282 ANNUAL REPORT SMITHSONIAN INSTITUTION, 1920. 



sideration of this topic would lead us too far astray but it will be 

 found discussed elsewhere. 13 



It may also be mentioned here that, as some of the minor con- 

 stituents are, in the course of analysis, precipitated and weighed with 

 others, and are later determined separately and subtracted from the 

 previous total figure, if these are not determined the figure for the 

 main constituent will be too high. This is notably the case with 

 alumina, with which are precipitated and weighed the oxides of 

 titanium, phosphorus, rare earth metals, zirconium, chromium, and 

 vanadium, with often some manganese. If the analysis is not com- 

 plete as regards these constituents, therefore, the figure for alumina 

 will be too high. 



As has been said above, the average composition of the igneous 

 rocks has been estimated by several petrologists — Clarke, Harker, 

 Loewinson-Lessing, Daly, Knopf, Mead, and myself. Clarke based 

 his earlier estimates very largely on the analyses of rocks from the 

 United States, as did Knopf, while Harker's average was of rocks 

 from Great Britain alone. In his latest estimates Clarke included 

 rocks from all over the globe, as did I in my own computation. This 

 also was the basis of Daly's and Mead's computations, though in both 

 their estimates, which were founded largely on personal selection of 

 what constituted " types " of various rocks, the personal equation 

 enters somewhat unduly. As we shall see later, continental averages, 

 or others selected from regional data, differ too much to be repre- 

 sentative of the average composition of the whole " crust." 



The basis for the present, and latest, estimate was the collection 

 of rock analyses that has recently been published. 14 This includes 

 practically all the analyses of igneous rocks, from all over the earth, 

 that have been published between 1883 and 1913, inclusive. These 

 amount to 8,602 analyses, of which 5,159 of fresh rocks were con- 

 sidered to be "superior" — that is, satisfactory as to accuracy and 

 completeness. Only these 5,159 analyses were used. The computa- 

 tions of the various averages, for the whole earth, the continents, and 

 various districts of the earth's surface, were made by Dr. F. W. 

 Clarke during the summer of 1919. To him I am greatly indebted 

 for his very painstaking and laborious undertaking, and would ex- 

 press my great appreciation of his kindness in permitting the present 

 publication of some of his results. It must be said that there are 

 presented here only a few of these, and that all the data in detail, 

 with certain considerations of them, are to be published by us jointly 

 in the near future, as a Professional Paper of the United States 

 Geological Survey. 



18 Washington, H. S., U. S. Geol. Survey, Prof. Taper No. 99, pp. 10-26, 1917. 

 « Washington, H. S. ( U. S. Geol. Survey, Prof. Paper No. 99, 1917. 



