i88 



DISCOVERY REPORTS 



temperature and the range are given, but sometimes the former could be distinguished while the 

 beginning and end of the gradient were too ill defined for measurement of the range. In some cases 

 where the change of temperature was small, and the convergence and its gradient could not be dis- 

 tinguished, the middle temperature is omitted, but, for calculating the average, a small figure is entered 

 as an estimate for the range. In one or two other instances an approximate figure could be given for 

 the range though the middle temperature was very obscure (e.g. No. 49). 



Table 3. The change of temperature markmg the Afitarctic convergence at the surface. 

 Middle temperature and range of the gradient 



* Doubtful. 



Analysis of this table indicates that the middle temperature varies mainly according to the time of 

 year, and to some extent according to the latitude of the convergence, while the range varies in different 

 longitudes and appears to depend on other factors. 



Taking the middle temperature first, Table 4 gives the average and maximum and minimum values 

 for each month in arbitrary ranges of latitude. Divided in this way the quantity of data is rather small, 

 and it is evident that the middle temperature can vary considerably in any one month and latitude, 

 but it is obvious from the average figures that it generally falls to a minimum about September and 

 October, and rises to a maximum about February. This is naturally to be expected since the sea tem- 

 perature varies with the time of year while the latitude of the convergence is little affected. The figures 

 also leave little doubt that where the convergence is in a high latitude the middle temperature tends 

 to be lower than where it lies in a low latitude, though the difference is not very great. This implies 

 that surface isotherms near the convergence do not always run quite parallel to it. 



