DAILY VARIATION— MAY, JUNE AND JULY. 71 



The temperature is above the mean of the day from 9 p.m. to 8 a.m. and below during 

 the rest of the day. No excess of temperature is shown at 4 a.m. and the day is colder 

 than the night ; thus neither of our effects is present. Mohn concludes that the daily variation 

 is fully accounted for by the daily variation of the wind : ' In the dark season the north 

 component of the wind's frequency exceeds the south component uninterruptedly from 7 a.m. 

 to 8 P.M. and the northerly winds are colder than the southerly winds. It seems reasonable to 

 ascribe the diurnal period of the temperature found in the dark season to the effect of the wind.' 



But Mohn has gone further ; he has determined the daily march of temperature for the 

 dark jnonths (a) during cloudless days ; (b) during overca.st da3's ; (c) during days with an 

 average wind velocity less than 4-5 metres per second ; and ((/) during days with an average wind 

 velocity greater than 4-5 metres per second. The results for each of these for the three winter 

 months, November, December and January, are shown as curves 2, 3, 4 and 5 of figure 26. It 

 will at once be seen that the temperature variations on cloudless and on windless days are 

 almost mirror images of the variations on overcast and windy days. The warm period found in 

 the Antarctic between 9 a.m. and 7 p.m. is certainly present in the north on overca.st days, 

 and the excess of temperature at 4 a.m. is very marked in the north on days with little wind. 

 This leads us to investigate the effect of cloud and temperature on the Antarctic temperature 

 variations. The variation was therefore calculated for cloudless day.-' and overcast days, but 

 it was found that the number of days was too small to give smooth curves, therefore all 

 days were considered in which the cloud amount was less than 5 at all ob.servations, and all 

 days in which the cloud amount was more than 5 at all observations. These curves were 

 still irregular but smoother than (he former, they were howevei the same in general run. 

 They have been plotted as curves 6 and 7 of figure 24, page 67. The days of May, June 

 and July of the four years were then divided according to wind by first taking out those 

 days in which the hourly wind velocity did not rise to 20 miles an hour in any one hour 

 throughout the day, these formed the first class ; the remaining days on which the wind rose 

 above 20 miles per hour during at least one hour in the day formed the second class. The 

 temperature variations for each class were determined and plotted as curves 8 and 9 of the 

 same figure. 



Examining the two latter curves we see that they are so similar to one another and 

 to the mean of all days, curve 1, that wind can have little direct influence in McMurdo 

 Sound on either of the features under discussion. 



The influence of cloud, however, although small, is quite marked. It will be seen that 

 curves 6 and 7 in their general run are similar to those found for the north ; days with 

 cloudy skies are warmer during the day time than during the night, while days with clear 

 skies are on the whole warmer during the night hours than during the day hours. But a 

 still more important relationship is shown. The excess of temperature at 4 a.m. is clearlv 

 marked with clear skies, 4 a.m. being the warmest hour of the day ; while there is no 

 indication of the excess at 4 a.m. on cloudy days. 



It therefore appears that our first effect, the day warmer than the night, is associated 

 with cloudy days, while the second effect, excess of temperature at 4 a.m., is associated with 

 cloudless or slightly clouded days. Except that the ma.ximum is at 2 a.m. instead of 4 a.m. 

 in the north the same relationship holds there. 



Mohn has given an explanation of the average conditions in the north, but he gives no 

 suggestions for the remarkable difference in daily variation according to cloud and wind.* 



* It should be. remarked that in the north cloud is so intimately related to wmd, windy daj's being nearly 

 always overcast, that it is impossible to separate the two effects. The curves for no cloud and little wind on 

 figure 26 are practically from the same data; similarly with the curves for overcast skies and windy days. 



