152 



WORZEL AND HARRISON 



[chap. 9 



deep ocean. The uncertainty about which density to use in the Bouguer reduc- 

 tion may make it preferable to compute the attraction of trial structures, 

 including the assumed sub-bottom mass distribution and the varying water 

 depth, and to compare this attraction directly with the free-air anomalies. 

 The methods are ultimately equivalent as the attraction of the varying depth 

 of water is taken into account in both. In general, the Bouguer anomaly is 

 more useful in fairly small-scale problems and the free-air anomaly in larger- 

 scale problems, though there is a wide scope for personal preference. 



Computation of the topographic correction is troublesome. When submarine 

 measurements were being accumulated at a relatively slow rate, it was possible 

 to consider computing this correction manually at each station, using the 

 classical methods inaugurated by J. F. Harford and W. Bowie. A template 



Fig. 12. Continental slope profile off Southern California. (After Harrison, 1959.) 



divided into zones by concentric circles, each zone being divided into compart- 

 ments by radial lines, was used and the average height or depth in each compart- 

 ment estimated. Tables (see Heiskanen and Vening Meinesz, 1958, §6-2, pp. 159 

 et seq.) are available from which the contribution of each compartment at the 

 station positioned at the central point of the template can be read, and the 

 total correction determined by summing all these contributions. This method 

 is laborious and it is no longer practicable to attempt to keep pace with the rate 

 of accumulation of gravity data, now that thousands of miles of continuous 

 gravity measurements can be taken yearly. The advent of high-speed digital 

 computers provides the ultimate answer to this problem, but it will take several 

 years of experience to discover and standardize the best procedures. Un- 

 fortunately, the correction depends mainly on the topography very close to 

 the station, and it must be described for the machine in considerable detail. 

 However, the depth of each unit area need only be read off once. Talwani, 

 Worzel and Landisman (1959) have described a method suitable for two- 

 dimensional structures. Quite a number of problems can be treated two-dimen- 

 sionally to a good approximation, and the method may also be as satisfactory 



