THE STATISTICAL METHOD IN CHEMICAL GEOLOGY. 



By F. W. CLARKE. 

 (Read April i8, 1906.) 



In an essay upon the relative abundance of the chemical elements, 

 published some sixteen years ago/ I attempted to apply the statis- 

 tical method to ascertaining the average composition of the earth's 

 crust. Since that time the data have been repeatedly revised, both 

 by myself and by others, and the results obtained have found appli- 

 cations which I did not anticipate. The composition of the litho- 

 sphere has furnished, so to speak, a sort of base-line to which other 

 computations could be referred ; the figures, therefore, have acquired 

 a peculiar importance, and it has become desirable to determine, 

 more critically than heretofore, the degree of their validity. To dis- 

 cuss the nature of the averages and to consider how far they may be 

 utilized is the purpose of the present communication. In order to 

 accomplish this purpose I must restate, very briefly, the main points 

 of the original argument. 



As a first step in the discussion, it was necessary to assume a 

 -definite mass of matter as available for statistical analysis. That 

 mass included the atmosphere, the ocean and a certain portion of the 

 lithosphere; the portion, namely, that may be supposed to lie within 

 our reach. For the last item it was assumed that a shell having a 

 thickness of ten miles below sea-level would represent known ma- 

 terial ; in other words, that it would consist of rocks essentially iden- 

 tical in general character with those which we can study at the sur- 

 face. How much thicker the rocky crust of the earth may be is 

 another question, upon which we do not need to enter. It is only 

 the known material which concerns us now ; and some of that is 

 brought to us by volcanoes from depths far below any to which we 

 can penetrate directly. The eruptive rocks enable us to determine 

 what sort of matter lies below the immediately observable surface. 



'^ Bull. Philos. Soc. Washington, vol. 11, p. 131, Oct. 26, 1889. Also in U. S. 

 •Geol. Survey Bull. 78, p. 34. 



14 



