Chap. vni. OXALIS. 323 



bases of the shorter stamens. Their 'stigmas are much 

 less papillose, and smaller in about the ratio of 13 to 

 20 divisions of the micrometer, as measured trans- 

 versely from apex to apex, than the stigmas of the 

 perfect flowers. The styles are furrowed longitudinally, 

 and are clothed with simple as well as glandular hairs, 

 but only in the cleistogamic flowers produced by the 

 long-styled and mid-styled forms. The anthers of the 

 longer stamens are a little smaller than the correspond- 

 ing ones of the perfect flowers, in about the ratio of 

 11 to 14. They dehisce properly, but do not appear 

 to contain much pollen. Many pollen-grains were 

 attached by short tubes to the stigmas ; but many 

 others, still adhering to the anthers, had emitted 

 their tubes to a considerable length, without having 

 come in contact with the stigmas. Living plants 

 ought to be examined, as the stigmas, at least of the 

 long-styled form, project beyond the calyx, and if 

 visited by insects (which, however, is very improbable) 

 might be fertilised with pollen from a perfect flower. 

 The most singular fact about the present species is 

 that long-styled cleistoga.mic flowers are produced by 

 the long-styled plants, and mid-styled as well as 

 short-styled cleistogamic flowers by the other two 

 forms ; so that there are three kinds of cleistogamic 

 and three kinds of perfect flowers produced by this 

 one species ! Most of the heterostyled species of 

 Oxalis are more or less sterile, many absolutely so, if 

 illegitimately fertilised with their own-form pollen. 

 It is therefore probable that the pollen of the cleisto- 

 gamic flowers has been modified in power, so as to act 

 on their own stigmas, for they yield an abundance of 

 seeds. "We may perhaps account for the cleistogamic 

 flowers consisting of the three forms, through the prin- 

 ciple of correlated growth, by which the cleistogamic 



