5 



■ 



Tlie external glands afford anotlier feature of doubtful signU 

 ficance. Tliey are present and j^la'iuly visible either as small 

 circular or oblong- bodies alternating with the perianth-segments 

 iu many species (<?.^<7., T. nigroviontanum] T, .lacinidatum, T\ 

 coriaceam, oic,—see figs. 3, 5, 8,. 9), but in others again tliey 

 cannot be recognised, and it is often a matter of doubt whether 

 they lire entirely absent or only very slightly developed. They 

 inay be found in species belonging l)oth to tlie bearded and un- 

 bearded sections of the genus. A detailed investigation"^' of tlieir 

 morphology is not possil)lc in dried material; it is therefore to be 

 hoped that a botanist in Sonth Africa may be stimulated to nialie 

 an examination into the morphology and physiology of these 

 obscure organs as Avell as of other points connected witli the 

 biolugy of the flowers of Thesiin)t. 



In the bearded section, Barhata, after tlie primary sub- 

 division on the question of the presence or absence of hairs behind 

 the anthers, the chief point of distinction iu the flowers lies in the 

 character of the apical beard and the texture of the perianth- 

 segments. More commonly th(* beard consists of a dense mass of 

 woolly, waved hairs, but sometimes the beard is composed of rela- 

 tiA'cly few, stiff, straight, con.b-lilve hairs. In some of the species 

 'with this ,type of beard the perianth-segments are h(»rny and 

 translucent and the margins of the segments are fringed with a 

 frill of short fine hairs, as for instance in 7\ apicatinn (fig. 15). 



The stiff and the woolly types of beard have been found to afford 

 a good character for se2)a rating ci^rtain species such as T. junceuiVy 

 T, fle.ruosmn, T. natfilense^ etc., wliich e:^hibit considerable 

 external resemblance. ' 



The position of the anthers, either included in the tube oi- 

 exserted and carried uj) against ilio perianth-tir'gnn^nts is an 

 important character iw i\\^ section Barlmta and is often found to 

 be associated with the len«»(h of tlie stvle. When the stiijma is 

 sessile tlie anthers are often inserted deep down in the tube, as in 

 y\ cm 'mat am {?i<^. 10). 



The possibility that heterostylisin may occur in the genus is 

 suggested by the floral arrangements of the two closely similar 

 8j)ecies, 1\ carinatnm and 7\ capitatum (cf, figs, 10 and 11), but 

 the relative lengths of anthers and style in the two species, thougl 

 highly suggestive of this state, are not quite convincing. So 

 similar in general appearance are these two species, however, 

 that it is frequentl}' impossible to assign a specimen to the 2)roper 

 species without dissegting the flower. ■ 



Througliout the genus it is unusual for the style to be of such a 

 length that the surface of the stigma is above the level of tlie 

 anthers. Sometimes it is at about the level of the middle of the 



\ 



The gland-like structures have been exaniiaed, as far as is possible 

 with herbarinm material, by Mr. L. A. Boodle. These portions of the periantli 

 are rather -thicker than the adjoining ]>arts, and may have dense or browu 

 contents in tbeir internal tissues. They are all the more striking in com- 

 jjarisoh with the large, nearly empty cells lying adjacent which are connected 

 vnth the statninal hairs. The epidermis bounding the structures externally 

 lias a considerably thinner wall than the epidermis oi the intervening parts, 

 nnd this featnre may indicate that an excretion i^s exuded from them on the 

 outer side of the perianth tube. 



