NOVEMBER. 267 



all know that prior to any trace of frost some forest trees 

 lose the green tint of their leaves, and assume quite brill- 

 iant colors ; and, too, a single branch may change, while 

 the other limbs of the tree remain unchanged ; again, in a 

 cluster of trees of the same species one or more may 

 change, but not the others this occurring either before 

 or after a frost or a succession of them. 



Fortunately, premature frosts, as we may call such as 

 occur earlier than September 20, are of such rare occur- 

 rence that their effects can not be satisfactorily studied. 

 It has often been asserted that when such frosts did occur 

 the foliage quickly responded by changing its color. In 

 but one instance have I been able to test the truth of this, 

 and I found that the trees apparently affected were in 

 every case those that change early in September, quite 

 irrespective of the temperature. Like the Virginia 

 creeper and the tupelo, there are several trees and some 

 small shrubs that undergo this change of color as the sum- 

 mer draws to a close, but such growths when scattered in 

 a wood are not apt to be noticed. Autumn leaves are 

 not generally known as a phenomenon of summer, so 

 are not looked for ; but they are a sturdy fact, neverthe- 

 less. 



So uncertain and contradictory seemed the whole 

 matter that I have for several summers followed the trees' 

 course from early spring until autumn, in hopes of learn- 

 ing something concerning the supposed relationship of 

 frost and the coloring of leaves ; the following may bear 

 significantly upon the question. 



The condition of the growth of the leaves in spring 

 appears to have much to do with the progress of the 

 autumnal change. As an instance of this I may men- 

 tion that three enormous beeches near my home were in 

 full foliage, May 1, 1886, and the rich yellow-green of the 

 growing leaves had wholly disappeared. During the first 



