3 6 DISCOVERY REPORTS 



and the northward current appears to have two main branches. The temperature dis- 

 tribution below the depth of 2000 m. suggests that there is a similar division of the 

 bottom current ; such a splitting of the bottom current, which may be caused by the 

 irregularity of the sea-floor, affords a possible explanation of the division of the surface 

 current. The temperature observations at Sts. 862 and 863 suggest that the stations are 

 not far from the Antarctic convergence, which probably bends southwards like the 

 isotherms. 



East of 1 00- 1 io° E the isotherms and the convergence start to recede gradually 

 southwards. It has already been shown that the position of the convergence depends on 

 the northward progress made by the bottom current, and the distribution of tempera- 

 ture and salinity in sections 10 and 11 (Plates XIX-XXIV), south of Australia, supports 

 this conclusion, arguing that the convergence recedes towards the south mainly because 

 the bottom current makes less progress towards the north. 



The movements of the deep and bottom waters are, however, more complex than 

 usual, especially in section 1 1 ; the isotherms and isohalines show that the upward slope 

 of the warm deep layer is not steep and unbroken, but gradual and in a series of steps. 

 The result of this abnormality is seen at the surface, where the Antarctic and sub- 

 Antarctic currents are not so sharply distinguished as they are where the warm deep 

 current climbs more steeply. The surface temperature records show that in section 

 10 there was a decrease from 5-5 to 3-7° C. between 52 04' S and 52 24' S, and in 

 section 1 1 an increase from 2-5 to 4-5° C. between 54 35' S and 54 20' S. These changes 

 appear to mark the principal boundary between the Antarctic and sub-Antarctic waters ; 

 and they probably indicate the normal position of the Antarctic convergence; but a 

 closer examination of the data suggests that the sub-Antarctic water sometimes flows 

 farther south. The distribution of temperature and salinity shows that the convergence 

 is formed at the surface as soon as the warm deep water lies deep enough to allow the 

 Antarctic current to sink towards the north below a southward current of sub-Antarctic 

 water. In section 1 1 the warm deep layer starts to slope, and the Antarctic water to sink, 

 between 57 and 58 S, and there seems to be a tendency for warm water to creep 

 southwards at the surface; but between 55 and 57 S the slope of the warm deep layer 

 is more gradual and the Antarctic current does not sink below the main body of sub- 

 Antarctic water until it reaches 54 35' S. Where the deep water behaves in such a way 

 and does not slope clearly and steeply, the absence of a sharp termination to the south- 

 ward movement of sub-Antarctic water is therefore not remarkable. 



The observations made at Sts. 883 and 891, just south of the convergence in sections 

 10 and 11, suggest that the surface water contains a considerable percentage of sub- 

 Antarctic water. At St. 883 the mean temperature in the first 100 m. was 3'73° C, 

 but between 150 and 300 m. it was only 2-16° C. Such a large difference so late in 

 the autumn (May 23) indicates that the two waters belong to different currents, the cold 

 water flowing northwards more rapidly than the surface water. The high temperature 

 at the surface also argues that the surface water is partly derived from a southward 

 movement of sub-Antarctic water. 



