NO. 4 TEMPERATURE VARIATIONS IN THE NORTH ATLANTIC 247 



for Liepe's stations III, IV, and V sho'wn in figures 56 and 59. It 

 can scarcely be doubted that the variations in the air pressure gradi- 

 ents, that is to say, in the intensity of the trade wind, is here an 

 important cause of the temperature fluctuations within the observed 

 fields which occur in intervals of a short number of years. It is 

 very doubtful that, however, if this is true of those variations of 

 longer periods of years. 



We see in figure 91 that the curves VIII and IX for the earlier 

 time up to 1892 lie close together. They then gradually separate 

 from one another and afterwards approach again in the year 1902. 

 The temperature at station VI and also at other of Liepe's stations 

 was considerably higher in the time interval 1893 to 1902 than would 

 be expected by reference to the curve of air pressure gradients. 



This relation is yet more clearly shown in figure 85, where we have 

 reduced the curves by two- and three-years smoothing. Curve B 

 shows the air pressure gradients in the trade wind region and curves 

 I to VIII the temperature at Liepe's stations. We see that here 

 curve B goes very well with the temperature curve of Liepe's sta- 

 tions III to VI for the first part of the time up to 1892, but that after 

 this time there is little agreement between the air pressure curve 

 and the temperature curves. The combined two- and three-years' 

 smoothing has eliminated the shorter fluctuations which in figure 91 

 and figure 56 show such great agreement. We must therefore 

 assume that in this region other factors began to come into play 

 after 1892. It might be that surface currents from the Canary 

 Islands bringing down water from the north had changed the 

 temperature. Liepe has remarked that temperature variations may 

 occur in this manner. The temperatures at Liepe's station I and. 

 partly also at Liepe's station II were particularly high in the years 

 1893 to 1900. It might be thought that warmer water thus carried 

 southwards would tend to hinder the depression of the tempera- 

 tures corresponding to the winds. 



Returning now again to figure 91, and comparing the curve VIII 

 for the already mentioned air pressure gradients with the curve II 

 for the difference in air pressure between Colombo and Hyderabad, 

 we see that the variations in these two curves at corresponding times 

 have a tendency to go oppositely. For instance, a small air pressure 

 difiference in India in the year 1897 coincides with a relatively great 

 air pressure difiference in the northeasterly trade wind. At other 

 times they run in the same direction, as, for instance, when a great 



