480 



must be aware of the close similarity between Scirpus palustris, multi- 

 caulis, pauciflorus, caespitosa and acicularis. There is also a strong 

 general similarity between Scirpus maritimus, sylvaticus, triqueter 

 and lacustris, which resemble each other much more than they resem- 

 ble any of the first-mentioned five species. Nevertheless, in Smith's 

 'English Flora' we find the four last united with two of the others (pau- 

 ciflorus and ca3spitosus) into one genus, Scirpus, the remaining three 

 retaining the generic name Eleocharis, under which Hooker includes 

 all the five. This disjunction of species closely alike one another, and 

 conjunction of two of them with species to which they have much 

 less resemblance, arises from strict adherence to a singly selected 

 character, namely, the base of the style being slender or dilated. 



The sectional or the generic characters among the Cruciferse, Um- 

 bellifera?, Composite and other orders, are too frequently founded 

 upon minute characters, difiicult to observe, and leading, after all, to 

 disjunction of resembling plants, and conjunction of those which are 

 less similar. This is strikingly shown in the small group of Schizo- 

 petaleae. " Nature " says Barneoud, alluding to A. P. De CandoUe's 

 embryonal classification of the Cruciferae, " appears to have created 

 the group of the Schizopetaleae to prove how little stable are frequent- 

 1}^ the majority of those sections or subdivisions of family which are 

 not founded upon a totality of characters of aflinity, as the true natu- 

 ral method requires. In the herbarium from Chili we find six species 

 of Schizopetalon, of which five are new. If we study these plants 

 with care before dissecting the seed, we are led to arrange them all in 

 the same genus ; all have a perfect similitude in the various organs of 

 the flower, the same aspect, and nearly the same habit ; in a word, we 

 find an almost uniform plan of generic structure. The anatomy of 

 the seed then demonstrates a considerable diff'erence between several 

 of the species. We find on the one hand, verj^ minute globular seeds 

 presenting an embryo with four linear and spiral cotyledons, with 

 curved radicle, evidently belonging to the Spirolobeae of De Can- 

 dolle ; and on the other, oval seeds, larger than the preceding, their 

 embryo with two incumbent, spathulate cotyledons, and with an al- 

 most straight dorsal radicle, evidently belonging to the section Noto- 

 rhizeae. This is the most striking character of the new genus Perrey- 

 mondia. Now it is quite plain that it is impossible to separate, 

 without violating the laws of natural affinity, in the methodical dis- 

 tribution of the Cruciferous plants, these two genera (Schizopetalon 

 and Perreymondia), so nearly related, and solely distinct as respects 

 the embryo, as it would be necessary to do according to the classifi- 



