26 



crest lies approximately in the plane of the peduncle, while in the 

 fruits here under consideration this plane passes approximately 

 between the two crests of the double fruit, through two of the 

 four parietal placentae, as indicated by the dotted line. 



The armature is similar to that of the normal fruits. In both 

 of the examples observed, the crests lie to either side of the up- 

 permost section and the plane of the peduncle. 



The determination of the morphological relationships of the 

 parts of the fruit is not easy or finally satisfactory with only 

 mature material available, but the following suggestions seem 

 helpful. 



The position of the four parts of the beak can offer little evi- 

 dence of value owing to the ease with which their form might be 

 modified. The inner flattened surface of the upper and lower 

 are approximately opposed to each other and the same is true of 

 the lateral elements of the four-parted beak. 



Internally, the anomalous capsules show four parietal placentae, 

 each expanded into two laminae upon the edges of which the 

 seeds are borne, very similar to those of the normal fruit. The 

 cavities are poorly defined since the ovule-bearing edges do not 

 reach the walls. 



Each of the claws (half of the beak of the fruit) of the ripe and 

 opened normal fruits represents not the distal prolongation of a 

 single carpel but the edges of both carpels, as is very clearly seen 

 from a section or from the conspicuous grooves marking on the 

 dry fruit the position of the placentae. This is also true of the 

 double fruits. 



On the peduncle the flowers are arranged spirally, so that if 

 we conceive of the abnormal fruit as formed from the primordia 

 of two successive flowers it will be seen that the synanthy will be 

 in part dorsal and in part lateral. This may explain the relation 

 of the parts of the fruit to the peduncle, as described above, and 

 the position of the crests on either side of the uppermost of the 

 four sections of the fruit. The accompanying diagram may 

 make clear a hypothesis as to the composition of the fruit. In 

 this diagram the walls of the two component fruits are indicated 

 by the difference in shading. The position of the crested outer 



