47 



tic ally disappear, if sea ice did not hamper the North American 

 coast line from February to March every year. The pressure dif- 

 ference between Bergen and Stykkisolm during the period October 

 to January was also found to be of importance. 



The use of pressure difference between various points furnishes 

 the best data for forecasting purposes, because there is no room for 

 the personal bias which may come in when charts are classified 

 according to types. A classification of the charts of pressure anom- 

 aly over the North Atlantic during the period December to March 

 has, however, been made, and this distinctly reveals two types of 

 pressure distribution— a plus type, in which an excess of pressure 



Fig. 8b.— The atmospheric pressure map constructed by averaging the pressures for the months Decem- 

 ber to March in the years 1885, 1890, 1903, 1912, and 1921. These years are characterized by a greater 

 amount of Arctic ice drifting into the western North Atlantic than usual. (See fig. 23, p. 76.) 



centered in the region of Iceland, more or less dominates the Atlantic 

 north of the Azores (see fig. 8a, p. 46), and a minus type when reverse 

 conditions prevail (see fig. 8b, p. 47). The plus type is subject to 

 further classification into (1) and (2), depending upon a relatively 

 great or moderate intensity of the excess pressure mass, both of which 

 are reflected in a relatively very light, or light ice year, respectively, in 

 the western North Atlantic. The minus type, although unmis- 

 takably showing a greater .amount of ice than normal, does not 

 permit subgrouping. In other words, the plus type of pressure 

 conditions (fig. 8a) exhibit a higher correlation with poor ice years 

 than do the minus type (fig. 8b) with correspondingly rich ice years. 

 This indicates the presence of other factors, such as variations in 



