H. G. Farmer in conversa.tions with, the author has described how he 

 would modify the WHOI wave pole by putting the tanks a.t greater depths so 

 as to improve the response of the instrument. This should certainly be a 

 subject for further investigation and study both theoretically and by means 

 of model studies. 



Composite f requency spectrum 



The results of the frequency analysis of the stereo data and the wave 

 pole datas as given in Tables llo5, 11.6 and 10.1, can now be connbined to 

 yield a composite frequency spectrum over a full range of frequencies. T?ie 

 spectrum for the stereo data is assuxned to be correct for low frequencies, 

 and the wave pole spectrum is surely quite reliable at high frequencies ex- 

 cept perhaps for a small amount of Vi'hite noise. As k varies from to 10 

 the entries in tJie second column of Table llo5 from the stereo data will be 

 used« As k varies from 11 to 22. the sums of the columms in Table 11.6 

 will be used in. order to remove swell from 'the spectrum. One can note 

 small differences between the entries in Tables 11.6 a.nd 11,5 due to round- 

 off errors a,t high frequencies. The errors are small compared to the vari- 

 ability in the sam.ple. For k from 23 to 27 the stereo values and the wave 

 pole values agree and an average weighted according to the computed num- 

 ber of degrees of freedom is used. For k greater than 2,7, the wave pole 

 values are used» This composite spectrum is given in Table 11,10. 



216 



