bH 



NATUILLL HISTORY OF PLANTS. 



is most reiiKirkable in the arrangement of these is, that those super- 

 posed to the five petals are external to those superposed to the sepals, 

 enfolding and hiding them entirely in the bud.' ]Moreover, the sepals 

 are unequal and quincuncial,the petals imbricated, the anthers introrse, 

 dehiscing by two longitudinal" clefts, as in Candollea. The carpels, 

 two in number, are free. The ovary contains one or two anatropous 

 ascending ovules ; the micropyle is originally^ turned inwards, and 

 even in the flower the umbilicus is surrounded by a rudiment of 

 the arillary collar. The style, grooved the whole length of its internal 



angle, tapers upwards to a point, 

 A. salicifolia is a small sufFrutescent 

 x^V ^mIIiI^ '^^3 J pl^^tgi^owiugin marshy soils, wherein 



its woody stock burrows, covered 

 with numerous adventitious roots. 

 The slender branches bear alternate, 

 very unequal leaves, placed close 

 together on the axis of a short branch 

 which ends in a nearly sessile flower, 

 leaves, the spiral of which they 



Adrastaa salicifolia. 



Fig. 125. 



Loiigitudiual section of flower. 



The sepals resemble the last 

 continue. 



Pachynema,^ of the same country as Adrastcea, has its flowers (figs. 

 126, 127) similarly organized. But of the stamens only seven or eight 

 are fertile. Their small anthers are two-celled and introrse; the 

 filaments supporting them are dilated from above downwards to 

 form a kind of pyramid. The two innermost stamens are reduced 

 to these filaments, each bearing at the tip a sterile gland instead of 



' On this account we have not chosen Adrastd-a 

 as tlie tirst tvpe of tiie DiUetiiacpce, and also be- 

 Giusethe exact alternation of these stamens with 

 one another, and their e.\act superposition to the 

 pieces of the calyx and corolla do not always 

 exist. Ilencc we may conclude, as we have else- 

 where said {Adansunia, vi. 265), "that here we 

 liavc not to deal with the usual androceal whorls 

 fotnid in regularly diplostemonous flowers." The 

 study of the evolution of the androceum will alone 

 reveal its true symmetry. But there is no doubt 

 that the stamens are not, as Bentham «fe 

 HooKEB assert, xympliri serie (rquali/er peri- 

 pherica." Some are so much internal to the 

 others, that in the bud they are not seen on re- 

 moving the corolla. 



- These clefts begin near the top of the 



anthers ; those of the two cells approach closely 

 at the summit, but without coalescing. The 

 filament, flattened and broad, is almost peta- 

 loidj the connective is continuous with it, 

 and the anther cells are so applied on the inner 

 face that nothing of them is seen on the 

 dorsum. 



^ When there is but one, as it developes it 

 undergoes a more or less decided twisting, so as 

 to turn the micropyle sideways, or even out- 

 wards. 



■» K. Brown, in DC, Syst., i. 412 ; Prodr., i. 

 70. — Deless., Icon. Sel., i. t. 73. — Endl., Gen., 

 n. 4756.— B. H., Gen., 15, n. 16.— Benth. & 

 F. MuELL., FL Austral., i. 47. — H. Bn., 

 Adansonia, vi. 279. 



