THE SHIFTING OF CLIMATIC ZONES. 



203 



takes place. If the increase in storm frequency for a single tier of squares (that is, for 

 those between two parallels of latitude 5° apart) is plotted, a similar phenomenon is ob- 

 served in the case shown in figure 67, where the increase or decrease in latitude 50° to 55° 

 has been plotted. Here again the curves are roughly parallel to those of figure 66, a 

 phenomenon whose significance is not apparent, but which seems worthy of study. In 

 latitude 45° to 50°, however, as appears in figure 68, the increase or decrease in storm 

 frequency assumes a character entirely different from that farther north. The diagram 

 shows a peculiar curved area of decrease and a diagonal distribution of increase with a 

 maximum in the east centering during the month of March. The diagrams of this same 

 phenomenon for the other tiers of latitude show an equally variant distribution, as may be 

 seen in part in figures 69 and 70. 



125 120 



Fig. 6S. — Changes in Storm Frequency by Months according to Longitude in Latitude 45° to 50°. 



Finally, the last diagram, figure 71, presents a summary of the absolute differences in 

 the number of recorded passages of storms during the earlier and later periods for each 

 square of the map. The appearance of the map varies Uttle, whether absolute differences 

 are used, as in figure 71, or percentage differences. This map offers an interesting answer 

 to the question whether the differences between the maps (as compiled by Dunwoody in 

 the period from 1878 to 1887 and by myself for the period from 1899 to 1908) are due 

 merely to differences in method and in observational material or to actual differences in 

 the number of storms. On the map I have indicated the barometric stations in western 

 Canada which were used for the earlier period: Fort Rae, Edmonton, Calgary, Medicine 

 Hat, and Qu'Appelle. It will be seen that in the square where the greatest increase in storm 

 frequency is noted there were three stations during the earlier period, which would seem to 

 cover that region sufficiently. This renders strongly probable the conclusion that the 

 increases and decreases shown in the various parts of the map are real — in great part at 

 least. It is also noticeable that the eastern area of decreased frequency hes directly north 

 of the area where the number of storms is greatest, which is what would be expected if 

 the ideal storm track (that is, the mean of all tracks for a given period) had been shoved 

 southward and westward. Concordant with this is the marked increase in frequency in 

 the southwest and south. Taken as a whole it appears as if we had an area of increased 



