398 H. S. WASHINGTON ISOSTASY AND ROCK DENSITY 



of the points, and among these those which are the most satisfactory as 

 to chemical representativeness and average altitude, fall on or near to the 

 hyperbolas, while others lie more or less far from the lines, although 

 roughly along them. We have noted above the probable reasons for the 

 departures in the case of some of the loci, such as those for Utah-Nevada, 

 Balkania, Oregon, Madagascar, the Algonkian region, and Eussia. The 

 departures of the others from the calculated positions may naturally be 

 ascribed to errors in the estimated altitudes, to the more or less unrepre- 

 sentative character of some of the chemical averages (and. consequently 

 the average densities), or to both combined. 



Notwithstanding all small discrepancies and defects in details, the data 

 given in Table III and the graphic representation in figures 1 and 2 show 

 clearly that the densities and altitudes vary together in the sense that as 

 the altitudes decreases the density increases, and the converse, although 

 not with strict proportionality. In other words, our study of the average 

 densities and altitudes shows that the higher parts of the crust are com- 

 posed of or underlaid by relatively light material, while the lower parts 

 are underlaid by heavier material. This, as was pointed out at the begin- 

 ning of this paper, is precisely the relation demanded by the theory of 

 isostasy. 



An illustration of what appears to be a consequence of this interrela- 

 tion is to be found in the ancient shields — the nuclei of the continents. 

 It is commonly recognized that these are made up to a large extent of 

 granitic rocks and gneisses, rocks of low density. We have no actual 

 knowledge of the exact chemical composition, and consequently average 

 density, of the rocks of the Canadian shield, but that it is largely com- 

 posed of the rocks mentioned is too well known to need more than men- 

 tion. The Baltic shield is composed of the same kinds of rocks. Here. 

 fortunately, we have some detailed knowledge, from numerous descrip- 

 tions and analyses. The average density of Scandinavia is 2.759 — that 

 is, it is one of the lightest land areas whose density we know. Only Utah- 

 Nevada, Balkania, Alaska, and the Andes have lower densities. We know 

 little in detail of the igneous rocks of Angara Land, but here also we 

 know that the prevailing igneous rocks are similar to those of the shields 

 just mentioned, although later extensive basalt flows occur. Much of the 

 same is true of Amazonia, although the eastern present border of this 

 shows a rather high density. Of the rocks of the less well defined Archi- 

 plata shield we know little. It is interesting to note, in this connection, 

 that Fisher 20 states that the Falkland Islands show a marked negative 

 gravity anomaly, being thus exceptional among island stations. Accord- 



80 O. Fisher : The physics of the earth's crust p. 250. 



