196 



DISCOVERY REPORTS 



Georgia and in the Bransfield Strait. The tracks were plotted on the usual semicircular charts. The 

 scale was 6 mm. to a degree of latitude, and it was thus possible to mark the ship's hourly positions 

 by dots at intervals of a millimetre or less (i mm. = 10 miles). The ship's positions recorded in the 

 logs, usually at 4-8 hr. intervals, were plotted first and the hourly positions interpolated. Such plotting 

 of the ships' tracks over many thousands of miles was naturally very laborious, but the resuhing charts 

 should in due course facilitate the plotting of various observations in addition to surface temperatures. 

 A small part of one of these charts is reproduced on the original scale in Fig. 8 A. These track charts 

 were then compared with the thermograph charts and the ship's track was marked in pencil at each 

 point at which the temperature line crossed the line of a whole or half degree C (-0-5, o, 0-5, i-o°, etc.). 

 The hues of the ships' tracks, together with the temperatures were then traced on to monthly tem- 



Table 8. Frequency of differences between the temperature at o m. and at 10 m. {including all records 



where the temperature at o m. does not exceed 5° C) 



perature charts, all data for any one month being included on one chart or set of overlapping charts, 

 irrespective of the year. The year was, however, marked against each track. A comparison of Figs. 5 and 

 8 will^show the method of plotting. For example, in Fig. 5 A (p. 187) the thermograph record crosses 

 the 3° level at noon on 14 October 1930. The position of this point is found in Fig. 8 A just north of 

 50° S, and hence 3° is marked in the corresponding position on the ship's track in Fig. 8 B. 



A large number of temperature records at single positions were then added to the temperature 

 charts. These included the surface temperatures read at stations made by the Discovery Commhtee's 

 ships when no thermograph chart was available (mainly derived from the 'William Scoresby'), and 

 published records from other sources. 



Part of a chart of October temperatures is reproduced as an example in Fig. 8 B. The mean isotherms 

 indicated here are in their final position, as shown in Plate II, after comparison with the corresponding 

 isotherms of other months. They represent the temperatures for both September and October, which 

 are very similar; but there is not room to show the observed temperatures for September on the same 

 chart, for in places the ship's tracks for the two months almost coincide. It should be mentioned that 



