BOTANICAL .SUBJECT^. 



235 



I'ig. yo. 8tameus of Polygala vulgaris, with onc-cellfd anthers 

 (Le Maout and Decaisne). 



margins. This is exactly wliat one would expect if the foregoing 

 theory had any basis of truth. Fig. 91 shows a carpellized stamen 



Fig. 9 1 . Carpellized stamen of Cheiranthus, after P. Brousse 

 (" Etudes des Fmits"), pi. S, fig. 3. 



of Cheiranthus. If the cherry petiole gland be the fusion of the 

 teeth-glands of a supj^ressed leaflet, and if it be homologous with 

 an anther-cell,* then it would follow that sometimes the anther- 

 cell would declare its nature by reversion to a leaf structure, vnXii 

 the re-appearance of the teeth-glands as ovules. The above fio-nre 

 shows that this does occur. Several cases of ovuliferous anthers 

 are given in Dr. Masters' " Teratology. "I 



We see that this route of analvzincr the anther, that is, through 

 the glands on the cherry-petiole, seems to lead to precisely the .'<ame 

 conclusion as that of analyzing it through the sori of ferns. Then 

 if this view of the anther be correct, we wotdd have the glands on 



♦Curiously enough the anther cells of Adhatoda betonica are on different 

 levels, as in cherry glands. 



•f In a case of Pcponia moutan (very double) an imperfect carpel had 

 ovules on one margin and an anther on the other margin. 



