434 



WILLIAM BOWIE 



the stations used as a single group for the various depths of com- 

 pensation had a range of only 0.002 dyne from 0.020 to 0.022. 

 Owing to the small range, it is not necessary to show the means 

 without regard to sign in the table. The anomalies in this table 

 have not a very wide range, and therefore there is no very strong 

 evidence to show that any one depth among the intermediate depths 

 is much better than the others. The evidence, such as it is, favors 

 the depth of about 85 km. 



TABLE V 



Anomalies at Different Depths of Compensation 



Character of 

 Topography 



Number 

 of Sta- 

 tions 



Mean Anomalies with Regard to Sign for Depth in Kilometers 



42.6 



S6.9 



8S-3 



H3-7 



156.2 



184.6 



Coast 



Near coast 



Interior, not in 

 mountainous re- 

 gions 



Mountainous re- 

 gions, below 

 general level . . . 



Mountainous re- 

 gions, above 

 general level . . . 



27 

 46 



—0.002 



+ .002 



— o . 003 

 + .002 



— 0.006 



+ .001 



- .003 



87 

 36 



20 l+o. 021 



+0.016 



+0 . 009 



— 0.007 

 + .001 



+ .001 



+0.003 



— 0.008 

 + .001 



+ .001 



+0.001 



— o . 009 

 + .001 



+ -003 



- o . 003 



— 0.009 

 + .001 



+ -005 



— -OQ3 



— 0.006 



In order to get stronger evidence as to the most probable depth 

 of compensation, the stations were divided into groups according 

 to the topography, as was done in several other tests. The mean 

 anomalies without regard to sign for the seven depths considered 

 had a total range of 0.009 from 0.017 to 0.026. Each of these 

 mean anomalies came in the group of stations in mountainous 

 regions, above the general level. The maximum range for any other 

 group of stations was o . 004 and that occurred in the mountainous 

 regions where the stations are below the general level. 



The means with regard to sign varied for the different depths 

 and the different classes of topography. These means are shown 

 in Table V. 



It will be noticed that there is only one group for which there 

 are very decided changes in the mean anomaly with regard to sign — 



