260 Mr. J. Miers on the Bignoniacese. 



whether it arises from the existence of an arillus round each 

 seed : if the latter be the case^ as is very probable, it would 

 offer a good discriminating character between the Crescentiacea 

 and Bignoniacea ; for no trace of any arillus has yet been ob- 

 served in the latter family ; otherwise there is little real distinc- 

 tion between the two orders. The floral characters in all the 

 Crescentiacece are similar to those of the Bignoniacece ; and there 

 is no essential difference in the structure of their exalbuminous 

 seeds, for it has been shown that the presence or absence of a mem- 

 branaceous wing no longer offers any line of distinction between 

 the two families. Setting aside the yet uncertain question of an 

 arillus, the claims of the Crescentiacece to an ordinal rank are 

 feeble, being reduced simply to the presence of pulp and the 

 indehiscence of the fruit. If these claims should be considered 

 of insufficient value, this small group, without inconsistency, 

 might still be retained, after the example of DeCandolle, merely 

 as a tribe of the Bignoniacece. 



The group of the Cyrtandracece has been considered by many 

 of the most eminent botanists as a tribe of the Gesneriacece, 

 among whom are Mr. Robert Brown (PI. Jav. Rar. 105), Prof. 

 Endhcher (Gen. PL 716), and Mr. Benthara (Lond. Journ. Bot. 

 v. 360). On the other hand. Prof. Lindley (Introd. 283) and 

 Prof. DeCandolle (Prodr. ix. 258) regard it as a distinct family, 

 more allied to Bignoniacece. It differs from Gesneriacece in its 

 perfectly and constantly superior ovary and its exalbuminous 

 seeds : it accords with Bignoniacece in the form of its calyx and 

 corolla ; in its stamens being often didynamous, when frequently 

 only two of them are fertile, as in Catalpa; in its anthers being 

 more or less divaricated at their base, and united at their summit 

 by a connective, which forms an apicular excurrent appendage ; 

 and in its free 2-locular ovary, seated within a fleshy disk. It 

 differs from Bignuniacece in the species being for the most part 

 herbaceous, with alternate or radical leaves, which are never 

 strictly pinnate ; in its prominent bifurcate placcntations, which 

 fi'equently produce the appearance of four cells in the fruit ; in 

 its very numerous minute and always pendent seeds, and its 

 terete radicle with very small cotyledons. These characters are 

 sufiicient to establish its claim to a distinct ordinal rank. The 

 ovary in Cyi^tandracece appears to be composed of two carpellary 

 leaves, placentiferous on their midribs and conjoined by their 

 sterile margins, as in Jacaranda and Crescentia; it is conse- 

 quently unilocular, with two opposite parietal pla- Fig. 17. 

 centse, which severally bifurcate (fig. 17). By the sub- 

 sequent growth and approximation of the placentfe,the 

 2-valved capsular fruit frequently becomes spuriously 

 2-locular or falsely -1-celled : the generally comose extremities 



