ANDROSTEPHIUM VIOLACEUM. CROWNED LILY. 171 



differs in the very short tube of the perianth (the lower portion of 

 the flower), in the tube of filaments having only a short tooth 

 between the filaments, and in the form of the capsule." But in 

 some of the allied genera the length of the tube would not be 

 of much consideration. In some unquestioned Millas, for in- 

 stance, the "tube of the perianth" is three-fourths the length of 

 the whole flower, while in other cases it is no longer than we find 

 in the subject of our present chapter. We thus see how difii- 

 cult it will be for the student to decide on the genera of these 

 Liliaceous plants, when he collects them for the first time. As a 

 general rule the union or the separation of the parts is regarded 

 as among the best characters. The theoretical structure of a 

 lily flower is to have three sepals (the usual calyx), three petals 

 (the corolla), and a similar series of two sets of three, resulting in 

 six stamens and three pistils ; and it is chiefly from the manner 

 in which these various parts are united or developed in proportion 

 to one another, that characters to distinguish the various genera 

 are found. Sometimes, as in the ordinary lily flower, the apex 

 of the pistil is divided into three distinct parts, but in our plant 

 as we see in Fig. 2, the apex is inclined to be capitate, or termi- 

 nating in a little pin-like head. This, though there were no other 

 characters, would at once suggest to the student that it was not 

 a Liliiun. Then there may be characters drawn from the 

 phases of growth, which, however, are not often referred to in 

 botanical works, because so much has to be derived from dried 

 specimens wherein these life-characters cannot be observed. In 

 many plants the stamens and pistils finish their growth at or 

 about the expansion of the corolla; but we see in our plant that 

 the pistil is nearly complete long before the stamens, which do 

 not take on their peculiar crown-like form until the segments of 

 the " perianth " have fully expanded. 



Again, the roots of these plants deserve more study for botan- 

 ical characters than they have received. In our specimen there 

 were several buds (Fig. 3), which seemed to indicate that new 

 corms may be formed by offsets. And then at the base of the 



