Jin Wu 
clean sea surface, where capillary waves were undoubtedly present 
and should contribute to the mean-square surface slope. On the other 
hand, some of their observations were taken in the interior of a dense 
artificial slick, where waves shorter than 30.5 cm (1 ft) were repor- 
ted by Cox and Munk to be absent. For the latter case, the maximum 
wavenumber for existing waves, designated as k, (with the minimum 
wavelength 7, ), is certainly well outside the capillary range, or in 
other words kg smaller than k,. Therefore, for this portion of 
the data obtained in the interior of artificial slicks. Equation (11) can 
be rewritten as 
2 BLU, ,?/s) kK. (13) 
in which the wavenumber kis substituted with e/a 2 1( Phillips 
1966). It isSnow obvious, if we replot the mean-square slope data in 
the semilogarithmic form and fit the replotted data with a straight- 
line, wecandetermine both B andk, independently. The kg so deter- 
mined can then be compared with the abserved ke 
Replotted in figure 17 in the proposed form and fitted with 
straightlines by means of the least-square principle are the data of 
three different groups : (a) clean sea surface with the airflow in the 
hydrodynamically smooth regime and the transition region, (b) clean 
sea surface with the airflow in the hydrodynamically rough regime, 
and (c) sea surface covered with slicks. For the last group, the va- 
lues of B and k, , determined from the slope and the intercept of the 
fitted straightline, are 
B=4.6 x 10.7: A, = 38cm (14) 
The value of B is identical to that obtained earlier by Phillips (1966). 
Taking into account the scattering of the data and the rather crude 
visual observations of the minimum wavelength in dense slicks, the 
A_ can be considered in rather close agreement with the observed 
value of 1 ft. This agreement also supports the technique used here 
to determine the spectral constant. 
For a clean sea surface, the contribution of the capillary 
waves to the mean-square surface slope, the second term on the 
right-hand side of (11), cannot be neglected. Relative to the data ob- 
tained in the slick, those obtained in the hydrodynamically smooth 
regime of clean water surface are seen in figure 17 to be shifted al- 
most parallel upward and twoward the left. Referring to (11) and (13) 
the upward and nearly parallel shift of the data indicates that the 
1452 
