CORRELATION OF DISTRIBUTIONAL FEATURES. 



395 



much more closely than do the climatic values for the nearest main- 

 land stations, which are often far distant. 



It has been particularly difficult to determine the southern limits of 

 southeastern plants in peninsular Florida, and our data as to the cli- 

 matic extremes for plants reaching southern Florida should be regarded 

 as only tentative in those cases in which the values for Key West or 

 Miami have been used. 



It is impossible, in general, to determine the optimum value of a 

 given climatic condition by any method which involves taking into 

 account the number of stations with readings of a given value, owing 

 to the irregular distribution of the climatic stations. As an example 

 of what might be done by this method, however, we have constructed 

 the diagram shown in figure 20, in which is shown graphically the 



1.20-1.30 



1.60-1.70 



Fig. 20. — Graphs showing the frequency in each of three different vegetations, of the grouped 



stations to which the progressive values of the moisture ratio tt/E correspond. 



Ordinates represent frequencies of occurrence of the ratio-value ranges shown as abscissas. 



number of stations in the Grassland, the Grassland-Deciduous 

 Forest Transition, and the Deciduous Forest, at which have been 

 determined the several progressive values of the moisture ratio ir/E 

 that are indicated. As there is a relative uniformity in the distribution 

 of the climatic stations in these three vegetations, and as results have 

 been expressed in percentages of the total number of stations in each 

 vegetation, we may regard this graph as a rough means of showing 

 which of the intensities of this condition is found over the largest area 

 in each of the given vegetations. No ratio values below 0.20 are found 

 in the Grassland, and the greatest number of stations have values 



