indicates that one would have to be extremely fortunate to extrapolate from the 

 conditions under which Priestley took his limited data to those of high wind speed/ 

 a very rough, free water surface/ and strong atmospheric instabi'ity. In order to 

 compare the theory of Phillips' with measurement, then, it has been necessary to 

 multiply all theoretical values of flC by a constant factor in order to bring their magni- 

 tudes up to those observed, it should be pointed out that while such multiplication 

 changes the magnitude, it does not affect the functional form ofd. Hence, 

 the shape of the curves shown in Figure 15 are presumably independent of the scaling 

 factor. 



The values of wind speed used to evaluate CC were 30 knots upwind and 33 

 knots downwind. These values were selected because, not only do they give the 

 best agreement between theory and observation, but also the wind is not known to 

 within the limits implied by these choices. In lieu of more exact wind measurements, 

 then, these selections allow the fairest test of the theory. 



The comparison between the theoretical and observed values appears to be 

 good. The agreement worsens with increasing frequency, but this is accounted for 

 by an increase in the signal to noise ratio for these frequencies. From equation (4) 

 it will be seen that the addition of noise to F causes an overestimate of CC . The 

 results are still considered quite reasonable. If, in addition to the present favorable 

 results, one considers those obtained by Snyder and Cox (1966), it seems not unreason- 

 able to say that a number of the original aspects of Phillips' theory have been verified. 

 Some uncertainty must remain though, for there are still too many loop holes (scaling 

 factor, wind speed, and Section 7.2.3) to allow an "absolute confirmation".* 



7.2.3 Exponential Wove Growth 



The estimates of the parameter of exponential wave growth, ^8 , are shown on 

 Figure 16 for both runs and all directional assumptions. That the estimates of /8 have 

 little dependence on the form of K is somewhat obscurred by the fact that the data 

 from the two runs are not as similar as one would like. There are at least two possible 

 explanations for this apparent result: 



(i) The data upon which the lowest frequency estimates of /3 are based is 

 rather scattered. One could put a straight line (/3 = 0) through the data and obtain 

 almost as good a fit since CC is but slightly affected and ^ is small anyway. For the 

 higher frequency estimates of /S (on the upwind run) the higher signal-to-noise ratio 

 will cause lower estimates for reasons discussed in the following paragraph. Hence, 

 the apparent near constant value of /8 with frequency. 



*lt has been rumored that such a thing does exist in geophysics. 



41 



