110 Mr, B. Clarke on the Anthers of 



of them may tend to remove the difficulty, if, as I will endeavour 

 to show, their structure proves to be not at all without a parallel. 

 Those who have noticed the stamens of Columellia, regard each 

 of them as consisting of a monadelphous bundle formed of three, 

 the anther itself being regarded as six-celled, and consequently 

 as having three pairs of lobes, indicating the presence of three 

 stamens : a careful examination, however, of the species has led 

 to the conclusion that they are always single, being never more 

 than two-celled, their plurilocular appearance being entirely due 

 to the sinuous character of the lobes*. (PI. VI. fig. 20.) 



The anthers of the species differ but very little from each 

 othei-, further than in the degree of convolution of the cells, which 

 seems owing merely to their variable length ; and a comparison 

 of them with those of Cucurbitacese, especially of Cucumis sativa 

 and Bryonia dioica, will, I believe, be a convincing proof, that if 

 the nearest affinity of this family is not with the Cucurbitacese, 

 yet there is no other to which it more closely approaches ; but, 

 for the pui'poses of further comparison, a short character is 

 added. 



Shrubs having the habit of Myoporacese and Scsevolese, especially 

 of the latter. Leaves opposite, without stipules. Flowers 

 solitary, axillary, with two conspicuous bractese at the base of 

 the ovary, or more or less adherent to it. Calyx superior, 

 with five segments, which are small and acuminated, the odd 

 one being posterior. Corolla monopetalous ; limb five-lobed, 

 in pestivation imbricated, but not quite regularly so, the same 

 segment not being always external. Stamens two, adhering 

 to the corolla near its base, mostly alternate with its segments, 

 being placed between the two posterior and two lateral, but 

 occasionally showing some irregularity, one of them, if not 

 both, becoming opposite or uncertain in their relation ; fila- 

 ments thick and very short. Anthers two-celled, opening 

 outwardly (or incumbent, the larger portion opening out- 

 wardly), the cells much elongated and tortuous, each of them 

 being doubled on itself towards the centre of the connective, 

 so that its two extremities closely approximate both to each 

 other and to the extremities of the other cell. Disk epigy- 

 nous, not in any degree perigynous. Ovary two-celled, the 

 cells anterior and posterior; style short and thickened ; stigma 

 bifid, consisting of two flat, semicircular lobes. Ovules nu- 



* The filament, although very thick, contains but one bundle of vascular 

 tissue, which is quite entire and compact down to the base of the corolla, 

 but above spreads out into tortuous meshes, along the margins of wliich 

 the cells of the anther are attached, so that it is not unlike an aquifoliaceous 

 leaf, the apex of which is bent down internally towards the base. 



