DILLENIACEJE. 



Ill 



towards the base. The five petals alternating with these, are also 

 imbricated in the bud. The indefinite liypog-ynous stamens each 

 consist of a free filament, and an extrorse 2 -celled anther, more or 

 less versatile on the summit of the filament, and dehiscing longi- 

 tudinally.' The gynseceum is free ; it consists of a thick central 

 axis surrounded by from twenty to thirty cells, each surmounted 

 by a spreading style reflexed on the top of the ovar}^, and stigma- 

 tiferous on the upper and inner surface. In the internal angle of 

 each cell is a placenta bearing numerous anatropous ovules. The 

 fruit, which is surrounded by the persistent calyx, becomes a multi- 

 locular berry, the pulp of which contains numerous seeds with 

 thick coats"" and abundant fleshy albumen, surrounding a central 

 elongated embryo with small cotyledons.^ In other Actinidias the 

 flowers often become polygamous, by abortion of the gynseceum. 



Fig. 161. 

 Fruit. 



Actinidia strigosa. 

 Fig. 162. 

 Longitudinal section 

 of fruit. 



Fig. 163. 

 Seed. 



Fig. 164. 



Longitudinal 

 section of seed. 



They are shrubs, often creepers, twining, with alternate simple 

 penninerved leaves. The flowers are axillary to the leaves, solitary, 



' The anther is extrorse after anthesis ; it is 

 also distinctly so in the young buds of A. rugosa. 

 The difl'erence between the stamen, as enclosed in 

 the bud, and that of the expanded flower, is that 

 the top of the filament is quite straight before 

 expansion, while afterwards it is bent twice on 

 itself before giving attachment to the connective. 

 In the flowers of A. strigosa, the anthers are also 

 extrorse. The connective is narrow and entire 

 above, but bifurcated towards the base of the 

 anther ; and as its branches diverge below, so do 

 the cells of the anther. 



^ These seeds have no aril ; a character cited 

 by several authors (Bentham & Hooker among 



otliers) to justify the separation of Actinidia from 

 DilleniacetB ; but we have elsewhere {Joe. cit. 

 258) called attention to the tact that in 

 Dillenia itself the seeds may want an aril. 



■^ This embryo is straight, and more developed 

 than that of most Dilleniace<p, which is at the 

 apex of the albumen. This is another of the 

 characters which removes Actinidia from these; 

 " oh embrgonem magis evolufum," say HKXTH.\xr 

 & Hooker (op. cit. 11). We do not ascribe much 

 importance to this character. l?iit tlieafliuities 

 of Actinidia with Saiintiija, of the order Tent- 

 siroemiacefE, are incontestible. 



