308 ANNUAL REPORT SMITHSONIAN INSTITUTION, 1920. 



factor in the consideration of isostasy, to be taken up later. It will 

 be seen from the table that the average density of the crust is cal- 

 culated, from the average chemical composition, to be 2.77. An 

 average might be arrived at by considering all the determinations 

 of specific gravity of rock specimens that have been made by the 

 ordinary physical methods, and that are found abundantly in the 

 literature. An average thus arrived at would seem to suffer from 

 several disturbing factors that are eliminated by the method based 

 on the chemical averages. Thus it would include the densities of 

 many lavas that are more or less glassy, which are decidedly lighter 

 than holocrystalline rocks, and which, furthermore, are surficial 

 rocks, not found at any considerable depth beneath the surface. It 

 would also be seriously affected by the porosity of the surface rock 

 specimens ; and at great depths this must be very largely or wholly 

 done away with by the pressure of the superincumbent crust, as 

 shown by Van Hise and others. On the other hand, however, the 

 density determinations are probably more equably distributed among 

 the various kinds of rocks than are the chemical analyses, which may 

 reasonably be expected to include a possibly undue proportion of 

 " interesting " and rarer types of rocks, as has been mentioned. 



It is impossible at present to evaluate the relative influences of 

 these several factors, but I might incidentally express my surprise 

 that such a simple means of arriving at an estimate of the average 

 density of the rocks of the earth's crust as is here suggested does not 

 seem yet to have been attempted — at least nothing seems to have been 

 published on the subject. An estimation that I am now making 

 along this line is not yet complete enough for publication in this 

 paper, but will be given later elsewhere. 



On the whole, after due consideration of the several factors in- 

 volved, I am inclined to put much greater weight on the final result 

 arrived at from the averages of the chemical compositions. This, 

 also, is subject to certain possible corrections in the future. It would 

 seem to be probable that it is somewhat too high, as it does not 

 include any, or at least a proportionate, number of analyses of many 

 large areas which are almost certainly generally granitic and there- 

 fore relatively light. This applies to the interiors of Asia, South 

 America, Australia, and probably Africa, to mention the larger divi- 

 sions, and also to smaller ones, such as Spain, Egypt, South Africa, 

 the Greater Antilles, and others. It is impossible now to estimate 

 the magnitude of this correction. 



On the other hand, if we are dealing with the rocks of the crust to 

 any (humanly) considerable depth, such as the 10 miles assumed by 

 Clarke, and which might justly be placed at 20 or more, we meet 

 with the possibility of a correction in the other direction; that is, 

 toward a higher density. This conclusion is based on the ideas of 



