NO. 4 TEMPERATURE VARIATIONS IN THE NORTH ATLANTIC I29 



surface temperature. The variations show a good agreement with 

 one another. From the zigzag curve A one sees that the agree- 

 ment descends even to the smallest peculiarities. For instance, see 

 how quickly the variations follow one another in 1889, 1894 to 

 1895, 1899, and so on, with almost complete parallelism. There is 

 to be sure a small displacement between the curves oftentimes, so 

 that strong maxima or minima show a tendency to appear earlier 

 in the air in Stockholm than in the Norwegian coast water. This 

 may be estimated as some days or even a couple of weeks, and shows 

 itself very frequently in the monthly means we have employed. The 

 reverse and earlier coming of the extreme in the water than in the 

 air is found only exceptionally. In the curves B and C, which 

 we compare, one finds a well-marked parallelism, with a similar 

 tendency to displacement to that shown also in the A curves. In 

 the C curves, where periodic variations of two years or rational 

 parts of it are eliminated, and principally only the longer periods 

 can come to observation, the displacement is, however, shown quite 

 distinctly. The maxima and minima fall in most cases earlier in 

 the curve for Stockholm than in that for the Norwegian light- 

 house stations. 



From the nearly simultaneous occurrence and well-marked agree- 

 ment of the features of these curves, it follows with great cer- 

 tainty that no causal relation exists between the variation of the 

 surface temperature on the Norwegian coast and the variations of 

 the air temperature in Scandinavia, but rather that both variations 

 must be due to the same cause, although the action takes place a 

 little earlier in the air than in the coast water. 



The direct common cause of short interval variations is, accord- 

 ing to our view, doubtless the variation in the air pressure distribu- 

 tion, which is rendered very probable by the above described investi- 

 gations on the relation between the air pressure distribution and 

 the surface temperature at Ona and Torungen. We shall later speak 

 of a yet more elegant proof of the accuracy of this assumption. 



Accordingly it is clear that the surface temperature on the Nor- 

 wegian coast cannot be used as a measure of the temperature varia- 

 tions of the water-masses of the warm Atlantic ocean currents in 

 the North Sea, as has been done by Pettersson and Meinardus. In 

 this connection it is interesting to remark that Pettersson found the 

 best agreement between the surface temperature on the Norwegian 

 coast and the air temperature in Sweden in February, and not so 

 good in January. This corresponds exactly to what we have found, 



