SAPINDAOE^. 379 



fruit, whose seeds are or are not provided with an aril and whose 

 embryo is generally exalbumiuous, more rarely surrounded by a layer, 

 usually thin, of this perisperm. The other characters, to which a lower 

 value only can be allowed and which serve in general only to dis- 

 tinguish genera, are drawn from : the squamiform appendages which 

 do or do not accompany the petals inwardly ; the number of the ver- 

 ticals of the androceum, sometimes comj)lete and sometimes incom- 

 plete ; the number of ovules and theii- direction ; for there are often 

 one or two in each cell, ascendent and more often descendent, and 

 sometimes an indefinite number ; in which case they are arranged on 

 two vertical rows ; the shape and consistence of the fruit, rarely 

 fleshy, nearly always dry, but sometimes capsular and dehiscent, 

 sometimes indehiscent and sometimes, in the latter case, prolonged in 

 samaroidal wings ; the dimensions of the aril when it exists, for it 

 may surround the whole seed or only forma cupula- shaped expansion 

 at its base ; finally in the shape of the embryo, which is straight, 

 arched, contortuplicatc or twisted in a spiral. 



There are two of these characters to which, it seems to us, we 

 must be very careful not to accord an absolute value for distinguish- 

 ing the tribes or series of this family : they are the position of the 

 leaves and the regular or iiTegular form of the corolla. By the first, 

 the j^sculece [Hippocustanece) are separated very distinctly, at first 

 sight, from the allied genera of the group of Sapniuleie. But we can 

 very well understand Bentham and Hooker having included them in 

 this series, and we have only kept them separate Avith hesitation, 

 because we see genera such as Valemiielia, having all the essential 

 characters of the flowers and fruits like those of Paullinecv fi.-om which 

 they cannot be separated, and presenting nevertheless an aspect and 

 foliage completely exceptional with decussate leaves. As to the other 

 character cbawn fi-om, the regularity or irregularity of their fiowers, 

 the number of the petals equal to or less than that of the divisions 

 of the calyx, the greater or less excentricity of the gynseceum or 

 its perfectly central insertion, flnally from the symmetry or unsymme- 

 try of th(! disk (which may surround circidarly the central portions of 

 the flower by a rampart equally developed everywhere or more on one 

 side than the other, or which may even be totally wanting on one side 

 and only represented on the other by a sort of crescent, a thick scale, 

 simple or double, and sometimes of considerable dimensions) ; all these 



