ANONAGE^. 251 



sliown that this character is at most only sufficient to justify sub- 

 divisions within a genus. The stamens are almost always inde- 

 finite in the Anonacea, and it is only since the time of A. de Saint- 

 HiLAiRE that it has been known that Bocagca may have an andro- 

 ceum of subdefinite elements. The study of B. heterantlui has 

 proved to us that the number of stamens may be even quite definite, 

 limited to three or six ; as is also the case with some species of 

 Oro2jh(sa. But at the same time we have had to unite the American 

 Bocageas and the Asiatic Alphomecm into a single genus. Now the 

 latter often have indefinite stamens. Moreover, when the stamens 

 of the Anonace(B are very numerous, they appear, when adult, 

 arranged in a spiral, while in the species of three, six, or nine 

 stamens the existence of trimerous or hexamerous verticils appears 

 quite incontestible. In this respect the Anonacece would resemble 

 the Banunculacefje, having the pieces of the androceum sometimes 

 arranged in whorls, sometimes in spirals.' 



6. The form of the stamens, the relative size, direction, and position 

 of the anther-cells and connective, especially with regard to the pro- 

 longation of the latter, are of great, though not quite absolute, value 

 in separating genera." There is no reason for hesitating very much 

 before placing a species in one genus rather than in the neighbouring 

 one because its stamens are those of the Uoarlea, not of the Miluise(B, 

 or the reverse. We have seen how Bentham & Hooker go much 

 further, at the very outset relegating all the Anonacece with stamens 

 of the Miliusea to a separate tribe, though the other characters of 

 the flower are extremely variable in the different genera of this tribe. 

 Adopting their standpoint, we should perhaps need to adopt a third 

 type of staminal organization — that so well marked in the group 

 Clathrospermiiiii of the genus Popoivia. Here the stamens present 

 certain characters of the UvariecB ; for it is impossible to class 

 Popowia in the same division as the Miliusoe ; but yet Clathro- 

 spermum proper has been put in the latter category.'' 



' The study of development will alone finally ^ We have seen, for instance, that certain 



settle this question. Tlie numerous stamens of Anonas may he considered to possess stamens of 



the AnomtcefE like those of the Dilleniacea'., Bocagea (p. 222), and that in Anaxagorea U\c 



miglit well be originally arranged in bundles. connective often recalls that of several species of 



(For the chief details ccmcerning the arrangement MiUusa by its elongated form, aiid by being tliit- 



and tlie varying number of the pieces of tlic andro- toned and tapering at the apex (p. 207j. 

 ceum, see Adansonla, viii, 312-329.) •* See Adansonia, viii. 311. 



