84 MB. BENTHAM ON LOaANIACEJE. 



than two cells to the ovary; and of the three most important 

 modes of insertion of the ovule, pendulous in Canthiece, erect in 

 'Psycnotriew, and laterally peltate in Ixorece, we have in Gart- 

 ner ece only the two last. 



The connexion with the allied free Monopetalce is very slight. 

 Pagamea and Gartner a are Rubiaceous even in their stipules. 

 Gardnera may indeed show some approach to Apocynece in its 

 anthers and less prominent stipules, but not in any other par- 

 ticular. 



23. Gabdnebia, Wall. 



The ovary in this genus is at the time of flowering so small and 

 fleshy, and the cavities so very minute, that it has always been 

 found very difficult to ascertain its structure. It has generally 

 been supposed to have two cells, with one ovule in each ; but this 

 is doubted by DeCandolle. Zuccarmi describes one pendulous 

 ovule in each cell, and the berry is certainly usually two-seeded. 

 But four-seeded berries are occasionally to be met with in one 

 species, and Wight represents more than one ovule in each cell of 

 the other. I have therefore been led to examine very carefully 

 the ovaria of a considerable number of specimens ; and although 

 at first I had some difficulty in tracing them in the G. ovata, I 

 subsequently found them in several flowers, but always solitary 

 and exceedingly minute. In the G. angustifolia. they appeared 

 more distinct, and always two in each cell, collaterally attached by 

 their centre. I found the same structure also in one of Siebold's 

 Japanese specimens of that species. 



On searching among Bubiacece for the nearest parallel to 

 Gardneria, I was struck with the very close resemblance in habit 

 and foliage of a Singapore plant, determined by Gardner to be a 

 species of Ccelospermum. It is in fruit only, but evidently belongs 

 either to that genus or to Gonochthodes, of neither of which have 

 we any authentic Javanese specimens, but which are probably 

 not really generically distinct. The seeds of Ccelospermum are 

 shaped like those of Gardneria-, but they are enclosed in the 

 crustaceous endocarp, so that the fruit, instead of being a berry, 

 is a drupe with as many pyrenes as seeds. The number of these 

 is four, as in many fruits of Gardneria angustifolia, and the genus 

 is consequently placed among those with a four-celled ovary. The 

 ovary, however, is not described by Blume ; and, from the fact of 

 the style being bifid only, one would be led to conjecture that it 

 was two-celled only, with two collateral ovules in each cell, show- 

 ing a still further analogy to Gardneria angustifolia. 



