59 



The column is peculiar. In its lower two-thirds (fig, 1) it 

 is expanded into wide wings, which pass upwards and forwards 

 and unite so as to form a combed hood over the anther. The 

 upper third is produced, external to this hood, into a dorsal 

 clavate undivided process (fig. 2). The roof of the hood along 

 the line of union of the two wings is tomentose, and reaches 

 to about the same level as the dorsal process of the column. 



The anther (fig. 1) is attached to the base of the column 

 and is, with the exception of its point, hidden until maturity 

 of the flower behind the stigma. Its point (figs. 1 and 2) is 

 produced into a long, blunt, recurved, finger-like process. This 

 does not reach the roof of the hood until the flowers are ready 

 to expand. The pollen is very friable, there is no sign of 

 a caudicle, and no very definite connection between the ros- 

 tellum and pollinia. The former is, however, sticky, and a 

 small quantity of pollen adheres to it when pulled forward 

 with a needle-point. It is not possible to remove the pollinia, 

 however, by touching the rostellum, as in the case of cross- 

 fertilizing plants. 



The stigma (figs. 1 and 2) lies in the lower part of the con- 

 cavity formed by the column and its wings. It is very sticky 

 and somewhat triangular in shape. The lower part of it is 

 concave, and projects obliquely forward. 



The rostellwrn is well marked, and occupies the upper 

 angle of the stigma. 



As the flower matures the anther is carried slightly up- 

 wards until its point touches the hood and the upper parts 

 of the pollen-cases become exposed (fig. 3). An oblique 

 dehiscence of the cases then takes place, leaving a compara- 

 tively small portion of the pollinia uncovered just above the 

 rostellum and upper border of the stigma. This soon dries, 

 and crumbling away falls upon the rostellum and upper part 

 of the stigma (fig. 4). In hot weather, when the flowers open 

 freely, the process appears to be facilitated by the hardening 

 and bending forward of the rostellum. As some pollen has 

 become adherent to it, a degree of slight traction is exerted 

 upon the masses, helping them to rupture and fall more freely 

 upon the stigma (fig. 5). More frequently, however, the 

 rostellum undergoes a degenerative process of black cicatriza- 

 tion or necrosis (fig. 6). The contraction thus caused increases 

 the area of pollen exposed and assists materially in pollination. 

 Then, too, the bending forward of the anther over the stigma, 

 which occurs about this time, is also a beneficent factor in the 

 process which must not be overlooked. Apart from these 

 physical factors, I believe that the breaking up of the pollen 



