public Health, Johns Hopkins University, No. 52. 
2 Pearson, K. On a form of spurious correlation which may arise when indices are 
used in the measurement of organs. Proc. Roy. Sac, 60, 1897 (489). 
THE AVERAGE CHEMICAL COMPOSITION OF IGNEOUS ROCKS 
By Frank W. Clarke) and H^nry S. Washington 
U. S. Geological Survey* and Geophysical Laboratory of the Carnegie 
Institution of Washington 
Communicated March 22, 1922 
We have recently completed computations of the average chemical 
composition of the igneous rocks of various countries, of the continents, 
and of the earth's crust. These are to be published as a Professional 
Paper of the U. S. Geological Survey, with discussions of the characters 
of the various averages, of the relative abundance, distribution, associa- 
tion, and evolution of the elements, and other matters. Inasmuch as it 
is probable that this paper will not appear for a considerable time, it is 
thought best to publish now the various average compositions arrived at, 
in order to put before geologists and others some interesting data as to the 
chemical composition of the earth. 
The data used by us were 5,159 chemical analyses of igneous rocks from 
all over the earth, and published between 1884 and 1913, inclusive. Analy- 
ses only of "superior" quality, and of fresh unaltered rocks, were con- 
sidered. These were taken from a collection of rock analyses made by 
one of us.^ In computing the averages of the various countries and the 
continents the sum total of each constituent was divided by the total 
number of analyses of rocks from the area dealt with. In the case of 
the earth average, both in terms of oxides and of elements, the same 
method was adopted for the major constituents, Si02, AI2O3, Fe203, FeO, 
MgO, CaO, Na-O, K2O and H2O (only water above 110° was considered), 
while the average for the minor constituents was obtained by taking the 
mean of the sum total of each divided by the number of analyses (which 
gives too low a result) and by the number of determinations (which is 
apt to give too high a result). 
In computing the average of the whole crust the relative proportions 
estimated by one of us^ some years ago were adopted. The latest esti- 
mates of the masses and compositions of the hydrosphere and atmosphere 
have been used. These relative masses are as follows: Lithosphere 93%; 
hydrosphere 7%; atmosphere 0.03%. 
