350 MB. J, O. BA.KER ON LTLTACEjE. 



now dealing. As 1 have lately liad occasion to show elsewhere, 

 between the extreme points of Scilla, as that genus is (as I 

 believe rightly) constituted in the ' Enumeratio/ four other ge- 

 nera which the author has adopted, must be i)laced, viz. Bar- 

 nardia, Ledehourta, Eratolotrys^ and Drimia, About half the 

 species placed by Kunth in Scilla have two, and the other half 

 have several ovules in each of the three cells of the ovary. 

 Lindley separated, under the name of Barnardia, two Asiatic 

 species, which only differ structurally from the plants just re- 

 ferred to by having only a single ovule in each of the cells. Tliis 

 character seems quite insufficient to found a genus upon, and he 

 either overlooked or did not know that a long-known species of 

 Scilla from Barbary (parvijlora of Desfontaiues) is also uniovulate. 

 Kunth, however, keeps up Barnardia^ but retains the Barbary 

 plant in Scilluy and in consequence this latter has since been 

 made into a monotypic genus by Steinheil under the name of 

 Stellaris. The other three genera quite coincide with one another 

 in structure, and only differ geographically, Ledehouria being an 

 inhabitant of India, Bratobotrys of Nubia and j^byssinia, and Dri- 

 mia^ as Kunth defines it, of the Cape of Good Hope. In the typical 

 species of Scilla the divisions of the perianth spread from the very 

 base when the flower is fully expanded; but in these plants, as in 

 the Common Wild Hyacinth of our English woods (the Scilla nu- 

 tans of Smith and Kunth), the divisions, though not properly con- 

 nate at the base, as they are in the cultivated Hyacinthiis orien- 

 talis, yet cohere permanently in a cup, and spread only for the 

 upper half or two-thirds. The intermediate gradation between 

 these two shapes of flower may be easily studied in Scilla cam- 

 pannlataj which is a South-European subspecies of nutans very 

 common in the gardens round London. But this is rather a di- 

 gression from the main question. I believe that no one who 

 has at all attended to the order will feel any doubt that a 

 thorough revision both of its genera and species is needed, di- 

 rected with a view to ascertain, from the consideration of all the 

 species which are now known, what are the best limitations and 

 diagnostic characters of the former, and to bring together the 

 species in one view, and define them more explicitly upon a 

 uniform plan. This is what, in the present paper, I have at- 

 tempted to do for a section of the order as fully as the material 

 at my command would permit. Liliacese is an order in which, 



Ml a orpnpral rnlft. the difttiiirtion hpf\rppn nllipd sriPcieS cannot 



