different wave pole spectra as originally tabulated shows no effect of wind 

 variability a,t the 5 percent level, and the low value at k w 15 occurs in all 

 three cases, 



The third hypothesis is one that cannot be tested except by doing the same 

 experiment over again using the sanae wave pole under similar meteorological 

 conditions. Xt will be rejected as a working hypothesis solely because it can- 

 not be tested. However, due to the possibility of this hypothesis combining 

 with some of the others in part, the possibility of incorrectly rejecting it with 

 a chance of more than 0,01 (say 0,15) must be borne in mind. 



The fourth hypothesis is a very attractive one. If the wave pole cali- 

 bration curve were more like the one sh,own by the dashed curves in figure 

 J. 1,25 th^^n. the theoretical one, there would, be agreement between both obser- 

 ved curves and the theory. It will therefore be assumed that this hypothesis 

 is the dominant explanation for 1;he discrepancies which have occurred. 



This hypothesis can be tested by modeling the wave pole in a scaled 

 down long crested Gaussian sea with the correct model frequencies present, 

 and connparing t^e record it makes with a record made by a wave pole held 

 fixed in position, 



An irregular sea is suggested for the tests because a non-linear effect 



.1, 

 of considerable magnitude may be present. In Part 8 it was shown that the 



wave pole moYes upward when the crest of a long period wave passes^. The 



submerged tanks are therefore closer to the mean level in the crest of a long 



period wave, If the crest of a shorter period wave is present at the same 



214 



