178 HETEROSTYLED TRIMORPHIC PLANTS. Chap. IV. 



mid-length stamens, produced capsules, and these 

 contained twice as many seeds as did the flowers 

 fertilised with pollen from the shortest stamens. 

 It thus appears (and we find some evidence of 

 the same fact with 0. speciosa) that the same rule 

 holds good with Oxalis *as with Ly thrum salicaria; 

 namely, that in any two unions, the greater the in- 

 equality in length between the pistils and stamens, or, 

 which is the same thing, the greater the distance of 

 the stigma from the anthers, the pollen of which is 

 used for fertilisation, the less fertile is the union,- 

 whether judged by the proportion of flowers which 

 set capsules, or by the average number of seeds per 

 capsule. The rule cannot be explained in this case 

 any more than in trfat of Ly thrum, by supposing that 

 wherever there is greater liability to self-fertilisation, 

 this is cheeked by the union being rendered more ster- 

 ile; for exactly the reverse occurs, the liability to self- 

 fertilisation being greatest in the unions between the 

 pistils and stamens which approach each other the 

 nearest, and these are the more fertile. I may add that 

 I also possessed some long-styled plants of this species : 

 one was covered by a net, and it set spontaneously a 

 few capsules, though extremely few compared with those 

 produced by a plant growing by itself, but exposed to 

 the visits of bees. 



With most of the species of Oxalis the short-styled 

 form seems to be the most sterile of the three forms, 

 when these are illegitimately fertilised; and I will add 

 two other cases to those already given. I fertilised 

 29 short-styled flowers of 0. compressa with pollen from 

 their own two sets of stamens (the pollen-grains of 

 which differ in diameter as 100 to 83), and not one 

 produced a capsule. I formerly cultivated during 

 several years the short-styled form of a species pur- 



