Thus the total variance of the white noise reading error is approximately 

 0.54 (ft)^, and the quantity 0.54/800 (ft)^ should be subtracted from each of 

 the tabulated values of U{r, s) to correct for this effect- 

 Moreover the spectrum for data Set 3 at the origin definitely shows the 



effects of curvature. The average spectrum also shows an effect of curvature 



■ 



in the peak at the origin of the spectrum and in the distortion of the contours 



near the origin. The magnitude of the effect can be estimated from the spectrunn. 

 ^ There is a hint of column noise in both of the covariance surfaces and in 



the average covariance surface. There is a fairly strong ridge along the verti- 

 cal axis of all three figures. However, these ridges do not produce the pre- 

 dicted effect of a ridge along the horizontal axis of the spectra. Thus if the 

 column noise is present it is masked by some other more serious source of 



terror. 

 White noise and curvature error both add positive quantities to the spec- 

 trum when they occur. Corrections to the total variance of the original data 

 can be calculated from the above information and the results are tabulated in 

 Table 11.1, 



As seen from Table 11.1 the corrected variance of the combined data is 

 3.80(f^^. From the study of the wave pole spectrum, assuming correct cali- 

 bration, confidence bounds on the E value were set and it can therefore be cal- 



2 2 



culated (by taking half the value) that the range from. 2.23 (ft) to 2.64 (ft) 



would enclose the true value of the variance of the wave pole data nine times 



out of ten. The true variances of the wave pole data and the stereo data should 



169 



