232 DISCOVERY REPORTS 



Nansen (191 2) explains the formation of similar water in the Arctic Ocean as due to 

 a convective circulation which carries water downwards from the surface to the bottom 

 in autumn and early winter. Brennecke (1921, p. 140) says that it is formed along the 

 continental shelf of the Antarctic Continent where the surface water is cooled right 

 through and sinks down the continental slope. Drygalski (1927, pp. 495 et seq.) states 

 that the bottom water is a mixture of the cold water formed on the continental shelf 

 with the deep water. Wiist (1928, p. 525) at first followed Nansen's explanation. He 

 thought that when the surface currents became slower in autumn and winter Antarctic 

 bottom water was formed by the sinking of highly saline surface water in the centre of 

 two cyclonic movements, one in the Weddell Sea and one farther east. 



The difficulty which prevented acceptance of the explanations of Brennecke and 

 Drygalski was that there was always found to be a continuous warm deep layer, even 

 at stations very far south, through which the cold water formed on the continental shelf 

 did not appear to sink. There are, however, insufficient observations to show that this 

 sinking of water does not take place, particularly in winter along the east coast of 

 Graham Land. 



In a more recent publication (1933, p. 48) WiJst has concluded from the distribution 

 of potential temperature (the temperature to which the water would be adiabatically 

 cooled if it were raised to the surface) that Antarctic bottom water in the Atlantic is formed 

 in the way suggested by Brennecke. He has also distinguished a slightly warmer bottom 

 water which he has called Antarctic deep water. He suggests that the two cyclonic water 

 movements distinguished by Meyer (1923) in the Weddell Sea, and in 30° E, do not 

 exist, and that there is instead a convergence region similar to that shown by Willimzik 

 (1927) and Moller (1929) in the Indian Ocean. He considers that the convergence 

 region lies between 60° and 65° S in the Weddell Sea, and between 56° and 60° S 

 farther east. Its position is closely related to that of the northern edge of the pack-ice. 

 Antarctic deep water — according to Wiist — would be water which sinks in this region 

 in autumn and winter. 



The existence of a convergence region in the Weddell Sea is however hypothetical ; 

 our observations and those of Brennecke seem to show that there is simply a diverg- 

 ence region between the currents flowing west and east. In this divergence region 

 warm deep water and Antarctic bottom water upwell, and tend to flow outwards (to 

 the left) in, or just below, the surface layer. At the edges of the sea, and particularly 

 along the east coast of Graham Land, heavy water will sink downwards. Most water 

 will probably sink in winter when the surface water is coldest and most saline. The 

 water which sinks need not have a salinity as high as the warm deep water because it 

 is colder, and because the sinking is partly due to dynamic forces. 



The observations of the ' Deutschland ' (Brennecke, 1921) show that the surface 

 water in winter has a salinity of almost 34-50700' ^^^ closer inshore it may be much 

 more. By the sinking of this water and its mixture with warm deep water, which in the 

 west of the Weddell Sea has a maximum temperature of 0-40° C. or less, both Antarctic 

 bottom water and Wust's "Antarctic deep water" can be formed. If they are both 



