-lO BLAKE: REVISION OF THE AVOCADOS 



with the avocado, has recognized three chief forms on the basis 



of leaf and fruit differences. These he separates by the following 



key: 



A. Leaves anise-scented : skin of fruit thin and soft, i . Mexican type. 



AA. lycaves not anise-scented: skin of fruit thick. 



B. Surface of fr. usually smooth: skin leathery, usually not more 

 than Vi6 in. thick; seed coats frequently distinct, the 

 outer one adhering to wall of seed cavity; cotyledons often 



rough 2. West Indian type. 



BB. vSurface of fr. usually rough or warty: skin brittle, granulir, 

 Vi6 -^/i6 in. thick; seed coats adhering closely to the nearly 

 smooth cotyledons 3. Guatemalan type. 



Mr. Popenoe, moreover, has become convinced that his "Mex- 

 ican type" represents a distinct species, Per sea drymifolia, 

 Schlecht. & Cham., and that Per sea schiedeana Nees is a distinct 

 species. Both of these forms were treated by Mez as varieties 

 of P. americana. 



In the study of the extensive material which has now been 

 brought together it became clear at once that Mez had erred 

 in reducing P. schiedeana to varietal rank, since it possesses 

 definite specific characters, not only in the pubescence of its 

 leaves and its long pedicels, but also in the technical features of its 

 floral parts, which were entirely passed over by Mez. P. drymi- 

 folia is a more doubtful form, not always distinguishable with 

 certainty in herbarium material, and I prefer, at least for the 

 present, to treat it as a variety of P. americana. The novel fact 

 appears, moreover, that the most commonly cultivated avocado 

 of Florida, known as the Trapp, represents a technically very 

 distinct and undescribed species, characterized by its sparsely 

 pubescent perianth and absolutely glabrous pistil and staminodes. 

 Another new species of the avocado group, characterized by its 

 glabrous ovary, narrow panicle, and sessile staminal glands, 

 has been collected by Purpus in Oaxaca and is here described as 

 P. cinerascens. 



While the relationships of the North American forms can be 

 considered as now established with a fair degree of definiteness, 

 this cannot be said of the South American types. In addition to 

 typical P. americana and its variety drymifolia, at least two 



