158 BOTANY OF SOCOTRA. 



expanding at the base, and joining on to the lateral short rounded lobes, 

 each of which is set on the corolla at an obtuse angle to the radius of the 

 flower, and forms with the base of the filiform lobe a small cup on the surface of 

 the corolla. The lateral lobes of each scale nearly coalesce across each corolla 

 lobe. The stamens, which have short filaments, stand embraced between the 

 lateral lobes of the scales and opposite the central lobe, and are attached below 

 this last to the corolla tube by an insertion extending to the bottom of the 

 tube. At the base of the tube the filaments expand laterally, become connate, 

 and thus form a ring around the pistil, and the interstaminal portions of this 

 ring are produced into fimbriated scales, which, bending inwards, conceal the 

 ovary. But the stigmatic structure is more remarkable. The stigma is 

 expanded into a large more or less peltate disc, with a low conical vertex, but it 

 is 5-angled at the side, and each surface of this small pyramid is clothed with 

 a membranous inverted hood, from the inferior extremity of which there 

 depends by an elastic membranous articulation a somewhat brick-shaped body, 

 hollowed on one side and closely appressed at first to the style. It is not 

 difficult to disengage these appendages from the style, and then they spring up 

 by means of their articulation and project from the stigma at a greater or less 

 angle. The stigma is altogether more Asclepiadaceous than Apocynaceous. 

 But as I have already mentioned, the stamens are quite free from it, and there 

 are no pollen corpuscles. What is the nature of these bodies pendant from the 

 stigma, remains doubtful. They can hardly be corpuscles, as the stamens are 

 always quite free and separate from the stigma, and the bodies do not approach 

 the anther sacs. They may be the stigmatic surfaces. Each has a viscid 

 concavity, and we find in some genera of Apocynads lateral portions of the 

 style sometimes stigmatiferous. Or they may be merely secreting glandular 

 appendages. Unfortunately we have but a few flowers for examination, and 

 are therefore unable to determine this point. 



Amongst Echitidece the genus is aberrant. The aphyllous habit is very 

 exceptional, not only in the tribe but in the order. Most genera in this tribe have 

 the carpels distinct from one another below the style, but there are a few genera, 

 — Vallaris, Lyonsia, Parsonsia, and others, — in which the carpels are connate, 

 and this is their character in our Socotran genus. The three I have mentioned 

 fall into the section of the Parsonsiece, characterised by the connivent anthers 

 forming an exserted cone ; and our genus conforming with that character must 

 be referred to the same tribe, and its position is in the vicinity of the above 

 genera. 



Its individuality is so strong that it is hardly necessary to refer in detail to 

 the characters by which it is diagnosed from the neighbouring genera. The 

 seed comose at the apex, and the straight cotyledons, and its distribution, exclude 

 it from all except some five old world genera, Pottsia and Isonema being the 

 two m addition to the three above mentioned. From them the stamens not 



