93 



cut off from the tree, and finally in many cases the entire inflor- 

 escence — when it does not support developing fruit. 



As the earliest flower buds expanded in the latter half of 

 January, the trees began to cast off the old foliage which had 

 served them ten or eleven months. As is frequently the case with 

 deciduous tropical trees whose flowers are produced when the 

 limbs are almost or quite leafless, both the process of defoliation 

 and the expansion of the flowers begin and progress most rapidly 

 at the top of the tree, with the result that this usually becomes 

 quite bare of foliage while the lower branches are still well-clothed 

 with the old leaves (see Wright 4, p. 475). Not only are the lower 

 branches the last to lose their old leaves, but they are the first to 

 acquire the new, which are in many cases put forth and sometimes 

 even attain full size before the old leaves in the same part of the 

 tree have been completely shed. Thus it happens that the lower 

 stories of some trees are never bare, while the upper half is 

 often denuded of foliage during at least a considerable part of the 

 period of flowering. Frequently, however, a tree will be found 

 quite leafless at the height of the flowering season, when it is yellow 

 with the myriad buds and blossoms it supports. As with the sassa- 

 fras, the flowers themselves are borne on leafless branches which 

 spring from the base of the newly expanding annual shoot, and 

 consequently are situated below the young foliage. The tender 

 leaves are a very light green and contrast strongly with the sombre 

 hue of the old foliage still persisting on the tree. By the second 

 week of March, 1929, some of the trees had completed their flow- 

 ering, while in others flowers were still produced on the upper 

 branches, after the lower had spent all their buds. About half the 

 trees were in full new foliage, but a few were still rather bare. 



Summary 



Observations were made on the flower behavior of eight avo- 

 cado trees in a dooryard planting in western Panama. Two of these 

 were typical class A trees, as defined by Stout, and two belonged to 

 class B^. The other four, while giving indications of the type of 

 behavior of class B trees, were erratic in that only a small per- 

 centage of the flowers opened in the afternoon. In most the first 

 period opening was suppressed entirely, and there was a single 

 long period of anthesis in the morning, when the pollen was shed 

 at the same time the stigmas of the same flowers appeared recep- 



