CORRELATION OF DISTRIBUTIONAL FEATURES. 535 



that the fringe completely surrounds the region of greatest abundance 

 makes this a case in which it is of interest to compare the character of 

 the two sets of ampUtudes. The amplitudes of conditions for the center 

 are narrow in all cases, except the number of cold days. The ampli- 

 tudes for the fringe are wide in a number of cases, notably the number 

 of cold days, the annual daily mean temperature, the length of the 

 longest rainy period, and the moisture-temperature index. In every 

 case the amplitude of the conditions for the center is less than that for 

 the fringe, and the extremes for the center lie within those for the 

 fringe in all cases except those of evaporation, humidity, and the 

 moisture ratios. The geometrical centers of the blocks representing 

 the extremes for the region of greatest abundance He within the blocks 

 for the extremes of the fringe in every case except that of evaporation. 

 This means that for all of the conditions except evaporation it is possi- 

 ble to find a locality in the fringe which possesses climatic values that 

 are near those of the absolute ecological optimum of the species. 



The fact that a straight line laid down across any portion of the map 

 of the United States passes through localities showing for long dis- 

 tances a progressive change in the values of each chmatic condition 

 is responsible for the maximum and minimum values of so many con- 

 ditions being respectively greater and less in the fringe than in the 

 center. The conditions which are favorable to Liriodendron in the 

 region of greatest abundance are, in some cases, still more favorable 

 to it in the northern part of the fringe, and in other cases still more 

 favorable in the southern part of the fringe. The center exhibits the 

 favorable constellation of conditions designated as the ecological 

 optimum. When a given condition shows minimum and maximum 

 values for the fringe which are not respectively lower and higher than 

 the extremes for the center (fig. 44, plates 53, 57, 59, 60, and 65), it is 

 probably an indication that the condition involved is not an important 

 one in determining the location of the center and fringe, however 

 narrow the amphtudes involved may be. 



The comparative uniformity in the amplitudes of all temperature 

 conditions for the distributional center of Liriodendron indicates that 

 these conditions are of nearly equal weight in determining the limits 

 of the center, with a slight indication of preponderant importance for 

 the mean temperature of the coldest 14 days. The daily mean pre- 

 cipitation and the number of days in the longest normal dry period 

 appear also to be conditions of importance in limiting the center. The 

 narrow amphtude of the moisture ratios for the center is largely to be 

 attributed to the low values which characterize the Ohio Valley (see 

 plate 59). An adequate series of evaporation and precipitation sta- 

 tions in the heart of the southern Alleghenies would doubtless give 

 maxima for the center nearly or quite as high as those for the fringe 

 (see values for Pisgah Forest, North Carolina, table 16). The west- 



