352 MR. J. G. BAKER ON LILIACEJE. 



under the superintendence of Dr. Harvey, which contains many 

 unpublished species. 



The section of the order dealt with in the present paper is one 

 that is circumscribed with tolerable definiteness. I have not 

 thought it needful to eiiter upon any points connected with the 

 general characterization and relationships of the order, because 

 I believe that these may be regarded as settled quite satisfac- 

 torily, and that any thing which I could say would only go to 

 confirm what botanists appear to be already fully agreed upon. 

 The great body of Endogens with a corolla-like perianth, with 

 divisions in sixes, and stamens in sixes or threes, falls into two 

 alliances, characterized, the one by a superior, and the other by 

 an inferior ovary and fruit, so as to leave very few of its genera 

 in a position at all doubtful or intermediate. In subdividing 

 the former alliance, we get characters, which are universally re- 

 garded as of ordinal value, in the extrorse anthers, separated styles, 

 and septicidally dehiscent capsules of Colchicacese. Deducting 

 also Pontederiaceje, a small order of not more than thirty species, 

 in which the perianth is twisted in aestivation, there remam con- 

 siderably over 1000 species, the great bulk of which agree 

 closely in all important points of structure. It seems to be 

 most natural to regard all these plants as constituting a single 

 order, and to arrange the great bulk of the species in two series 

 in one of which the fruit is a berry, and in the other a capsule. 

 In these two series the great bulk of the order will readily rank, 

 and there will remain over only a few exceptional groups, each 

 containing a few species only, such as Hoxhicrghia^ Lapageria, 

 and Fhilesia, in which the ovary is 1-celled, with parietal phi- 

 ceutation, Conantherse, which approximate to Amaryllidacese by 

 their partially adherent ovaries, Uvularieae, which approximate 

 to Colchicacese by their extrorse anthers, Paridea?, vvhich approxi- 

 mate to Colchicaceaj by their disunited styles, and a few others. 

 Taking the capsular series of genera, the most natural and con- 

 venient method is, I think, to dispose them in two subseries, one 

 characterized by having the segments of the perianth free from 

 one another down to the very base, and the other by having 

 them joined together for at least the lower tliird or quarter. If 

 we follow this method there is no difficulty in deciding clearly m 

 which of the two subseries each genus will fall ; and if we use for 

 tribal characters the general arrangement of the inflorescence and 

 the nature of the root-stock, the tribes of the two subseries will, 



1 



