130 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. "JO 



that the variations in the surface temperature at Ona, and the other 

 Norwegian lighthouse stations in February are in closer agree- 

 ment with the variations of the wind conditions than in January. 



IX. THE PERIODICITY OF THE VARIATIONS OF THE SURFACE 



TEMPERATURE OF THE ATLANTIC OCEAN AND OF 



THE AIR TEMPERATURE OF THE CONTINENTS 



If we now go on to the investigation of the possible causes of 

 these variations, it is obviously necessary to investigate whether 

 they are entirely aperiodic or are in some way arranged in deter- 

 mined periods which can be recognized. 



Unfortunately the series of observational material which was 

 available to us for the Atlantic Ocean is too short in order to study 

 the periods by the general methods of harmonic analysis. How- 

 ever, we have endeavored to get an approximate analysis out of 

 the mean temperatures which we have found for the whole investi- 

 gated path across the North Atlantic Ocean for the series of years 

 1898 to 1910. 



The table following figure 29 shows the observed anomalies of 

 mean surface temperatures in all of the regions investigated by us 

 of the North Atlantic Ocean between America and Europe. We 

 have not employed the Danish fields north of 50° north latitude 

 We have compared these values in different ways corresponding to 

 the periods we wish to eliminate. The elimination was performed in 



the ordinary manner according to the formula: A''= ^ • •• -"t" n^ 



We have eliminated first a two-year period, then a three-year period 

 and finally a five-year period. 



The results are given graphically in figure 54. The curve a shows 

 the orginally found anomalies of the mean temperature. The curves 

 h, c, and d show the values after elimination of the two- three- and 

 five-year periods. In the two latter curves {c and d) we give the 

 results after the elimination of the short periods by full-drawn lines. 

 The dotted line c shows the result of the three years' eliminatio'n as 

 obtained directly from the observed values shown in curve a with- 

 out regard to the two years' elimination. In the same way is shown 

 by the dotted line d the result for the five years' comparison directly 

 from the original values. The curve d has a very regular march. 

 The dotted curve extends over eight years. The residual values 

 which this curve yield may indicate periods of longer duration. One 

 naturally thinks first of the eleven-year sun spot period and the 



