546 JOSEPH BAKRELL 



co-ordinate axes of the mass unlike. This would require some 

 observations in two directions at right angles from the epicenter. 

 More complete data consisting of both deflection and gravity 

 observations would of course give still closer approximations toward 

 the real form and depth of the mass. What it is desired to show 

 here, however, is that the interpretation of this large mass as a 

 sphere gave a depth to the center of mass about twice too great and 

 a mass perhaps four times too great. This result is in line with the 

 general deduction previously made in regard to the direction of 

 the error involved in the interpretation of deflection residuals as 

 due to spheres. It contributes its individual testimony to show 

 that the masses producing the notable gravity anomalies and deflec- 

 tion residuals are situated within the zone of isostatic compensation 

 and more especially in the upper part of that zone. 



In regard to the large positive mass No. 5 the data are less 

 determinative. At a distance of 60 km. southeast from the epi- 

 center the anomaly is only +0.005 dyne. At 150 km. northeast it 

 is —0.027 dyne. It thus appears that there are some large nega- 

 tive masses easterly of No. 5. As this, however, is on the side 

 away from the line of traverse the problem of the real form and mass 

 of No. 5 is at present indeterminate. 



The adjustment between the deflection curves due to spherical 

 masses and the values of the deflection residuals has been made 

 closer, perhaps, than the probable values of the residuals. Further- 

 more, the residuals of Solution G, if they had been given for this 

 region, would have required a somewhat different distribution of 

 masses. A solution for that depth of compensation which would 

 reduce to the smallest quantity the sum of the least squares of the 

 residuals of this area 1,000 km. long and 400 km. wide would be 

 still somewhat different. That local solution which would give the 

 smallest residuals would be such as would make small the algebraic 

 sum of the positive and negative masses, but the difference in mass 

 between positive and negative centers would not be much reduced. 

 The depth to the centers of mass would be the quantity most 

 affected by a change in the hypothesis regarding the depth of com- 

 pensation. These epicenters then, in so far as the accidental errors 

 do not vitiate the values of the residuals, are realities in nature. 



