BANUNCULACE^. 63 



petals. Not one of these appendages is truly alternate with any 

 of the four valvate sepals of the calyx. As we approach the centre 

 of the fiovver, we find the kind of spathulate expansion at their summit 

 gradually transformed into a connective, bearing on its inner face 

 two anther cells dehiscing longitudinally. At the same time the 

 basilar portion becomes contracted to form a true filament. The 

 " petals" of Atrageiie are therefore only staminodes.' Besides, the 

 numerous carpels of Atragene have this in common with those of 

 Clematis, that the ovule observed on the inner angle of the ovary 

 has above it four small ovules in two vertical rows, which never 

 become developed.^ 



The genus Naravelia? was estabHshed for certain Indian species of 

 Airagene, whose leaves, instead of being three-lobed, have the middle 

 segment abortive, or transformed into a tendril. If we analyse 

 N. zeylanica' we find a pubescent calyx of four, five, or six valvate 

 sepals, and within this a large number of imbricated stamens, with 

 flattened filaments, and two-celled introrse anthers,' dehiscing 

 longitudinally, each surmounted by a small prolongation of the 

 connective. The carpels are numerous, covered with stiff", erect 

 hairs ; and each encloses a suspended ovule with the micropyle 

 turned upwards and inwards. This is the only one completely 

 developed; but above it, when young, are four others in pairs on 

 each side of the suture of the carpel, of which traces are with 

 difficulty found when the fertile one is adult. This is another rela- 

 tion with Atragene, of which Naravelia also possesses the corolla 

 formed of very long petals dilated at the summit and varying in 

 number.^ But these are only sterile stamens. For a long time 

 they are very small, shaped like the outer stamens, and presenting 

 above an anther- like sweUing, which never becomes fertile. Hence 

 we cannot separate Atragene and Naravelia generically from each 

 other, and therefore not from Clematis. 



' We must unite these types into one genus, existence of these small sterile ovules. Even 



for the same reason as we include Pulsatilla five or six may be found, 

 in Anemone, and because, as we shall soon ' Xaravael Heem., Zeylan., 26. 



find, we cannot separate from the true Hib- * N. zeylanica DC, Prodr., i. 10. — Hook. & 



berfias those species in which the outer sta- Th., Fl. Ind., i. 3. — B. H., Gen., 4,.— Atragene 



mens are transformed into staminodes, &c. zeylanica L., Amfen., i. 405. 

 In several cultivated species of Clematis the " Endlicheb {Gen., n. 4470) believed these 



flower becomes double, like that of the Ane- stamens to be extrorse, which would have re- 



mones. moved these plants further from Atragene. But 



"■* Here also it is the study of organogeny in N. zeylanica they are distinctly introrse. 

 which revealed to us (Adansonia, i. 33 i) the '' From five to fifteen. 



