212 THE STRUCTUEE OF FLOWERS. 



When the short or mid-stjled forms were illegitimately 

 crossed bj the long-styled, then the two parent forms alone 

 were reproduced, but in no case did the third form appear. 



When, however, the mid-styled form was illegitimately 

 fertilised by the longest stamens of the short-styled, the seed- 

 lings consisted of all three forms. Tbis illegitimate union was 

 noticed as being singularly fertile, and the seedlings themselves 

 exhibited no signs of sterility, but grew to the full height. 



Finally, of the three forms, the long-styled evinces some- 

 what the strongest tendency to reappear amongst the off- 

 spring, whether both, or one, or neither of the parents are 

 long-styled. 



Although L. Salicaria has not, as far as I know, shown 

 any signs of variability in the lengths of its filaments and 

 styles, yet, as is perhaps generally the case with heterostyled 

 plants, there are one or more species of the same genus which 

 are normally homostyled. Thus L. hyssopifolium, which is 

 not social, and is a dwarf form and an annual, bears only six 

 to nine stamens, the anthers of which surround the stigma, 

 which is included within the calyx. The three stamens, 

 which vary in being present or absent, correspond with the 

 six shorter stamens of L. Salicaria. The stigma and anthers 

 are upturned as in the last species, and so indicate the fact 

 that it is a degenerate form from L. Salicaria or some other 

 intercrossing species, though it has now reacquired its self- 

 fertilising properties. Oxalis is a genus having trimorphic 

 species. Many of them are extremely infertile with their 

 " own form " pollen. Such are the long-styled form of 0. 

 tetraphylla, versicolor, Brasiliensis, and co7}ipressa. On the 

 other hand, in the long-styled form of 0. incarnata, rosea, 

 and Piottce, and in the mid-styled form of 0. carnosa, no self- 

 sterility occurs.* 



* According to Hildebrand, Bot. Zcitg., xlv., pp. 1, 17, 33. 



