SECT. 3] THE SOUTHERN OCEAN 287 



those made in summer (Brennecke, 1921 ; Lyman, 1958) reveal water with 

 salinities as high as 34.60 to 34.75% and cooled to freezing point lying in 

 depressions on the continental shelf. Surface observations at the northern end 

 of Graham Land (Deacon and Marshall, 1954) indicate a similar winter maxi- 

 mum. There are no vertical series of temperature and salinity observations 

 near the continental slope in winter among which we can look for evidence of 

 cold, highly saline water sinking from the shelf, but there can be little doubt 

 that it takes place on a large scale. It may also take place along the east coast 

 of Graham Land, though the observations of the Deutschland as she drifted 

 northwards through the middle of the sea during the winter of 1912 show a 

 continuous though much weakened warm deep layer. The closest indication of 

 complete convection from top to bottom observed so far is in the region between 

 the northern tip of Graham Land and the South Orkney Islands (Deacon, 

 1937, p. 108). 



From the Weddell Sea the temperature of the bottom water increases to- 

 wards the east and north all round the continent and the increase is accom- 

 panied by a small increase of salinity and decrease in dissolved oxygen content. 

 The ocean bottom temperatures off the Pacific side of the Graham Land 

 peninsula are 1°C higher than those on the Atlantic side. Very cold, highly 

 saline water is found on other parts of the Antarctic shelf, but as far as we know 

 it is confined in depressions on the shelf and has little or no influence on the 

 deep layers of the ocean outside. The regions where it has been found are 

 the Ross Sea. where the salinity reaches at least 34.87% in winter (Deacon, 

 1939 and 1939a; Lyman, 1958), and Adelie Land, where summer observations 

 show cold water with a salinity of 34.70% at the bottom of a depression on the 

 shelf. Winter observations made by the Gauss on the shelf in 90°E (Drygalski, 

 1926) show that the salinity of the surface water was never as high as 34.51% , 

 but later observations by the Ob indicate that this salinity is exceeded a little 

 farther east. More observations, especially in winter, are needed for a complete 

 picture, but those available so far indicate that the Weddell Sea is by far the 

 greatest source of Antarctic bottom water, and they suggest that its bottom 

 water is an essential constituent of the mixtures which spread northwards in 

 the Pacific and Indian Oceans as well as in the Atlantic Ocean. The tempera- 

 ture and salinity characteristics of the newly formed bottom water are not 

 likely to vary since they are determined primarily by those of the deep water, 

 but it is reasonable to expect that much more is produced in winter. This is a 

 very interesting question to pursue. Stommel and Arons (1960) estimate that 

 the flow from the Weddell Sea must be at least as great as that of the highly 

 saline water from the North Atlantic Ocean. There is some evidence from the 

 behaviour of sounding lines and dredges, from the movements of plankton and 

 the hardness of the sea floor that the bottom current is unusually strong. 

 Radiocarbon "age" determinations south of New Zealand have been inter- 

 preted as evidence of very slow movement, but there are many uncertainties. 

 It is doubtful whether there is time for the carbon dioxide of the shelf water to 

 get into equilibrium with that of the atmosphere, and we know that the shelf 



