272 OBSERVATIONS ON THE NATURAL FAMILY 



In the mean time it may give some plausibility to the 

 hypothesis to remark, that there are families of plants 

 strictly natural in which a series of degradations exist, if I 

 may so speak, from the assumed perfect pistillum, to a 

 structure as simple as that of Compositae. 



Thus in Protcacece we have the type of the perfect pis- 

 tillum in the many-seeded folliculus of Embothrium ; the 

 first degree of imperfection in that of Grevillea, where only 

 one ovuhnn of each series remains ; a further reduction in 

 the indehiscent monospermous fruit of Zeucospermum, in 

 which the insertion of the ovulum is lateral ; and the sim- 

 9i] plest form in Protect itself, where the single ovulum is 

 inserted at the base of the cavity. Proteaceae, however, 

 exhibit a series of obliterations in the parts of a single pis- 

 t ilium only. An illustration more in point, though some- 

 what less perfect as a series, may be taken from Goodenovia, 

 an order of plants very nearly related to the class of which 

 we are treating. In the greater part of Goodenovia. the 

 ovarium is bilocular, each cell having an indefinite number 

 of seeds ; in the greater number of Sc<evolce, each cell is 

 reduced to a single ovulum ; while in some species of the 

 same genus, and in all the species of Dampiera, the ova- 

 rium, though retaining its external characters, is reduced 

 to a single monospermous cell, with an erect ovulum, as in 

 Compositae. The natural order Crvciferce exhibits also 

 obliterations, more obviously analogous to those assumed 

 as taking place in syngenesious plants ; namely from a 

 bilocular ovarium with two polyspermous parietal placentae, 

 which is the usual structure of the order, to that of Isatis, 

 where a single ovulum is pendulous from the apex of the 

 unilocular ovarium. And lastly in the genus Bocconia, in 

 the original species of which {B . frutescens) the insertion of 

 the single erect ovulum has the same relation to its parietal 

 placentae, as that of Compositae has to its filiform cords, a 

 second species {B. cor data) exists in which these placentae 

 are polyspermous. 



My sixth observation on Compositae regards the order in 

 which the florets expand. To understand the relation this 

 order has to that of other families, it may be necessary first 



