THE NORMATIVE AVERAGE DENSITY 389 



normative minerals and some corrections have been here taken into ac- 

 count which were not considered before, and the altitudes because Wag- 

 ner's estimates differ slightly from those of Murray formerly used. 



In my former paper the oceanic depths which are represented by the 

 rocks of the Pacific and Atlantic islands were considered to be the average 

 ocean depth. Study of the bathymetrical charts, however, shows that the 

 Atlantic islands from which we have rock analyses rise from either the 

 long central ridge or from what are deeper parts of the continental shelf, 

 while the Pacific islands may be considered to -represent more truly the 

 average depth. In the later calculations I have assumed as a basis both 

 the "ridge" depth of the Atlantic (estimated at about 1,000 fathoms) 

 and also its average depth, while the average depth of the Pacific as here 

 given (4,520 meters) is accepted. We shall see later that the more satis- 

 factory results are obtained by using the "ridge" depth of the Atlantic as 

 the basis for calculation. 



The averages given in Table II call for some critical comments, so that 

 the reader may the better understand them. Some of them are averages of 

 many analyses, which represent rocks and localities well distributed over 

 the areas and with probably a not undue proportion of analyses of "in- 

 teresting" rocks. The averages for North America, Europe, and probably 

 Australia would belong among these. Other areas are represented by few 

 analyses and these not well distributed. South America, for instance, 

 has but 138 analyses and these almost wholly from the Andes, British 

 Guiana, and eastern Brazil. Africa is represented by but 223 analyses 

 and almost all of these from localities near the coasts. The same is true 

 of Antarctica, of which we know the rocks only at isolated localities along 

 a narrow coastal fringe, while the petrology, as well as the elevation, of 

 the interior are quite unknown and probably always will be. Asia, by far 

 the largest continent of all, is very meagerly represented, the 114 analyses 

 being of rocks from few and rather "bunched'' localities, while the rocks 

 of vast areas, such as the Pamir, Tibet, and much of Siberia, China, and 

 India, are almost unknown. The Atlantic is represented by rocks from 

 but a few islands in the southern part, the basalts of the northern portion 

 not being represented among the data used, although some are now avail- 

 able. While the analyzed rocks of the Pacific are rather well scattered 

 over the area, yet the analyses are very few and many of them are of 

 "interesting" and somewhat unusual rock types, as Iddings pointed out 

 in his last paper. 



This matter of the representatives of the various averages, those of the 

 smaller areas as well as those of the larger, is discussed at some length in 

 the professional paper already mentioned. The conclusion arrived at is 



