NO. 4 TEMPERATURE VARIATIONS IN THE NORTH ATLANTIC 5 



the observations for the two three-decade intervals. The observa- 

 tional material for these fields is on the whole very full, though 

 the number of observations in each decade group in the weakest 

 cases is less than 20. But in general and for the later years after 

 1904, the numbers exceed 40 or even 50 in the decade groups. 



In the southern region (between 10° and 40° west longitude 

 and 37° and 45° north latitude) the observations, as the plates i 

 to 14 show, are so scattered that the mean temperatures which are 

 determined for two-degree fields are very untrustworthy. For this 

 region, in which the local temperature differences are comparatively 

 small, we have therefore reduced the observations in larger fields 

 of 2° latitude and 10° longitude. These 10° longitude fields are 

 indicated in figure i and also in plate 15 by cross hatching. 



With the help of the mean values of the temperature for the 

 decades and decade groups of each year, we have computed the 

 normal temperatures of the surface and of the air for each of the 

 chosen fields. There were in all sixty fields, forty-eight northerly 

 2° fields, and twelve southerly 10° fields. In this computation we 

 used only the values for the eleven years from 1900 to 19 10 inclu- 

 sive, because the observational material for the two first years, 1898 

 and 1899, is not satisfactory. Finally the anomalies for the single 

 decade and decade groups for each year were computed. These 

 anomalies may be found in tables iW, 3W, 6L and 8L, where also 

 the normal temperatures for the water and the air are given. 



Concerning the accuracy of the temperature observations in our 

 material one must admit that this is only moderate. This remark 

 holds for the temperatures of the ocean surface but more particu- 

 larly for .those of the air. The readings, including those of the 

 water temperatures, are often given in whole degrees, frequently 

 in half degrees, and even the accuracy of the numbers themselves 

 is often doubtful. At single stations, the temperatures given are 

 sometimes impossible, as for example, water temperatures of 

 -3° C. or even -4° C. ! An explanation of these errors is hard to 

 give. It appears as if at many stations the surface and air temper- 

 atures were interchanged. We have cast out the obviously false 

 observations. In tables iW, 3W, 6L and 8L, the computed mean 

 values in such cases are indicated by bold-faced type. 



In single cases where observations have been wholly lacking or 

 where the computed mean value on account of too small a number 

 of observations seemed improbable, we have introduced a new value 

 by interpolation. In forming this value, the temperature relation 



