192 DISCOVERY REPORTS 



convergence is higher or lower than the relevant average figure in Table 5 ; nor does it appear to vary 

 according to the latitude of the convergence in its mean position ; but there is almost certainly a tendency 

 for the range to increase a little if the convergence is displaced to the north or south of its mean position. 

 This effect is not obvious, for it is obscured by the more important regional variation, and more data 

 would be needed before it could be measured. 



For measurement of the regional variation the average range is plotted in Fig. 7 in convenient groups 

 of longitude, usually between meridians 10° apart. Thus between 20 and 30° W there were five crossings 

 with an average range of 2-4° C ; between 5° E and 5° W there were ten with an average range of 1° C. 

 Here again a curve is drawn to mark the trend of the plotted points. The data for this curve are rather 

 uneven. It is well supported in the Atlantic region, but in the Indian Ocean there is not very much 

 to rely on, and in the Pacific sector, between 100 and 160° W, the curve is little better than guesswork. 

 The reason for these variations in the range of the convergence gradient is not quite certain and will 

 need some further investigation. 



For the purpose of placing the isotherms about the line of the convergence in Plates II-XII, I have 

 found it convenient to take the range of the gradient at every 10° of longitude from the curve in Fig. 7. 

 The resulting figures are shown in Table 7, which gives an overall average range of 17° C. 



From Plate I and Tables 5 and 7 it is now easy to estimate the expected position of the convergence 

 and change of temperature at the surface, so far as the available data suggest. For example a ship 

 steammg south in 20° E in December would expect to meet the convergence any^vhere within about 

 50 miles north and 50 miles south of 48° 20' S (see Table i and Plate I). Table 5 gives the middle 

 temperature of the gradient as 4-1" and Table 7 gives the range as 1-4°. We should therefore expect 

 the temperature to fall from 4-8 to 3-4° C. The chances of the gradient being warmer or colder than 

 this can be judged from Table 6, but the chances of finding a larger or smaller range are more difficuh 

 to assess. 



Table 7. Provisional estimate of the average range of the convergence gradient in different longitudes 



It is to be hoped that further data will be collected in the future, especially in the winter, and from 

 the Indian Ocean and Pacific sectors. It is desirable, for example, that Tables 5 and 7 should be 

 recalculated from fuller material, and much useful information might be obtained if the line of the 

 convergence at the surface could be followed and mapped out for considerable distances. 



