Jin Wu 
from waves of capillary range is independent of the wind velocity and 
the coefficient B has a different value for the hydrodynamically rough 
flow regime; or k,, is proportional to U, ¢ The former cannot be 
true, because any downward shift of the Hess data shown in figure 
17, in order to get rid of the contribution from waves in the capillary 
range, would mean that the contribution from gravity wave compo- 
nents to the slope at the lower portion of high wind velocities is lower 
than that at the upper portion of low wind velocities. Furthermore, 
referring to (13), the cutoff wavenumber obtained from the intercept 
of the straight line fitted to the data at high wind velocities is simply 
too great to be reasonable. 
In summary, the data seem to indicate that the mean-square 
slope is contributed by wave components from only the gravity range 
at low wind velocities ( Dio < 7 m/sec), and from both gravity 
and capillary ranges at high wind velocities. In addition, the cutoff 
wavenumber, or as considered by Phillips (1966), the neutrally stable 
wavenumber is proportional to U , the square of the wind velocity 
measured at he standard anemometer height, Following this conside- 
ration, we now plot U 2 versus the difference between the slope mea- 
sured at high wind velocities and that contributed by gravity wave 
components, and rewrite (11) as 
st°- By (ky U,)°/e)=Biln(k, /ky) (16) 
Therefore, the straight line fitted through the replotted data shown 
in figure 18 allows an independent determination of the spectral coef- 
ficient B' and the dependency of k, onU,_. It is noted that the trend 
of the data shown here provides a good verification of the form of the 
Phillips' capillary-wave spectrum, which has not previously been 
verified by observations. 
The value of B' is found to be 
Bi = 3" 15x10 7* 
which is about the value 1.5 x Lo. offered by Phillips (1966). Com- 
paring the curves, fitted by the spectral coefficients shown in figures 
17 and 18 with those shown in figure 4.17 of Phillips (1966), one is 
inclined to believe that the former may be more accurate than the 
latter. Moreover, Phillips adopted Miles (1962) calculation of the 
neutrally stable wavenumber as the cutoff wavenumber of the slope 
data in his process to obtain B' , while the present technique allows 
an independent determination of B' and the cutoff wavenumber. The 
1454 
