202 



THE CLIMATIC FACTOR AS ILLUSTRATED IN ARID AMERICA. 



of greatest storm frequency, we should clearly get a single maximum. If, however, the 

 mean track were to swing somewhat farther south, so that at the extreme southern limit 

 (where, as before, storms were most frequent) it touched latitude 46°, it would cut the 

 parallel of 47.5° in two places. As 46° is not the median hne of any set of squares, we 

 should not have any maximum corresponding to it, but we should have a double maximum 

 in the tier of squares having 47.5° as their median line. 



Let us turn now to the question of how far the apparent increase in storm frequency 

 during the later as compared with the earlier of our two periods is genuine. The total 

 number of observations of storms, as obtained by adding all the numbers on each of the 

 two year-maps, shows that the total frequency for the later period is 17.4 per cent more than 

 for the earlier. At first it might be thought that this is due merely to a difference either 



1^ 85° 80° 75° 70° 65° 60° Z% 10% IS^ 20% 25% 30 ^I Z5° I20'' 115° 110° 105° 100° 95° 90° 65° 80° 75° 70° 65' 



65 



150 



66 



zoo 250 



67 



Lines drawn around the 



Fig. 65.— Changes in Storm Frequency by Months aci-ordiug to Longitude. 



Numerals indicate amount of increase or decrease in storm frequency in the given longitude for each month. ^ marked " maxi- 



areas show an increase of oyer 40. The line marked " zero mdicates no change m frequency. The dotted Une marked maxi 

 mum " indicates maximum increase. 



Fig. 66.— Changes in Storm Frequency by Months Irrespective of Longitude and Latitude. 



= Absolute increase in storm frequency by months. 



= Per cent of increase. Average per cent equals 17.4. ^ ^ 



Pio. 67.— Changes in Storm Frequency by Months according to Longitude in Latitude 50 to 55 . 



in the observational material or in the way in which it has been treated. Closer inspection, 

 however, seems to show that this is not the case. If it were the case, we should expect to 

 find that the increase indicated by the later figures is distributed somewhat umformly over 

 the whole country, and in each of the different months. In order to test this I have plotted 

 the total increases in storm frequency for each month of the year in the solid line of figure 

 66 and the percentage of increase by a broken line. It is evident that the increase m storm 

 frequency has varied from 7.5 per cent in some months to 30 per cent in others. Moreover, 

 the curve shows a seasonal variation with two periods of minimum increase in the late 

 winter and late summer, and two of maximum increase centering approximately m May 

 and December. In addition to this, I have plotted the increase and decrease for each 

 month of the year in each north and south column of 5° square, as is shown in figure 65. 

 Here it appears that the lines of no increase and of maximum increase both shift eastward 

 or westward according to the seasons, being farthest to the west toward the end of the 

 winter and early in the fall, at about the time when the greatest increase in storm frequency 



