i86 DISCOVERY REPORTS 



the same relative position at each of these points rather suggests some balance of pressure between 

 water masses drifting through those parts of the Southern Ocean which are hmited to the north as 

 well as to the south by land masses. Sverdrup, Johnson and Fleming (1942, p. 607) say: 'The south- 

 ward displacement of the Antarctic Convergence to the south of Australia and New Zealand can 

 probably be ascribed to the relative narrowness of the passage between these regions and Antarctica, 

 and the similar displacement off South America can be attributed to the southerly location of Drake's 

 Passage separating South America from Graham Land.' 



It is also worth while to compare the position of the convergence with the mean northern limit of 

 the pack-ice. The pack extends farthest north about September and October, and the line in Plate I 

 is a reproduction of the September-October line shown by Mackintosh and Herdman (1940, plate Ixix). 

 It is evident that the position of this line bears some relation to the position of the convergence. The 

 extreme limit of the ice will of course lie nearer to the convergence, but it is doubtful whether the pack 

 ever reaches the convergence itself. 



The distance between the convergence and the mean northern limit of the ice shown in Plate I 

 varies from 120 miles in the Scotia Sea to about 550 miles in the Indian Ocean sector, with an average 

 of 370 miles. This may be taken as a belt which is practically free of pack-ice throughout the year. 

 There is some reason to suppose that it has a characteristic plankton fauna and flora. On the basis 

 of the distribution of macroplankton in the Falkland sector I distinguished a ' northern zone ' imme- 

 diately south of the convergence (1934, p. 150), and Hart (1942, p. 280 and Fig. 2) found that the 

 phytoplankton could be suitably divided into certain biogeographical areas, one of which he took as 

 the 'northern region' lying between the convergence and a line 330 miles to the south of it excepting 

 certain special areas. This region was in fact distinguished largely on the grounds that it is normally 

 free of pack-ice at all times. This is a matter to which I hope to return in a future paper. 



THE CHANGE OF TEMPERATURE AT THE SURFACE 



The convergence is generally distinguished by a sharp change of temperature at the surface, which 

 usually appears as a steep gradient on the thermograph, and it is important to consider what variations 

 occur in the position of the gradient on the temperature scale, and the range of the gradient. This is 

 best done by examining first the variations in the middle temperature of the gradient, and then the 

 variations in the range. The aim of the present section is to estimate (as nearly as the data permit) 

 what change of temperature we may expect to find in any longitude at any time of year, and hence 

 how to place the isotherms at the convergence when drawing charts of surface temperatures. 



Deacon (1933, fig. 10, p. 190) gives an example of the gradient as shown on the thermograph. This 

 was No. 59 in Table 9, and it is an exceptionally well-defined crossing. Some further typical examples 

 are shown here in Fig. 5 A-E. The middle temperature is taken as the mean of the temperatures at 

 the beginning and end of the gradient representing the convergence, or, if the gradient is not very 

 clearly defined, whatever seems best to represent the central temperature of the convergence. It is 

 easy enough to decide the middle temperature and range in such examples as Nos. 33 and 106 (Fig. 5). 

 In such as No. 88 it is difficult, but these small and ill-defined gradients cannot be excluded from 

 calculations of the average. In this case the middle point was taken as 3-0° and the range as o-8°, on 

 the assumption that the rise of temperature immediately to the right of St. 1476 represented the 

 convergence. 



Table 3 is a list of all the middle temperatures and ranges which I have been able to measure. It 

 includes some rather doubtful measurements, such as No. 88, but should serve for provisional estimates 

 of the changes of temperature to be found in different months and positions. Usually both the middle 



