292 DISCOVERY REPORTS 



due as much to the difference between the beginning and end of the month as to the 

 difference between different years. In December, for example, it will be seen that the 

 more northerly positions are mostly recorded early in the month, and the more southerly 

 positions towards the end of the month. 



The Weddell Sea has an important effect on the distribution of ice in the Atlantic 

 sector. The low latitude reached by the ice-edge between 30 W and 20 E may be 

 ascribed to the flow of cold water from the Weddell Sea which exerts an influence far to 

 the east, and maintains a belt of ice in this region for some time after the pack has begun 

 to melt and break up farther south. In the Weddell Sea itself the distribution of ice is 

 very variable and little understood. The Ross Sea, which is shallow and has a less 

 effective current system, has comparatively little influence on the ice beyond its im- 

 mediate neighbourhood. 



SEPTEMBER (PLATES LXXI-LXXIII) 

 Observations in this month are scarce, but the ice is always far to the north. The 

 ' Discovery II ' found the ice-edge north of the South Shetlands and South Orkneys in 

 1934 (see Plate LXXI), and there are three records in about 55 S between South 

 Georgia and 20 E. The only other record is to the north-east of the Ross Sea (about 

 62 S). All these observations were made by the ' Discovery II ' or ' William Scoresby \ 



OCTOBER (PLATES LXXIV-LXXVI) 



There are more data for this month, especially between o° and io° E, and some records 

 from factory ships are included. The three observations close to the South Shetlands 

 (191 1, 1926 and 1930) probably lie near the mean position of the ice-edge at this time of 

 year, but it appears that in some years (e.g. 19 10 and 1924) factory ships reached the 

 Bransfield Strait in October without meeting pack-ice. (An account of ice conditions in 

 the Bransfield Strait is given by Clowes, 1934, p. 59). In October 1930, the ice lay very 

 near South Georgia, but the ice conditions in that spring were exceptional. Between 

 30 W and 20 E the October positions on the average resemble those for September. 

 There are some useful observations by factory ships between 77 and 98 E, and in 

 H7°-i27°Etheice was found well to the south in 1930. There are no observations in the 

 Ross Sea and greater part of the Pacific sector. The ice plotted in the Bellingshausen 

 Sea in Plates LXXIV and LXXVI (1932 and 1934) almost certainly lies south of the 

 normal position for mid-October. Both observations were made at the end of the 

 month. 



Little difference can be distinguished in the latitude of the ice-edge in September and 

 October, and the lines showing what is here assumed to be the mean position in the two 

 months are therefore drawn to coincide (see Plates LXIX and LXX). 



NOVEMBER (PLATES LXXVII-LXXIX) 

 For this month there are records from nearly all longitudes. The only considerable 

 gap is on the Pacific side between 120 and 180 W. There is a tendency for the ice to lie 

 on the whole a little south of its position in September and October. In 19 14 and 193 1 



