176 DISCOVERY REPORTS 



in the eastern South Atlantic there seems to be a tendency to an eastward movement, particularly 

 in the Gulf of Guinea, where it apparently diverges cyclonically from the main northerly trend. Thus 

 in the South-west African region the flow is again ill defined. 



In the subtropical region between the surface-waters, and the salinity minimum of the antarctic 

 intermediate water the T-S curves follow a nearly straight line relationship to within about 200 m. 

 of the surface, above which the circulation becomes more complex within the tropical and sub- 

 tropical surface and subsurface layers. This water mass, represented by the nearly straight line 

 part of the curve, has been termed the 'South Atlantic central water' (Sverdrup, Johnson and 

 Fleming, 1946). The process of its formation is still rather obscure. Sverdrup et al. (1946) 

 suggest that it is probably formed by a simple process of sinking along surfaces of equal density in 

 the region of the subtropical convergence, as the vertical T-S relationship of the South Atlantic 



15- 



10' 



67 , /'4O0 1 ^ 

 (ISO)/ ^ 



1000- 



•\ N A DEEP 



3250 



ANTARCTIC BOTTOM WATER 



M 50 



SALINITY 



/" 



3500 



35 50 



Fig. 31. A typical temperature-salinity curve for the south-east Atlantic, drawn from observations at ' Discovery' station 673. 

 The water masses represented on the curve are named. The additional points in open circles are the T-S relationships at 

 the stations (675, 671 and 668) at the depths (in brackets) shown. These stations lie in a north-south line across the subtropical 

 convergence, so the points show the horizontal T-S relationships at subsurface depths across this region, indicating the close 

 similarity to the vertical T-S curve for the South Atlantic central water. Depths are in metres. 



central water corresponds closely with the horizontal T-S relationship in the subtropical convergence 

 region (see Fig. 31). Clowes (1950), however, has taken exception to this view on the grounds that 

 an examination of the surface T-S relationships in the region o° to 20 E. and 30 to 40 S. fails to 

 show any similarity to the vertical T-S curve of the South Atlantic central water. He considers that 

 a more likely explanation is a direct mixing between the antarctic intermediate water and the sub- 

 tropical surface- and subsurface waters, which would probably take place in the region of the sub- 

 tropical convergence. 



North of the antarctic convergence in about 50 S., where the antarctic surface-water sinks to form 

 the intermediate water, there lies a belt of warmer water, the subantarctic water, whose characteristics 

 suggest that it is a mixture of antarctic water and warmer water from the north. Extending north- 



