K Barlow 55 



The fact that the two sets of pollen on one plant behave differently 

 towards the same female shows a segregation in the individual of the 

 factor (or absence of factor) causing incompatibility'. This constitutes a 

 marked difference between heterostj-led plants and such self-sterile 

 species as Linaria vulgans (de Vries (4)) and Cardamine (Correns (5)). 



Oxalis valdiviana was selected for experiment, owing to its annual 

 habit and its long flowering period. This plant has a very high 

 degree of self-sterility, and it was found difficult to proceed with it 

 alone. Lythrum salicaria has been found to be more self-fertile 

 (C. D., Forms of Flowers) ; also the normal capsules set over 80 seeds, 

 whereas in Oxalis there are onl}' from 10 to 1.5. Experiments were 

 started on this species in 1911 so that we have as yet only the F^. 



The plants were obtained from the Cambridge Botanic Gardens, and 

 from wild seed. Two plants of the variety rosea, growing in the Chelsea 

 Physic Garden, were also used. 



Oxalis valdiviana. 



Procedure. 



In 1907 four long-, four mid-, and four short-styled plants were 

 chosen from self-sown garden plants. All families subsequently raised 

 are derived from these twelve original parents. 



During the summers of 1907, 1908 and 1909 the plants were kept 

 under wire cages covered with muslin ; or each flower-head was sepa- 

 rately covered with a paper bag. No crosses were made in 1910. In 

 1911 and in 1912 the plants used were either isolated in bee-proof 

 greenhouse partitions or covered with paper bags. Bees visit this 

 plant very freely. It was sometimes difficult to frustrate their per- 

 sistent attempts to get at flowers temporarily uncovered for artificial 

 fertilization. Insect visits may be a cause of irregularity in the 

 results, and will be referred to later, with other possible causes of 

 error. 



Owing to the violent expulsion of ripe seed, each capsule has to be 

 separately covered with fine muslin before it reaches maturity. 



The immature anthers were removed from the females used for 

 crossing in 1907. In subsequent years castration was not thought 



' A flower of the short-styled Oxalis valdiviana was observed to have one of the mid 

 length tier of stamens at the level of the long stamens. The pollen from this aberrant 

 stamen was carefully tested on a long- and on a mid-styled plant, and was found to behave 

 exactly as "long" pollen. The length factor of the stamen is bound up with the inhibitory 

 effect of the pollen ; the abnormal growth of this stamen had brought with it the inhibition 

 of its pollen for a mid-styled plant. 



