320 CLBISTOGAMIC FLOWERS. Chap. VIII. 



are thus provided. Lastly, in V. canina never more 

 than two of the stamens, as far as I have seen, bear 

 anthers; the petals are much more reduced than in V. 

 hirta, and according to D. Miiller are sometimes quite 

 absent. 



Oxalis acetosella. — The existence of cleistogamic 

 flowers on this plant was discovered by Michalet.* 

 They have been fully described by Von Mohl, and I 

 can add hardly anything to his description. In my 

 specimens the anthers of the five longer stamens were 

 nearly on a level with the stigmas; whilst the smaller 

 and less plainly bilobed anthers of the five shorter 

 stamens stood considerably T below the stigmas, so that 

 their tubes had to traveL some way upwards. Ac- 

 cording to Michalet iiiese latter anthers are sometimes 

 quite aborted. In one case the tubes, which ended in 

 excessively fine points, were seen by me stretching up- 

 wards from the lower anthers towards the stigmas, 

 which they had not as yet- reached. My plants grew 

 in pots, and long after the perfect flowers had with- 

 ered they produced not only cleistogamic but a few 

 minute open flowers, which were in an intermediate con- 

 dition between the two kinds. In one of these the pol- 

 len-tubes from the lower anthers had reached the stig- 

 mas, though the flower was open. The footstalks of 

 the cleistogamic flowers are much shorter than those 

 of the perfect flowers, and are so much bowed down- 

 wards that they tend, according to Von Mohl, to bury 

 themselves in the moss and dead leaves on the ground. 

 Michalet also says that they are often hypogean. In 

 order to ascertain the number of seeds produced by these 

 flowers, I marked eight of them; two failed, one cast 

 its seed abroad, and the remaining five contained on an 



• ' Bull. Soc. Bot. de Fiance,' torn. vii. 1860, p. 468. 



