THE ANTARCTIC CONVERGENCE 191 



columns in Table 4 headed 'No.' (number of observations). A larger body of data would no doubt 

 result in corrections to most of the figures, but there are two purposes for which the table should be 

 useful. First it is helpful in drawing the charts of surface temperatures with which the latter part of 

 this paper is concerned, and secondly it should sometimes make it easier in the future to locate the 

 convergence. For instance, there is sometimes more than one sharp change of temperature in the 

 neighbourhood of the convergence and the figure given above under the appropriate month and latitude 

 may help to show which actually represents the convergence. 



Fig. 6 shows a rapid increase of temperature from October to February, though the increase is 

 slightly retarded from November to December. Although the curves are drawn tentatively there is 

 evidence for this check between November and December. It can be seen in each set of plotted points 

 in Fig 6 and the averages for these two months are based on more data than most other months. 

 Before their work was interrupted by the war Mr A. J. Clowes and the late Mr J. A. Nicholson made 

 a calculation of the seasonal rise and fall of temperature in part of the Antarctic surface layer by 

 statistical treatment of the vertical temperature readings at a number of stations made by the 

 'Discovery IF on the Greenwich meridian, and the curve so obtained showed a similar reduction m 

 the rate of increase of temperature. This work was not published, but they suggested that the reduction 

 might be a latent heat effect resulting from the melting of ice. 



Table 5 further indicates that the middle temperature varies through an average annual range of 

 about 2-7° C and by about i-o° C for 16° of latitude. The actual highest and lowest middle tempera- 

 tures were 5-89° (No. 15, Feb., 58° S) and 1-30° (No. 86, Oct., 62° S). Table 6 is a comparison of 

 actual middle temperatures with the expected temperatures given in Table 5. 



Table 6. Deviation of observed temperatures from smoothed average middle temperatures 



of the convergence 



When the convergence lies north of its normal position the middle temperature is usually (though 

 not always) a little higher than the average, and vice versa. This suggests that variations m the position 

 of the convergence are not dependent on variations of the sea temperature. 



It is not certain that Table 5 is equally apphcable to all longitudes. There are insufticient data from 

 many parts of the Southern Ocean to test this, but there seems to be no reason why the middle 

 temperature should vary in different longitudes. „ j , 



When the convergence is crossed the amount of change of temperature at the surface, called here 

 the 'range', may be as much as 4-0° (No. 54, 162° E) or may be so small as to be mdistinguishable 

 from other minor fluctuations of temperature (e.g. No. 138 in 01° E). A typical well-defined con- 

 vergence is seen in Fig. 5 B (p. 187), and here the range is taken as from 1-5 to 3-5 , ^-e- 2-o C. Ihe 

 range varies mainly according to longitude; that is to say, it is generally distinctly greater in some 

 regions of the Southern Ocean, than in others. There seems to be no regular variation in the range 

 at different times of year; it does not seem to be affected when the middle temperature of the 



