STRUCTURE.] OP AMARANTHS. 43 



furrow, which renders the seed slightly reniform. A white, 

 circular membrane, evidently arising from the umbilical cord, 

 is found fixed round the point of attachment at the bottom of 

 the furrow ; passing beyond the latter the membrane covers 

 the micropyle, but extends a very little on the surface of the 

 testa. Here then again we have a true aril ; but we may be 

 surprised to find this envelope in a genus lost, as it were, 

 amongst a crowd of others which have no trace of this organ. 

 In the anatropal and amphitropal seeds, to which our 

 attention has as yet been directed, the distinction between 

 the aril and proper exterior integument is quite plain. The 

 latter, as I have before said, a true ovulary leaf, traversed by 

 the raphe, with the nervures depending on it, forms a com- 

 plete envelope, the orifice of which, scarcely visible (micropyle) 

 is more or less near the hilum. The aril, on the other hand, 

 often reduced to the dimensions of a rim or of an unilateral 

 tongue, never has any nervures ; even when extending over 

 the seed as a hood or sac, it presents a large opening on 

 the side next the chalaza, opposite the micropyle. The same 

 distinctive characters can be easily applied to campylotropal 

 seeds, in which, instead of a raphe, there generally exists a 

 vascular network in the external integument, and the micro- 

 pyle is always close to the point of attachment. But these 

 characters, though, when combined, of the utmost import- 

 ance, have not taken separately the same comparative value 

 between themselves. The position of the micropyle, generally 

 determined by that of the radicle, furnishes a very steady 

 character, when compared with the inverse direction of the 

 opening in the aril. The presence of the raphe or of nervures 

 in the proper integument is far from being so constant, and 

 there exists a great number of seeds which have only one 

 entirely cellular integument, whether it proceeds from the 

 secundine, or from the very thin nucleus. If, in such seeds, 

 we suspected an arillary envelope, the position of the micro- 

 pyle or of the radicle, or, better still, the observation of their 

 development, would resolve all doubts. But if we found a 

 cellular envelope of doubtful nature, in an orthotropal seed 

 with only one cellular integument, we could not then decide 

 with certainty as to its nature, and we ought to remember 



