40 HARPER 



tend to counterbalance each other when the statistics are 

 consolidated. 



These phaenological diagrams are somewhat of an innovation, 

 but their importance will probably be better realized when they 

 come into more general use. They will doubtless afford a valu- 

 able means of comparing different habitat-groups in the same 

 region, and similar habitat-groups in different regions. And their 

 application is by no means restricted to habitat-groups, but 

 may be extended to structural and taxonomic groups, such 

 as trees, shrubs, families, genera, etc. 



By dividing the area of any phaenological diagram by the 

 number of species entering into it the average duration of the 

 flowering period for single species is obtained. Although ex- 

 treme accuracy could not be expected on account of the present 

 incompleteness of my observations, the figures obtained in this 

 way for different diagrams are remarkably consistent. 



Another interesting feature brought out by these diagrams, 

 which might have escaped attention otherwise, is that most of 

 them show a falling-off in the number of species in bloom about 

 the first of May. This is not a peculiarity of the Altamaha Grit 

 region, for I found the same to be true in Middle Georgia in 1896, 

 and I have recently ascertained that in Massachusetts (in some 

 habitats at least) a similar decrease comes about a month later. 

 This falling-off can hardly be correlated with any climatic feature, 

 but it probably indicates that there is a more or less fundamental 

 distinction between spring and summer flowers, connected most 

 likely with the leafing-out of the deciduous trees. 



Last but not least, each habitat -group, where practicable, is 

 illustrated by one or more photographs. 



It should be borne in mind that Nature draws few hard and 

 fast lines, and the nearest we can get to her methods is only an 

 approximation. I may have drawn the limits of the different 

 habitat -groups too far apart in some cases and too closely in 

 others, but further study will always bring us nearer the truth. 

 By the frequency method above described the typical and 

 characteristic members of each group are placed at the head 

 of the list, while those whose membership is more or less 



