AND THEIR CULTURE. 121 



•characters. Here we may trace at least four distinct types, with 

 intermediate gradations : — 1, oblanceolate, broadest above the middle, 

 narrowed gradually to the base, as in Longiflormn, Candidiim, and 

 Japonicum ; 2, obovate-spathulate, broadest about the middle, 

 narrowed suddenly into a distinct claw, as in Bulhiferum, Cafeshcei, 

 and Philadelphicum ; 3, ovate-lanceolate, decidedly broadest below 

 the middle, without a claw, and narrowed gradually in the upper 

 half, as in Speciosiim and Auratum ; 4, lanceolate, nan-owed gTadually 

 from about the middle to both ends, as in Marfagon, Choice donicum , 

 and Pomponium. In the colour of the flower we have two principal 

 series : — 1, unspotted white, liable to be tinged with yellow, green, 

 or purple ; 2, bright shades of red and yellow, very liable to inter- 

 change or mix with one another, spotted usually with dark purple. 

 Only about four of the species will not rank readily in one of these 

 two series. The character of the keel and surface of the inner face 

 of the segments in their lower thirds should be noted carefully in 

 the living plant. Here we have three principal types: — 1, the face 

 not at all papillose and keel not distinctly hollowed out, as in 

 Candidum and its allies ; 2, the face more or less distinctly papillose, 

 but the groove indistinct and glabrous, as in Speciosum, Pomponium, 

 and Martagon- ; and 3, the face papillose, and the groove a distinct 

 hollow, bordered by raised hairy lines, as in Bvlhiferum and 

 Tigrinum. 



Stamens. — The filaments generally place the anther on. a level with 

 the stigma, and fall not far short of the mouth of the flower. Both 

 style and filaments are decidedly shorter than the perianth in the 

 erect-flowered Concolor, and in several of the drooping-flowered 

 species of the Martagon group ; and the stamens are decidedly shorter 

 than the style in Candidum. In the direction of the filaments, and 

 shape of the fully expanded flower, there is a uniform correlation. 

 In the funnel-flowered Lilies all the filaments run parallel with one 

 another and the style towards the mouth of the flower, resting against 

 its lowest divisions, and curving a little upwards towards their points. 

 In the bell-flowered Lilies the style is distinctly curved, and the 

 filaments diverge on all sides fi-om the centre. 



Pistil. — In all the species there is a more or less distinctly 

 clavate ovary. In all, there is a moderately short style, clubbed at 

 the end by a stout, bluntly three-lobed stigma. 



Capsule. — The Capsules, as far as I have seen them, of all the 

 Lilies are quite uniform in general structure. To the question of how 

 far there are differences in shape that may be used for specific cha- 

 racters, I have paid too httle attention to living plants to speak with 

 confidence on this subject. 



The follomng is an outline of what I believe to be the most natural 

 primary division of the genus. I believe that no one who will take 

 the trouble to master the characters of these five groups, and once 



