DISTRIBUTION OF SURFACE TEMPERATURE 201 



observed temperatures, but there were no data here for December. The December isotherm, however, 

 must he between those for November and January, and since the observed temperatures nearly 

 everywhere indicate that it Hes nearer to the November than to the January isotherm there was no 

 difficuhy in deciding where to draw it here (see Plate IV). A good interpolation of this kind may some- 

 times give a more reliable position for the isotherm than a limited set of direct observations. Farther 

 west the November and December isotherms are both interpolated between those for October and 

 January, and in the winter months in the Pacific sector the 0° isotherms for five or six consecutive 

 months have to be interpolated (see Plate XIV). Here of course there is less certainty, especially as 

 the September isotherm rests on observations only from one year (1932). Even so there is not very 

 much choice in the positions in which they might be drawn. 



NOTES ON THE DISTRIBUTION OF TEMPERATURE 



It will be realized that Plates II-XII are semidiagrammatic charts which represent only the major 



features of temperature distribution. Surface temperatures are naturally subject to some appreciable 



widespread variations and to more numerous local variations ; and if the isotherms could be drawn in 



full detail for any area at a given time they would usually 



reveal elaborate complexities, and bear little resemblance 



to the smooth lines in Plates II-XII. Fig. 9 shows the 



apparent actual distribution of temperature between 



South Georgia and the Falkland Islands in September 



and October 1934, and may be compared with Fig. i 



(p. 182) and with the corresponding region in Plate II. 



The isotherms are correctly placed where they cross the 



ship's tracks, and between these tracks they are drawn 



in what seem to be the most probable positions ; but 



there is not much doubt that, if additional lines of 



observations could have been made at the same time, a 



still more complex picture would be presented. 



co- 



Fig. 9. Observed temperatures in the Scotia Sea, 

 1st September to 14th October, 1934, showing the 

 apparent positions of the isotherms. Half-degree iso- 

 Most charts of temperature distribution as observed therms are shown as pecked lines, and the ship's tracks 



at a particular time would be likely to show some features, 



such as tongues of warm or cold water and other bends in the isotherms, which are temporary, and 



which become smoothed out when the isotherms are drawn in their mean positions. Plates II-XII 



are intended to show only those features which seem to be constant, or which at least are found in 



most of the years for which observations are available. Even such constant features have been drawn 



with caution, and it may be that some of them are not sufficiently emphasized. 



The following points may be noted. South of the Falkland Islands in about 55^ W there is some 



evidence of a northward thrust of cold water. This is noticeable in the chart for September and October 



(Plate II) where the isotherms are crowded up to the convergence. For this chart there are data only 



for one year (see Fig. 9), but the bend in the convergence here and the similar shape of the isotherms 



in other months suggest that this is a constant feature. It is perhaps more noticeable in spring than m 



summer. A more conspicuous tongue of cold water projects northwards around the South Sandwich 



Islands (near 30° W). This is often very pronounced, and it is probably connected with a comparatively 



narrow thrust of cold water off the north-east side of South Georgia, which was found during intensive 



observations around the island. Warmer water is found to the east of the South Sandwich Islands, 



and it is possible that the southward bend of the isotherms about 10-20° W is not sufficiently emphasized 



in the charts. It is best seen in the charts for March and April, but it is not certain that it is really 



