68 



TEMPERATURE. 



the means for May, June and July during the four years, have already been seen on figure 

 20, page 54. The means for each year taken separately are shown as curves 2 to 5 on 

 figure 24. The results of the numerical tests for (a) and (h) arc entered in the following 

 table, where a plus sign indicates that the feature is present and a negative sign that it is 

 absent : — 



Table 28. 



The Reality of the Feature: Excens of Temperature During the Hours 9 «.«(. to 7 ^j.h(. during 

 May, June and July. — From column 2 of table 28 we see that taking all the observations 

 together the mean excess of temperature during these hours is positive. It is also positive 

 in each of the years 1902, 1903 and 1911, each considered separately; it is absent, however, 

 ill 1912. When the months are taken separately the feature is seen to be present in May 

 and June but not in July. Thus out of the seven separate sub-divisions of the data five 

 show the effect while only two do not. This must be taken as very good evidence of 

 the reality of the effect. We can therefore conclude that in McMurdo Sound during the 

 three winter months when there is no direct solar radiation the day is warmer than the 

 night. In other words in spite of the absence of a var}'ing solar radiation the variations of 

 temperature are in the same direction, but much less in intensity, than when the sun 

 shines. 



