Observations on South African Plants. 33 



The question naturally arises whether the flower I have described 

 represents a case of atavism, a reversal to an original type. I will, 

 however, not venture on any speculation on this subject, for on the 

 two spikes which I had at my disposal there were some flowers 

 which resembled the one which I have described, but were not quite 

 so regular, and did not possess any staminodes, but there were 

 also several with a different structure. They had a floral envelope 

 composed of two alternating whorls, each composed of two hooded and 

 saccate structures similar to the sepals of the flower first described ; 

 they had four stamens equally well developed as the stamens of the 

 first flower, and they had (I believe) two stigmata. ;;: Here we have, 

 therefore, orchidaceous flowers which are dimerous right through. 

 I think it right to put these cases on record, but I do not feel justified 

 in drawing any conclusions from the facts which I have ascertained. 



2. The structure of the flower and the method of 'pollination in 

 Grassula canescens, Schult. 



The following observations were made on plants grown in my 

 garden at Grahamstown, within about half a mile from a spot where 

 the species is growing in its wild state. 



Grassula canescens, Schult., belongs to the section Globulea which 

 is characterised in Harvey and Sonder's ' Flora Capensis ' as having 

 panduriform petals which bear an oblong or ovate fleshy "gland" 

 immediately behind the blunt apex. I do not know what has led to 

 the belief that the knob-like structures on the petals are glands, but 

 I may state at once that they are not excretory organs, and for want 

 of a better term I will call them " epipetalous knobs." I have reason 

 to believe that in the other species of Cassula § Globulea these knobs 

 are in structure and function similar to those of C. canescens. t 



The peduncle of Crassula canescens is scape-like, and bears a few 

 densely capitate cymules. The appearance of the epipetalous knobs, 

 which are white, announces the time when the flowers are ready to 

 be fertilised ; before this time they are, with the rest of the corolla, 

 enclosed by the calyx. As soon as this stage is reached the flowers 

 are plentifully visited by a hymenopterous insect (Synagris emarginata, 

 Sauss. ?) I " A dissection of the flower does not show any appre- 

 ciable quantity of nectar, but when covering up a cluster of flowers 



* I could not be quite sure about their number. 



f This is certainly the case in C. radicans, Haw, C. cultrata, L., and C. obvallata, 

 L. As regards structure of flower, insect visitors, and fertilisation, my remarks 

 on C. canescens apply also to these species, as I was able to ascertain after writing 

 the above account. 



\ Mr. L. Peringuey, who kindly tried to identify the insect for me, could not 

 state its name positively as I could only send him a female. 



