236 PROCEEDINGS OF THE AMERICAN ACADEMY. 



do not form an infinite series, but that the varieties cluster about 

 "center-points." Osann's great compilation proves that Rosenbusch's 

 classification is an objective and "natural" one to a highly useful 

 degree. 



2. The obvious error involved in computing "the average composi- 

 tion of the primitive crust of the earth," or "the average igneous 

 rock," or "the mean composition of the accessible parts of the earth's 

 crust," by averaging a large number of analyses compiled at random, has 

 not deterred a goodly number of authors from using such results as 

 those deduced by Clarke, Washington, and Harker. These averages 

 are bound to breed further errors when used as a basis for quantitative 

 studies in geology or oceanography. The discovery of " the average 

 igneous rock " is of the highest importance for many problems such as 

 the chemical denudation of the lands and the chemical evolution of 

 the ocean. The mean composition of the accessible crystalline rocks 

 of the globe must ultimately be obtained by taking account of the 

 relative volumes of the different rock-types. In computing the mean 

 the average analyses for the principal individual species must be em- 

 ployed. Since the only approach to success is through the quantita- 

 tive study of geological maps and memoirs, it is clear that for many 

 years to come the averages for the types recognized in Rosenbusch's 

 system are to be basal to the calculation. 



A glance at Table II. shows, however, that this new world-average 

 will differ little from the earlier world-averages with respect to one 

 oxide, namely, soda. For each of the areally and volumetrically im- 

 portant rock-types the average soda never departs far from a mean 

 of about three and one half per cent. The soda in the averages of 

 Clarke, Washington, and Harker (calculated as water-free) is, respec- 

 tively, 3.63 per cent, 3.34 per cent, and 3.90 per cent. 1 The agree- 

 ment is fortunate, since, for example, the quantitative problem relative 

 to the sodium in the ocean can be pursued without waiting for the 

 close determination of "the average igneous rock." Incidentally, it 

 may be remarked that the estimates of Joly 2 and Sollas 3 regarding 

 the age of the ocean, as determined by the sodium content, need revis- 

 ion, since neither author has allowed for the great variations in the 

 area of the lands during geological time. 



3. The recurrence of the main types of igneous rock in every conti- 

 nent shows that general processes of differentiation have been at work 



1 F. W. Clarke, Bull. 228, U. S. Geol. Survey, 1904, p. 16. 



2 J. Joly, Sci. Trans. Roy. Dublin Society, 7, 23 (1899). 



3 W. J. Sollas, Quart. Jour. Geol. Soc, Presidential Address, 65, p. lxxix 

 (1909). 



