210 



NATURAL HISTORY OF PLANTS. 



ascendincf ovules. In tlie two first-known species of this genus, 

 Ji. <dh(i' and rindh; both from the same country, the carpels are 

 pluriovulate, but there are only three of them, while there are no 

 more than six stamens. Tinally, of the small flowers of B. heterantha' 

 (figs. 243-247), which grows in the islands to the east of Africa, 

 some have also six stamens, of which three, the shorter ones, are 

 superposed to the interior petals. But in other flowers these three 

 are completely lost. Thus they only possess a calyx of three divi- 

 sions, three outer petals, three inner ones contracted at the base, the 

 three stamens answering to the outer petals, at the same time that 

 the gynaiceum is reduced to three carpels, usually uniovulate ; so 

 that these flowers present us with the greatest degree of simplifica- 

 tion as yet met with among Anonacece. 



Moreover, in this group, as in many others of the same order, 

 without any diminution of the number of pieces of the androceum, 

 those of the gynaiceum may be reduced to a single one. This does 

 not appear constant in B. canesccns f for the single biovulate ovary 

 observed in some of them is' inserted laterally, on one side of the 

 receptacle, while the place for the other carpels remains unoccupied. 

 In B. Gai(dicUaudiancC' (figs. 248-250), which we have made the tj^pe 

 of a particular section," the ovary of the single carpel forming the 

 gynaeceum is apparently terminal, which would seem to indicate 

 that it has been solitary the whole of its existence. Its ovules are 



Jlerh., n. 2668. In this species we have demon- 

 strated that there are often more than twelve 

 stamens, and from ten to twelve or fifteen car- 

 pels. There are some carpels with two ovules, 

 and others containing as many as four. The 

 asstivation of the corolla is certainly valvate. 

 The inflorescence usually springs from a sort of 

 excrescence on the wood of an old hi'aiich, whence 

 arise flowers of many successive generations. 



' A.S.ll., loc. cit. The stamens arc here six 

 in number, as in the following species. The 

 anther-cells are linear adnate, and almost mar- 

 ginal. 



- A.S.H., loc. ci/., t. 9. Here the carpels are 

 superposed to the outer petals. The inflorescence 

 of this species jtrcsents remarkable peculiarities 

 (see Adansonia, viii. 16 1). 



3 H. B>'., Aflntuionia, viii. 173. The very 

 small flowers of this species are borne on 

 very long ca])illary pedunclo;;. The carpels, sup- 

 ported on little slender feet, rarely contain two 

 ovules. We liave seen that the inner corolla of 

 this species is already in some respects that of 



the MitrephorecB. Nevertheless, it is inseparable 

 from the Old World Alphonseas. 



* SPBrCE, exs., n. 3549. — H. Bn., Adan- 

 sonia, viii. 171. — Trigyneia ? canescens Benth., 

 Journ. Linn. Sac, v. 70. In this species the 

 petals ai-e all similar, short and concave. The 

 calyx is gamosepalous, with its three angles not 

 prominent. The stamens may be as many as 

 fifteen in lunuber, and are pretty regularly ar- 

 ranged on the receptacle. The short filament 

 is surmounted by a large fleshy body like an 

 elongated cone, to the sides of which are applied 

 the two anther-cells close to the base. The 

 ovary is surmounted by a short reflexed style. 

 On the sides of the receptacle we see the places 

 for the aborted carpels. 



^ H. Bn., Adansonia, Nnii. 183. 



* Sect. EremodeJphis (See Adansonia, loc. 

 cit.). Bentham & HooKEK {Geh., 21), had 

 already pointed out an Alphonsea with a uni- 

 carpellary gynseceum. There is no reason 

 to separate generically the species in which this 

 occurs ; for in other generic groups there are 



