66 TEMPERATUKE. 



position of the point for May ; for one would have expected this to be abont half way 

 between the points for April and June, as was the case at the Gauss station. Here I think 

 Mohn's explanation has weight. Compared with April, May was a cloudy and windy month 

 and these together would have reduced the temperature amplitude if there had been no 

 change of surface. 



Therefore the effect of the cloud and wind must be added to the change of surface, 

 thus accounting lor the low position of May relatively to April, but it is interesting to notice 

 that in spite of the cloud and wind the point for May is higher than the mean position of 

 the other points, showing that the surface had not got into the normal state for sea ice. 

 The low position of the Fra7)i point for July is in all probability due to the explanation 

 given by Meinardus ; for the mean temperature of this month was the freezing point, and 

 the daily amplitude of temperature could not be great above a wet snow surface. 



From this discussion we see that the variation of the temperature amplitude in the Fram 

 type is mainly accounted for by changes in the surface and solar radiation, these are the 

 chief factors, small variations may be due to other causes such as' cloud, wind or local conditions. 



We must now examine why the McMurdo type diilers from that of the Fram type. 

 The explanation is that owing to the local conditions there was no large change at any of 

 the stations of this type in the surface conditions between the spring and summer. At 

 Framheim the surface was that of the Barrier which can undergo no change. At McMurdo 

 Sound and Snow Hill the observations were on the coast between a snow-covered land and 

 a frozen sea. In McMurdo Sound there was little change of the snow-covering of the land 

 throughout the year and the sea ice was almost swept clear of snow by the blizzards during 

 the winter. In fact there was more snow on the sea ice in January than in October, and the 

 effect is seen by the relatively high position of the point for the former month in the 

 diagram. Similar conditions probably held at Snow Hill which was subject to very high dry 

 winds which must have prevented any large accumulation of snow on the sea ice. 



Summary and Conclusions : Temperature Variation in the Period of Solar Activity, September 

 to March. — (a) The daily temperature amplitude during this period over sea ice nearly free from 

 snow is a linear function of the daily amplitude of solar energy. 



{b) In consequence the temperature amplitude increases from the return of the sun to 

 midsummer and then decreases, except that the warmest month has generally a lower ampli- 

 tude due to the inability of the surface to warm up above the freezing point. This gives 

 rise to the McMurdo type of daily temperature variation. 



(c) A layer of loose snow increases the temperature amplitude. 



(d) In consequence, in situations where there is a layer of loose snow covering the sea 

 ice in the early spring months the temperature amplitude is relatively high. As the solar 

 energy increases the snow-covering becomes more dense and thinner, in consequence the 

 temperature amplitude generally decreases from spring to midsummer in spite of the increasing 

 amplitude of solar energy. This gives rise to the Fram tyj^e of daily temperature variation. 



(e) In places where the surface consists of a very deep layer of snow — for example the 

 Barrier — the temperature amplitude is very large. 



Daily Variation of Temperature during May, June and July. Period with No Direct Solar 



Radiation. 



During May, June and July the sun does not rise above the horizon in McMiudo Sound 

 and the amount of indirect radiation from the sky is so small that it may be entirely 

 neglected. 



