Observations on South African Plants. 39 



sympodium are always antidromous, whereas the flower terminating 

 the ill-developed "free" shoot is homodromous with the flower 

 terminating its relative main axis. The result of the arrangements 

 which I have described is such that all organs of the plant which 

 require light get a maximum of it. Given the same forms of stems, 

 leaves, and flowers, and the well-known straggling habit of M. 

 angulatum, I doubt whether any other combination of these parts 

 could secure to them more advantageous positions than they occupy 

 in nature. A fuller consideration of this point, which is of the very 

 greatest importance, I will leave for the present but hope to revert 

 to it in some future paper. 



b. The development of the flower in Mesembrianthemum angulatum. 



There is nothing remarkable in the structure of the flower of this 

 plant, except that the placentation remains permanently axile, 

 whereas, as is well known, in a great many species of Mesembrian- 

 themum the placentation, though originally axile, becomes parietal 

 in the fully developed flower. My reason for studying the develop- 

 ment of the flower of M. angulatum was chiefly to form an opinion 

 of my own on the origin of the so-called petals in the genus. The 

 result agreed entirely with the facts found by Payer and Hagen. The 

 very numerous so-called petals and the stamens arise together from 

 five protuberances or " primordia " (Plate II, B pr) after the sepals 

 have made their appearance. These split up by a series of radial and 

 concentric divisions, and the outer of them become the petals, where- 

 as the inner ones develop into stamens (Plate II, C st). Strictly 

 speaking, therefore, the so-called " petals " are staminodia, and the 

 genus Mesembrianthemum is to be regarded as apetalous, as is now 

 usually done by botanists. In our case the three outermost rows of 

 serial structures resulting from the five primordia of the androccium 

 develop into staminodes (Plate II, C st'), and as there are usually 30 

 radial series formed, their total theoretical number is 90. I counted 

 82 in one flower. , As is usual in structures arising in close proximity 

 to one another, some succumb to the pressure exercised by the 

 others, and thus the smaller actual number may be accounted for. 

 Altogether there are usually eight structures in each radial series — 

 this would leave five for the real stamens, and their total theoretical 

 number would be about 150. 



The staminodes soon flatten out and overlap, but even in a full- 

 grown flower one cannot help being struck with the similarity 

 between the filaments of the proper stamens and the staminodes, 

 and this similarity is greatest where the two adjoin, and from this 

 fact one might also be led to the conclusion that the so-called 

 " petals" are stamens without anthers, or in other words staminodes, 



