100 Linear Theories — Viscous 



only about one half the observed values [see table 5]. It does not 

 seem reasonable that the oceanographic observations should be 

 ofiF by more than, say, 20 per cent, nor that the theoretical ex- 

 pression. . .for the transport should account for the discrepancy, 

 since it is almost independent of the eddy viscosity and the shape 

 of the ocean basin. 



Maury and others have ascribed the North Atlantic circulation, 

 in particular the Gulf Stream, to differential heating between 

 equator and pole, to the freezing of ice, and to other processes that 

 make up the thermohaline circulation. If we assume that the circu- 

 lation were half wind-driven, half thermohaline, it would be a 

 strange coincidence that the geneTal pattern of the circulation, such 

 as the boundaries of the gyres, should conform so closely to the 

 general atmospheric circulation. Furthermore, Fuglister [Munk 

 refers to an unpubhshed manuscript, later published, 1951 a] has 

 found a high correlation between variations in the current with 

 variations in the wind. It should also be noted that the thermo- 

 haline circulation insofar as it is related to the outflow of river water 

 along the Atlantic seaboard would tend to reduce rather than to 

 strengthen the Gulf Stream. It would seem therefore that the sub- 

 tropical gyre, and probably also the subpolar gyre, are predomin- 

 antly wind-driven. 



Methods for computing wind stress from the observed wind 

 speeds according to the equation 



T = C^Palr£/^ [38] 



are discussed by Reid [1948a]. . .Underestimates of r may first 

 of aU result from underestimates in the Mind speeds on the chmato- 

 logical charts from which the appropriate averages were taken. 

 The preponderance of coastal stations, and the tendency of ships 

 at sea to avoid regions of high wind, would lead to consistent 

 errors, but these cannot account for more than a fraction of the dis- 

 crepancies. The weakest link is the drag coefficient C^ which is 

 based on measurements of Baltic storm tides, and a few other 

 measurements ... In accordance with the views presently accepted 

 we have assumed 6*2, = -0026 at high Avind speeds, Cj)X-OOS at 

 low speeds, with the discontinuity occurring at Beaufort 4 [Munli, 

 1947]. In the trade- wind belt of the eastern Pacific, where the 

 winds are predominantly Beaufort 4 and above, Sverdrup [1947] 

 and Reid [1948 a] have obtained satisfactory agreement between 

 computed and observed transports. Supposing a value of -0026 

 were apphcable at all wind speeds, then the eflfect of the south- 



