Philosophical Character of DecandoUe. 223 



lar descriptions of species which are contained in the Prodro- 

 mus and the general observations on the structure of plants 

 considered in the aggregate, which are found in the Organo- 

 gi-aphie ; constituting the groundwork of the former, and the 

 data upon which the latter was constructed. 



Thus, in his Memoir on the Crucifer^, he carries us in de- 

 tail through the structure of all the parts, first, of vegetation, 

 and afterwards of reproduction, belonging to this important 

 natural family ; and he shews, that the only distmction which 

 can be relied on for separating its members into natural groups, 

 are drawn, either from the form of the embryo, or from that 

 of the seed-vessel. If we adopt the former as the basis of our 

 system, we shall divide the Cruciferae into five natural groups, 

 according to the position of the Radicle with reference to the 

 Cotyledons; if Ave adopt the latter, we shall distinguish them 

 into six, according to the position of the valves of the Seed- 



This latter method he shews to be preferable to the old 

 Linna^an division, depending upon the length of the pod, as the 

 latter admits of no exact limits, and as it placestogether gene- 

 ra in no way allied, and divides others which are naturally con- 

 nected • but he nevertheless regards it as of inferior moment 

 to the distinction founded upon the embryo, both because the 

 latter is an organ of greater importance than the seed-vessel, 

 and because there is not such a gradation in its form, as is 

 found in that of the pod which incloses it. 



He adopts, therefore, as the basis of his classification, the 

 principle suggested by Robert Brown, with respect to the 

 manner in which the radicle is folded upon the cotyledons, 

 and afterwards subdivides the groups so formed accordmg to 

 the form and mode of opening of the seed-vessel. 



He thus, by means of these two characters, constructs 

 twenty-one natural groups, and satisfies himself of the cor- 

 rectness of the principles upon which he has proceeded m his 

 classification, by finding that the genera thrown together by 

 virtue of this arrangement, are really such as stand most nearly 

 allied one to the other. 



Thus, as in the physical sciences, we commence by making 



