106 W. D. MacMILLAN 



compensation, (c) compensation in a subterranean layer, (d) the 

 Chamberlin compensation. In so far as merely satisfying the obser- 

 vations it is found that any one of these hypotheses is as good as 

 any other, since each of them reduces the sum of the squares of the 

 residuals by about 90 per cent. This is a very notable reduction 

 and it places the hypothesis of isostasy upon a solid foundation of 

 credibility. We say "credibility" and not "proof" advisedly, 

 for there still remains the possibility that some other hypothesis, 

 non-isostatic in its nature, may satisfy the observations even 

 more closely, and it is a very difficult matter to show that no such 

 hypotheses exist. 



There seems to be nothing inherently improbable in the notion 

 that the density of the materials under the continents is less than 

 that under the oceans. Hayford's success, which must be con- 

 sidered a notable one, consists in showing, by very complete com- 

 putations which extend over a large mass of data, that assumptions 

 of very moderate differences of density are sufficient to bring the 

 observations and theory into fairly close accord. Whether or not 

 any other hypothesis will or can be equally successful must of course 

 be left for the future. Until some such hypothesis makes its appear- 

 ance we are fairly entitled to put our faith in the broader outlines 

 of isostasy and leave it to further observations and discussions to 

 make the details of the theory more precise. 



Notwithstanding the fact that all four of the hypotheses dis- 

 cussed by Hayford satisfy the observations equally well, it would 

 seem as if Hayford prefers the hypothesis of uniform compensation 

 to a depth of 122 kilometers and usually has this hypothesis in 

 mind when thinking of isostasy. This preference, which does not 

 seem to be warranted by his own discussions, 1 is somewhat danger- 

 ous in that conclusions which are peculiar to this hypothesis are 

 given a prominence to which they are not entitled. Thus, one is 



1 Second, it is not possible to ascertain whether this compensation is more 

 probable than the G compensation, uniformly distributed from the surface to the 

 depth 70.67 miles, since the two sets of computed deflections agree so closely that 

 their differences are much smaller than the accidental errors. — The Figure of the 

 Earth and Isostasy from Measurements in the United States, p. 159. 



A corresponding statement is made on p. 162 with respect to the Chamberlin 

 compensation. 



