which contains aircraft oscillatory motion caused by auto-pilot "hunt- 

 ing" and any turbulence. In this case the oscillations had a period of 

 about 80 seconds and an amplitude of about 90 minutes of arc. "Smooth- 

 ing" consisted of drawing a mean line through the oscillations. This is 

 a somewhat primitive but effective technique, and accuracy of the mean 

 line is estimated at + 3 minutes of arc. Automated filtering routines 

 are available that would accomplish the same purpose, but I doubt that 

 they would achieve any significant improvement in accuracy. The TH (G- 

 Ge) may be determined for any desired time interval, since the analog 

 data are continuous. In this case the data were sampled at 1.0 minute 

 time intervals (approximately 6.4 km). Considering that the average 

 wavelength of the F anomalies at this flight elevation (3.1 km ASL) is 

 about 25 km, this sample interval should be sufficient to adequately 

 describe the anomalies. 



The MH analog trace contains the same oscillations as does the G 

 trace and is reduced in the same manner. 



With these two observations (TH; MH) , D: (TH-MH) is established. 

 These computed observations are shown in Figure 10 together with obser- 

 ved F and I. 



Figure 11 shows anomalous (subscript a) magnetic components (IGRF 

 removed) computed from the information shown in Figure 10. Y' is rela- 

 tively smooth as it should be if the source anomalies are in fact 2- 

 dimensional. Some small (<100 nt) anomalies are present that may result 

 from local 3-dimensional bathymetric features (Blakely and others, 1973) 

 that are not particularly well defined on the bathymetric chart (Fig. 2). 



32 



