ANONAGEJE. 253 



SO mnch expect to find such diflerences of direction. Tlie ovules of 

 Phceanthus and EUipeia are horizontal or slightly ascending, though 

 inserted some way up the placenta. That ovules should be ascend- 

 ing or nearly erect when there are one or two nearly basilar, is yet 

 more inteUigible. The micropyle in this case looks outwards and 

 downwards ; this we find in Anona, Polj/althia, certain species of 

 Trigyncia, &c. But a good proof that a solitary ovule has not neces- 

 sarily the same ascending direction in all the species of a genus, is 

 afforded by the plant we formerly called Tngyneia ri/fescens' which 

 has a slightly descending ovule, with the micropyle looking upwards 

 and inwards, though possessing all the other floral characteristics 

 of its congener, the form lanceolata of Anona Perrottetii A. DC.,' 

 whose ovule is ascending. We intend at some future time to make 

 known a third species very near these two, in which the ovule is 

 even more markedly pendulous. 



10. The arrangement of the ovules, w^hether in one or in two 

 rows. — This character can have no great importance, being only 

 determined at a rather advanced age of the gynteceum. At their 

 origin, all the ovules, w^hen numerous, are probabl}' arranged in two 

 parallel rows. It is only later on that those of the one row become 

 interposed to those of the other, both sets gradually approachino- 

 the ventral median line. On splitting up the carpel through the 

 longitudinal internal groove, we usually effect a separation of the 

 ovules into two equal sets, one each side of the cleft, though they had 

 appeared ranged in a single vertical row. In certain genera quoted 

 as having sometimes one, sometimes two rows, we have always found 

 two.^ We shall never use this character to separate two genera. It 

 has no more value in the fruit than in the flower ; for ovules that 

 were in the flower arranged in two rows may correspond to seeds 

 superposed in a single row in the fruit, and ovules so close together 

 as to appear in a single vertical row, may develope into seeds 

 arranged in two very distinct ranks.' 



11. The presence or absence of contractions in the fruit answer- 



^ Adansonia, viii. 180, n. 1. * " The fact of the arrangement (of the 



' Adansonia, viii. 179, note 5. ovules) in two rows, probably exists in all cases, 



3 In Rexalobus for instance, of which all the but is not always quite so clearly shown ; and it 



species are alike in this respect, except H. mada- is only worth while to base o-enera on this cha- 



gascariensis A. DC, which is unknown to us racter when the two rows are very far apart 



(see p. 227, note 6), and should possibly be re- instead of close together. But that does not 



ferredtothe genus Monodora. occur in Anonacece." (A. DC, Mru/.. 7.) 



