DAILY VARIATION— MAY, JUNE AND JULY. 



69 



It would be of great interest to know if this were a feature of the whole of the Antarc- 

 tic in high latitudes, but unfortunately all the other stations in the Antarctic from which we 

 have data aie either very near or just outside the polar circle, hence they received some 

 radiation during these months. They each have warmer days than nights, but this may be 

 due to the sun and therefore cannot be taken as proof of the presence of the feature found 

 in McMurdo Sound. 



The Reality of the Feature : Marked Excess of Temperature at i a.m.— The results of the test 

 for this feature are shown in column 3 of table 28. Four a.m. is warmer than the mean 

 of 2 A.M. and 6 A.M. in six out of the seven sub-divisions into which the data have been 

 divided. It was absent only in the year 1911. There is thus even better evidence for the 

 reality of this feature than was found for the days being warmer than the nights. 



A glance at the curves for McMurdo Sound in figure 20, page 54, -vvill show that this 

 excess of temperature at i a.m., which is present throughout the winter months, becomes more 

 marked at other periods. While it is feebly marked during May and June it becomes so 

 large in July and August that it then forms the main maximum temperature of the day. 

 It is still well marked in September, but now the returning sun has produced a larger maxi- 

 mum later in the day. From October to March the 4 a.m. temperature shows no maximum, 

 but it is noteworthy that of these months October, November and March show a similar 

 effect at 2 a.m. ; may not this be the same effect but displaced by the minimum due to the 

 solar radiation falling very near to 4 a.m. ? After March the effect 

 reappears and is very markbd in April. 



We now see that in McMurdo Sound there is an excess of tempe- 

 rature at 4 A.M. compared with 2 a.m. and 6 a.m. in every month 

 from April to September inclnsive, and the same effect is possibly 

 present in October, November and March although during these 

 months the minimum of the solar temperature amplitude is a disturb- 

 ing factor. 



The same effect is clearly seen in the temperature observations 

 at the Gauss station although data for one year only are available. 



The Gauss wintered just outside the Antarctic Circle and therefore 

 direct solar insolation was never entirely absent ; yet during four 

 months, July to October, the maximum at 4 a.m. is clearly distin- 

 guishable. As the curves given in Plate V of Meinardus's work have 

 been smoothed the effect we are discussing is not clearly shown. I 

 have therefore plotted the actual data from midnight to 8 A.M., and 

 reproduce them here in figure 25. In July and August the maximum 

 is clearly visible as an actual peak on the curves. It is visible in 

 September and October as an interruption in the smooth course of 

 the curve at 4 a.m. 



The Snow Hill observations also show the same effect in July and August as will be 

 seen from the following table : — - 



Table 29. 



Departures from Mean Temperature at Snow Hill. 



Fio. 25. Daily variation 

 of temperature. Gams 

 station. Midnight to 

 8 A..M. 



