Plate 
ta which might not be applicable. The data at small fetch parameters 
are all by Mitsuyasu, and do not agree with the data of Hidy and Plate 
(1966) reproduced in figure 1 and which allow at most the conclusion 
that R~3. 10-2 for small values of the fetch parameter, Also, as 
Liu states, the value at the largest fetch parameter is a result per- 
taining to a spectrum not in equilibrium, because the wave pattern 
has not yet adjusted to the decrease in wind speed which had taken 
place. While one may debate if the rate of change of 8 is properly 
expressed by the curve shown in figure 4, it appears conclusive that 
8B as wellas £4 decrease with increase of fetch parameter, or more 
exactly with decrease of “y. But is it proper to associate this de- 
crease with the decrease of the drag coefficient, as the model of 
Longuet-Higgins and equation 2 implies? I think that while the conclu- 
sion might be correct, - as one cannot disprove at this time, - the 
model is certainly oversimplified, in more than one respect. To be- 
gin with Longuet-Higgins' model neglects the part of the wind energy 
that is transmitted to the drift current. It would differ by a constant 
factor if this was a constant percentage of the total, but all indica- 
tions are that this partition depends on the ''wave age'' c/uy where 
c is the phase-speed of the dominant wave. It is at present not clear, 
which fraction of the work done by the wind goes into the waves, and 
which fraction increases the energy of the drift current. Opinions on 
this differ, from assumptions like those of Longuet-Higgins to those 
of Manton (1971) according to whose model a maximum of only 52% 
of the total shear goes directly into wave motion. 
A second aspect not considered in the model of Longuet- 
Higgins is that wave growth changes with fetch. Consequently, some 
of the energy fed into the waves is convected further downwind rather 
than being dissipated through the wave breaking, a feature that must 
be considered in energy balance models. It is the reason for the exis- 
tence of the fetch graph. In fact, by assuming zero wave breaking and 
a wave shape which remains one of constant maximum slope at all 
fetches Deardorff (1967) was able to quite adequately predict the fetch 
graph, without however being able to provide an explanation on how 
a wave of this sort can exist. We must therefore point to this effect, 
although a mathematically and physically accepted model for the wave 
development with fetch appears at present not to be available. 
V. THE DOMINANT WAVE AND THE SPECTRUM 
Apart from the properties of the wave acceleration limit 4 
there exists an important difference of laboratory and field data in 
the shape of the dominant wave itself. It was already mentioned that 
1380 
