200 THE CLIMATIC FACTOB AS ILLUSTRATED IN ARID AMERICA. 



its eastern member has moved one square west of its position in February. The western 

 decrease, however, and the Canadian increase have both disappeared ahiiost entirely. 

 This speaks strongly against the idea that the differences between the old and new series 

 of maps are due to differences in the observational material on which they are based. 

 If such were the case it would scarcely be possible that in one particular month there shou d 

 be a sudden reversal of previous conditions, so that the number of observations should 

 suddenly show a relative increase in the western United States in latitude 40° to 45° and a 

 large decrease in Canada 10° farther north. The April map for both periods (figure 55) 

 shows a southwestern maximum; in the later period, however, the center lies farther south 

 and west than in the earher. In this map we again note the decrease north of the south- 

 western increase, and the western Canada increase. Each period shows only a single 

 eastern maximum, that of the later map lying well to the west of the other. 



May (figure 57) shows the southwestern increase and a sHght decrease in the Oregon- 

 Idaho section. Here again we notice only a slight increase in western Canada, which 

 gives strength to the reality of the large increases shown in other months. Also the double 

 maximum has almost disappeared, although traces of the western center may be seen in 

 the 27 tracks which crossed the square between 45° and 50° N. and 100° and 105° W. In 

 this case the earlier map also showed a double maximum indicated by the figures 28 in 

 the square between 90° and 95° W. and 34 between 75° and 80° W. June (figure 57) 

 shows marked western Canadian increase, a moderate southwestern increase, a double 

 maximum, a slight western increase, and a distinct southward shifting. July (figure 58) 

 shows strong western Canadian increase, double maximum and southwestern increase. 

 August (figure 59) shows western Canadian increase and a westward shift of the maximum 

 in latitude 45° to 50°. September (figure 60) shows general westward shift, double maxi- 

 mum, and western Canadian increase. October (figure 61) shows the southwestern 

 increase, the decrease just north of it, a pronounced increase over western Canada, and a 

 double maximum including within it a double maximum of the earlier period. The agree- 

 ment of the pecuhar loop in the southeastern part of the 1 line in both maps is striking. 

 November (figure 62) presents the highest increase over western Canada, a double maxi- 

 mum as compared with the single maximum of the earlier period. The same holds also for 

 December (figure 63). The Oregon-Idaho decrease is here marked. There is also a 

 general southerly shift. 



Combining these month maps we have a year map (figure 64), showing general agree- 

 ment with that of 21 years before, but characterized by the phenomena alreadj^ noticed 

 in the month maps — a marked increase over western Canada, a slight decrease in the two 

 western squares of latitude 40° to 45°, and a general southerly and westerly shift of the 

 lines of equal storm frequency. In latitude 45° to 50°, along the storm track proper, a single 

 maximum of 484 in the earlier period is replaced by a double maximum, 480 and 491, in 

 the later period. If we take the central area, which may be supposed to have had 400 

 storms in each case, its center would seem to fall about 2.5° farther west in the later map 

 than in the earlier. Little rehance, however, can be laid on this. Corresponding to the 

 general westerly shift, an eastern decrease is noted, but I lay less weight on this feature, 

 since I am convinced that the tracks as given in the monthly summaries are not continued 

 over the ocean in all cases as far as might be the fact. 



Turning now to the discussion of the meaning of the maps, let us consider first the 

 phenomenon of a double maximum. If the average track of all storms be plotted it forms 

 a curve corresponding to the line of greatest frequency. If such a curved line merelj^ 

 touches a given parallel of latitude, say that of 47° N., there will be a single ma.ximum of 

 storm frequency along that line. If, however, the curved storm track is shifted enough 

 so that its most southerly point Hes a few degrees south of latitude 47°, say in latitude 45°, 

 the storm track itself will cut the given parallel in two places. Thus along that latitude 



