NO. 4 TEMPERATURE VARIATIONS IN THE NORTH ATLANTIC 43 



apparently disappears. That is, hoXvever, what one must expect. In 

 the eastern regions the relations of the individual fields are so simi- 

 lar that even a few observations in a small field are sufficient to fur- 

 nish a fairly good mean value, while for the more variable region 

 further westward a much more extensive observational material is 

 necessary, and for this purpose, as Ave have seen, i° fields are not 

 adequate. 



If one forms the mean of the temperature anomalies, taking Peter- 

 sen's stations two by two which lie within our io° longitude fields 

 (between io° and 20° west longitude, etc.) it would be expected that 

 more trustworthy values in comparison with ours would be obtained. 

 In figure 13 the full curves show results found in this way for Peter- 

 sen's stations for February and March from 1898 to 1902, and the 

 dotted lines give the corresponding curves for our 10° longitude 

 fields. The agreement is better, especially in the eastern fields, than 

 we had expected. 



In comparing the Petersen curves for March with ours for the 

 last decade group, one should not forget that these last extend from 

 March 15 to i\pril 13 and the times for the curves do not fall 

 together, which in part explains their discrepancies. But it does not 

 explain the extraordinary deviation between the curves for the field 

 50° to 59° west longitude. In this field our curves for the first and 

 last decade groups for the years 1898 to 1902 fall almost exactly 

 together and we seem justified in supposing therefore that during 

 the time intervening the same relations must hold in this region. 

 There is therefore no place for the great disagreement which Peter- 

 sen's curves show, and we must conclude that these are not repre- 

 sentative. 



The principal result of Petersen's investigations is that the yearly 

 variations in the surface temperatures of the Atlantic Ocean depend 

 on the air pressure distribution, which controls the winds. He has, 

 however, made no attempt to reduce this relation to a quantitative 

 basis. 



He finds that the changes of position of the Icelandic air pressure 

 minimum are of great importance for the variations of the surface 

 temperatures in the Atlantic Ocean. He says " the non-periodic 

 changes of the position of the Icelandic depression cause correspond- 

 ing variations in the direction of the wind, which, after one or two 

 months interval, express themselves by the increase or diminution of 

 the ocean temperatures. Thus, for example, a very westerly posi- 

 tion of the depression calls forth, by means of the wind, high tern- 



