APPENDIX. 3j 



evidence has been given to prove, as I have remarked in regard 

 to AnthocerciSj that neither the structure of the anther, nor even 

 the didynamous character of the stamens, are by themselves 

 sufficient to retain a doubtful genus within the limits of the 

 Scrap huIariacecE. The fact, in combuiation with other circum- 

 stances, that manifestly determines the position of Anthotroche 

 among the AtropacetB^ rather than in Scropltulariacecey is the 

 structure of the seed ; this I'eserables that of Anthocercisy CypliaU' 

 thera and Duboisia ; it is cylindrical, slightly curved, containing, 

 enveloped in albumen, a nearly straight cylindrical embryo, with 

 minute cotyledons, and a terete radicle pointing to the base of the 

 seed, the hilum being on the lateral ventral sinus. Whatever may 

 be its true position, one conclusion is certain, that where Antho- 

 €€7'cis is placed, Anthotroche must accompany it. This agreement 

 is evident, not only in their similarly extrorse stamens, but in 

 the aestivation of the corolla ; this last feature is not so easily 

 distinguished in the last-mentioned genus, on account of the 

 shortness of the lobes of its border, and their being densely 

 covered on both sides with long branching hairs, which, being 

 closely interlaced, conceal the margins. It differs however from 

 Anthocercis and Cyphanthera in the more expanded and almost 

 rotate form of its corolla, its shorter and more obtuse lobes, and 

 in its stamens being all perfect and equal in number to the seg- 

 ments of the border, although one of them is somewhat shorter. 

 The peculiarity in the form of its anther consists in its being, 

 like that of Cyphanthera and Duboisia, roundly discoid and deeply 

 reniform ; its attachment to the filament is in the deep sinus 

 upon the side looking toward the style, while the external face 



runs 



open 



almost peltate form, by its two nearly equal gapmg valves, 

 showino- a globular prominence in the middle, which is the re- 

 ceptacle of the pollen. This structure is the result, as I have 

 already shown {app. 23),of thetotal abortion of one of the lobes that 

 ordinarily constitute an anther, and not, as generally supposed, 

 of the confluence of the two lobes. Anthotroche offers another 

 distinguishing feature, in its perfectly free cup-shapcd disk, that 

 invests the base of the ovary : this must not be confounded with 

 a somew^hat similar appearance in Anthocercis and Duboisia, 

 where the corolla breaks away by a circumscissile line, leaving a 

 membranaceous cup surrounding the base of the ovary, as m 

 Lycium, Fabiana, Nierembergia, Sessea, Vestia, Cestrum and some 

 others: the disk, although also existing in Anthocercis ^^n^ Du- 

 boisia, is not free to the base, as in Anthotroche, but is wholly 

 adnate and often inconspicuous, its obsolete lobes being some- 



times almost free. 



lyx 



f2 



