324 THE JOURNAL OF BOTANY 



Let us take these characters seriatim. So far as I am able to 

 judge from our dried specimens, the rootstock character is not a 

 constant one ; plants which from other points would be classed 

 under T. sessile show a horizontal rootstock comparable with that 

 of Underivoodii. 



The size of the plant is very variable ; that of Catesby's plant 

 {i. e. type of U^iderwoodii) closely resembles those of the Gronovian 

 and Plukenet specimens. On the other hand, of two specimens 

 collected at Lexington, Kentucky, by Dr. Short, which differ in 

 nothing but size, and would undoubtedly be considered as T. sessile 

 in the most limited sense, the stem in one case is 8 cm. long by 

 3 mm. wide, in the other 26 cm. by 6-7 mm. wide. Cultivated 

 specimens also often show remarkable differences in size, though 

 alike in other respects. 



The leaves of all the three historic specimens are similar ; the 

 difference in shape implied by Catesby's figure is not borne out by 

 his specimen, the leaves of which fall within the limits of size 

 given for T. sessile. Moreover, the leaves of the Gronovian plant 

 still show distinct traces of having been mottled, and the same 

 character is obvious in Plukeuet's figure. 



The sepals of Catesby's plant (3-5 cm. long), though larger than 

 in the other two specimens, do not reach the lower limit of length 

 assigned to Underwoodii. The petals (5 cm. long) are, however, 

 conspicuously longer than the sepals, though they also fall short of 

 the lower limit assigned to T. Underwoodii. The stamens are scarcely 

 more than one-third the length of the petals, while in T. Ufiderwoodii 

 the proportion is about 1 to 2. 



The length and form of the filaments is similar in all three 

 specimens, but the anthers in Catesby's plant are one-third longer 

 than those of the other two. 



The specimens which I have examined do not show any longi- 

 tudinal demarcation of style and stigma. The ovary appendages 

 are similar in all, consisting of erect or slightly apically recurving 

 processes which are apparently stigmatic down the whole of the 

 inner face. Those in Catesby's plant are intermediate in length 

 between those in the other two specimens. 



Therefore, if we restrict our observations to the type specimens, 

 it is evident that although many of the characters indicated by Mr. 

 Small are not confirmed, yet that the Carolina plant is distinguished 

 from the Virginian by its larger sepals, by its narrowly oblanceolate 

 petals which are also longer both actually and relatively to the 

 sepals and stamens, and by its longer anthers. 



It is possible to arrange a limited number of herbarium speci- 

 mens in a series of more or less distinct forms. Thus dealing with 

 specimens from the Atlantic slope, we have: — 1. What we may 

 regard as typical T. sessile conforming to Clayton's Virginian 

 specimen and that of Plukenet, with lanceolate to oval-lanceolate 

 petals about equal in length to the sepals and generally about twice 

 the length of the stamens. The plants seem, as a rule, smaller and 

 less robust than the other forms ; the leaves are bluntly ovate to 

 orbicular. 



