Discussion of wave pole, stereo and theoretical spectra 

 If a physicist were to measure the acceleration of gravity at the same 

 place by two different methods and obtain 980 cm/sec by one method ajid 

 1400 cm/sec'^ by another method, he would be positive that there was some- 

 thing wrong with the second method. In this study one is not in so fortunate a 

 position. There is no background of previous experience, and sampling varia- 

 tion must always be recognized as a source of any disagreement. 

 The results obtained so far are that: 



(1) A frequency spectrum obtained from stereo wave data agrees with a 

 theoretical curve derived by Neumann after correction for the presence of 

 swell and the effects of white noise and column noise in the original data. 



(2) A frequency spectrum obtained from a wave pole observation does not 

 agree with either the one derived theoretically or obtained from the stereo data 

 at two points at the one percent significance level. However, the wave pole 

 spectrum does agree with the theoretical spectra given a one knot variation in 

 the v/inds as pointed out in Part 10. 



The following hypotheses are among those that could be advanced to ex- 

 plain the results: 



(1) The agreement between the stereo spectrum and the theory is 



fictitious. It has been obtained by choosing just the right weighted average 



of winds reported quite a few hours before the actual observations of the 



waves and by rather prejudiced choices of just the right amounts of noise and 



swell to get agreement. Also the reduced stereo data may still be distorted. 



212 



