168 Dr. W. Seller on some Plants obtained 



useful descriptive character. In our specimen it appears in 

 almost all tlie siliculse that have been opened. In DeCandolle's 

 description of the Cruciferae, he mentions, as occasional, the pre- 

 sence of a stria or a rima in the axis of the fruit-septum ; and as 

 far as I have observed, the stria, v^rhich may be regarded as indi- 

 cating the tendency to the rima or fenestra, occurs generally in 

 the species of Cochlearia. This stria or fenestra in the axis seems 

 to suggest the idea that the dissepiment in the fruit of the Cru- 

 cifera is composed of two portions extended from the opposite 

 sides to meet there. And if this be deemed probable, then the 

 conclusion would follow that their seed-vessel is composed, not 

 of two, according to the received view, but of four carpels. As 

 Mr. Brown says he met with one specimen of C. fenestrata in 

 which many of the siliculse were three- valved and three-celled, I 

 was curious to ascertain if any of those in our specimen presented 

 this anomaly, but was disappointed. Mr. Brown does not say 

 how the second dissepiment was placed. It is impossible to sup- 

 pose that there were two dissepiments parallel to each other. I 

 infer then, particularly as Mr. Brown uses the word " dissepi- 

 mentum^^ in the singular number, that the additional septum 

 joined the normal septum in the axis. Mr. Brown^s discovery 

 of three-celled pericarps in a cruciferous plant is an encourage- 

 ment to botanists to search for the farther anomaly of four-celled 

 pericarps among the same ; which can hardly fail to occur, if the 

 theory of their fruit being composed of four carpels or carpellary 

 leaves be correct ; for on this view it must be by abortion that 

 placentae and a septum fail to appear opposite to the cleft of the 

 stigma, at the place in the valves occupied by the carina, when 

 that is present in this kind of fruit. Mr. Brown also remarks, 

 in his description of the C. fenestrata, that the umbilical cords 

 are joined together at their bases by a narrow membrane. This 

 narrow membrane farther iUustrates the structure of the fruit in 

 the CrucifercB. It represents the margin of the interior layer of 

 the carpellary leaf stopping short close to the inner side of the 

 middle rib, which here enters into the replum or frame of the 

 dissepiment, while the dissepiment itself is composed of the outer 

 layer joined with its fellow of the adjacent carpellary leaf and ex- 

 tended to the axis. This accords in so far with DeCandoUe^s 

 account of the structure of the septum in the Cruciferce, though 

 he describes it in different terms ; he says the septum is formed 

 by the reflexion inwards of the epicarps, while the endocarps stop 

 short close to the suture and produce the placentse. But if there 

 be four such shortened borders of the endocarp, two at each 

 margin of the septum, as in all siliculse with a double row of 

 seeds in each cell, then there must have been four original car- 

 pellary leaves, two entering into each valve, and two into each 



