OF CENTRAL AFRICA. 291 



fore, exceptions to all the assumed characters of ^Miniosese 

 arc found, and there is some approacli in both genera to 

 the habit of Cirsalpineoc. It is still possible, however, to 

 distinguish, and it will certainly be expedient to preserve, 

 those two tribes or orders. Abandoning divisions strictly 

 natural, and so extensive as the tribes in question, merely 

 because we may not be able to define them with precision, 

 while it would imply, what is far f]'om being the case, that 

 our analysis of their structure is complete, would, at the 

 same time, 1)e fatal to many natural families of plants at 

 present admitted, and among others to tlie universally re- 

 ceived class to which these tribes belong. No clear cha- 

 racter, at least, is pointed out in the late elaborate work of ]M. 

 De Candolle/ by which Legiiminosa3 may be distinguished 

 from Terebintacese and llosaceae, the orders supposed to be 

 most nearly related to it. It is possible, however, that such 

 characters, though hitherto overlooked, may really exist; 

 and T shall endeavour to show that Leguminosae, indepen- 

 dent of the important but minute differences in the original 

 structure and development of its ovulum, niay still be dis- 

 tinguished at least from Ilosacea\ 



In the character of Poly galeae, which I published in 1814," 

 I marked the relation of the parts of the floral envelo})es to 

 the axis of the spike, or to the subtending bractea. I in- 

 troduced this circumstance chiefly to contrast Polygalea3 

 with Leguminosae, and to prove, as I conceived, that Secu- 

 ridaca, which had generally been referred to the latter 

 family, really belonged to the former. 



M. Dc Jussieu, who soon after published a character of 

 Polygaleac, entirely omitted this consideration, and con- 

 tinued to refer Securidaca to Leguminosa?. M. De Can- 

 dolle, however, in the first volume of his ' Prodromus,' has 

 adopted both the character and limits of Polygalea3, which [236 

 I had pro])osed, though apparently not altogether satisfied 

 wdth the description he himself has given of the divisions of 

 the calyx and corolla. 



The disposition of the parts of the floral envelopes, with 



1 Memoires siir la TamUk des Legumincuses. 



2 Flinders' s Voy. to Terra Anstr. 2, p. 51-2. (Jtilt',j}p. 13, li.) 



