CLIMATIC CONDITIONS OF THE UNITED STATES. 195 



Most of the data of table 2 have been plotted on a map of the United 

 States, to represent them in a graphic way. After the numerous 

 stations had been located and the average length of their respective 

 frostless seasons had been placed beside the points representing them 

 on the map, lines were traced as accurately as possible through points 

 having a common length of frostless season, a line for each increment of 

 20 days, beginning with 80. By this means the map was subdivided into 

 15 sorts of seasonal areas. The first sort has, according to our data, an 

 average frostless season of less than 80 days, the second has a season of 

 80 to 100 days, and so on by steps of 20 days until the average frostless 

 season for the fourteenth sort of areas is 320 to 340 days, and for the 

 fifteenth, over 340 days. The only one of our stations that is unques- 

 tionably without frost, thus having an average frostless season of 365 

 days, is Key West, Florida. 



In tracing the equiseasonal lines we have followed the data of table 

 2 as accurately as possible, making no attempt to smooth the lines. 

 Topography has been allowed to exert a deciding influence in many 

 cases where observational stations are too far apart satisfactorily to 

 determine the positions of the lines. This is especially the case for that 

 portion of the chart which lies west of the one hundred and fifth 

 meridian of west longitude. In a very few cases the data of single 

 stations have been ignored, where the length of season given in the 

 table is obviously a marked exception for its region, thus suggesting 

 the possibility of error or inadequacy in the data themselves. Wherever 

 a small local area is indicated by the data from two or more stations, 

 however, the area has been shown on the chart. 



It was found at once that the approximate equiseasonal lines for 

 20-day increments were altogether too crowded in the mountainous 

 region of the West, when drawn upon a chart of any convenient size, 

 and for this region all lines have been discarded in this region, excepting 

 those for 80, 120, 180, 240, and 300 days. This method probably 

 presents the details for this part of the country as accurately as is to 

 be expected from the data now at hand. The omitted lines, all drawn 

 for the East, have been abruptly terminated wherever they would 

 enter the more generalized portion of our chart. Plate 34 is a repre- 

 sentation of the chart just described. For the sake of clearness, the 

 chart has been shaded so as to fall primarily into the five classes of areas. 

 These areas denote regions with average lengths, in days, of frostless 

 season of (1) less than 120, (2) 120 to 180, (3) 180 to 240, (4) 240 to 

 300, and (5) over 300. The other lines, representing greater detail, 

 are shown in dotted form. 



A comparison of plate 34 with Day's plate V (loc. cit.) shows, at first 

 glance, a remarkable series of differences. Closer scrutiny brings it 

 out, however, that the apparent discrepancies are mainly due to the 

 fact that the lines of Day's chart have been obviously subjected to a 



