486 



SCIENCE PROGRESS 



and shows this relationship. The line I. shows the variations 

 in the amount of heat contained in the water beneath the 

 Sognefjord section: it represents the area of water of salinity 

 of 35 per mille. and over, multiplied by the mean temperature. 

 The line II. represents the mean temperature-anomaly in 

 Norway. The mean air-temperature has been determined from 

 the records of 22 meteorological stations distributed over the 

 whole of the country and is based on thirty years' records. 

 The annual deviations from this mean are then calculated for 

 the same 22 stations. The water-temperature values are those 

 obtained during May, and the air-temperatures are for the 

 period November to April of the following winter. We see 



1900 1901 1902 1903 \go4- \qos 



i2oo.- 



II0O-- 



looo-l- ??<**. o° 



^/^srll ^<^-- apr, . l - 



Fig. 4. — Air and sea-water temperatures at Norway. 



Curve I. represents the heat-contents in the Atlantic water beneath the Sognefjord section in May. 



„ II. represents the average temperature-anomalies in Norway in the following November-May period. 

 (From Helland-Hansen and Nansen, loc. cit. p. 345.) 



at once that a maximum Gulf Stream flow in May is followed 

 by a maximum air-temperature in the following winter. 



The temperature of the air over the land is therefore a 

 function of the amount of heat contained in the water of the 

 adjacent sea-area. It would not do to attempt to compare 

 the mean surface temperature of the sea with the land- 

 temperature. In the winter half of the year the surface waters 

 of the sea are warmer than the overlying atmosphere, and heat 

 is radiated from the sea to the air. If this liberation of heat is 

 rapid, ascending air-currents are produced, with accompanying 

 cyclonic atmospheric disturbances. Obviously the latter will 

 affect the adjacent land-areas, but how exactly will depend 

 on a variety of conditions. When the surface waters of the 



