342 THE DEEP SEA 



The two-layering process suggested here represents a balance 

 struck by the ascending cold water and the turbulent descent of 

 heat. In the last few years many different turbulent mechanisms 

 have been suggested as responsible for this vertical heat flux. 

 Three such mechanisms are given. Their variety indicates the al- 

 most complete lack of knowledge of the actual oceanic process. 



First, the energy to drive a convective heat flux in a vertically 

 stable fluid could come from the horizontal flows induced by the 

 wind. If this is so, the total mixing should be related to the wind 

 speed and the density contrast in the sea. Hence, one would expect 

 seasonal variations in the vertical temperature structure where the 

 winds show marked seasonal variation and little seasonal variation 

 in regions where the winds are steady. 



A second energy source for convection could be internal waves. 

 It appears that internal waves would ha\"e to break to effectively 

 mix. There is plenty of tidal energy a\-ailable to drive internal 

 wa\'es but no one has yet explored the fashion in which this energy 

 might be made available to the convection. 



A third convective mechanism is that one responsible for the 

 reincorporation of the seasonal thermocline. This appears to be due 

 to winter cooling of the surface waters. In fact the long persistence 

 of the warmed surface layer produced in summer suggests that 

 other mixing processes are quite weak. Howe\'er, parts of the 

 tropical ocean have almost no seasonal thermocline and here con- 

 vection must be due to other causes. 



These different processes for the vertical flux of heat vv^ould have 

 motions with quite different characteristic space and time scales, 

 yet whatever the actual process may be, the continuity require- 

 ments of our idealized circulation prescribe the total amount of 

 heat which must be convected. In order to establish how much 

 larger than the microscopic heat transport the effective turbulent 

 transport must be, we may define an average turbulent coefficient 

 K as equal to HDJ b,T. I suggest no other function for an "eddy 

 conductivity" than this integral estimate. For example, use of 

 such a constant K in a dynamical treatment of the seasonal mixing 

 process is clearly of little value. 



To determine H, D, W, and K we must know AT, Ts, and A. 

 An estimate of AT" from the obser\^ations is approximately 15°C. 



