NO. 4 TEMPERATURE N'ARIATIOXS IX THE NORTH ATLANTIC I39 



and VI show better agreement with the magnetic curve M. On 

 the other hand, as ah-eady remarked the minima of curve III fell 

 in the years 1884 and 1894 a year after the sun spot maxima and 

 the maximum in the year 1890, a year after the sun spot minimum. 

 These yearly temperature curves I, II, IV, and VI show also un- 

 doubted similarity to curve f of figure 54. There appears, however, 

 to be some phase displacement, but this may be due to some special 

 accidental causes. 



The curves of the eastern fields of the Atlantic Ocean show at 

 least in part, as already remarked, a direct similarity with the sun 

 spot curve. We have computed the mean of the temperature 

 value for Liepe's stations I, II and III and have carried through 

 a three years' smoothing of the observations. The temperatures 



Figure 58. Three-year smoothed curves of surface temperature. I : for 

 Liepe's stations I to III. II : for the most easterly Danish field from 0° to 9" 

 west longitude 58° to 59° north latitude. Ill for Liepe's stations V and VI. 

 IV : the two most northerly fields from the Central Bureau at 30° and 36° 

 north latitude, see also figure 55, curves V and VI. V : for the air temperature 

 in southwest Siberia. S : curve of the vertical sunspot numbers with the 

 scale on ihe left. 



for his stations V and VI, have been treated in the same way. 

 The results given in curves I and III of figure 58, curves II and lY 

 representing the temperatures of the Danish fields 0° to 9° west 

 longitude 50° to 59° north latitude are similarly obtained, and the 

 two fields of the Central Bureau, at 36° north latitude and 20° north 

 latitude (see fig. 55, curves V and VI) and finally curve V for the 

 air temperature in southwest Siberia are also given. Above is the 

 curve S for the sun spot numbers. 



During the sun spot period 1889 to 1901, the smoothed tempera- 

 ture curves I to IV show quite good agreement with the sun spot 

 curves, except that the temperature curves I and III show four 

 shorter periods within this long period. In the next sun spot period, 

 after 1901, curves II and IV have a tendency to go opposite to the 



