majority of orchids, transitional stages, as well as com- 
plete absence of it, are also well documented. In 4 pos- 
tasioideae and Cypripedioideae all three stigmas are fertile ; 
therefore no rostellum is produced. In the remaining 
groups (Neottioideae, Ophrydoideae and Kerosphacroid- 
eae) it is assumed that a rostellum must be present. 
In some members of the Neottiotdeae (e.g. in the gen- 
era Spiranthes, Goodyera, Erythrodes, etc.) the style is 
modified into a wedge-shaped structure with the two 
separate stigmas situated laterally, while the third one 
is transformed into an elongate rostellum, all on the same 
plane. In this situation the nether surface of the rostel- 
lum is, however, still a functional stigma, as has been 
demonstrated by experimentation. We may look upon 
this condition as an intermediate step in the reduction 
or modification process, because in the more evolved 
members of the Neottioideae the rostellum ceases to be 
a functional stigma. 
There are also a number of species in the Ophrydoideae 
without any reduction in number of stigmas. Several 
attempts have been made recently to explain the pres- 
ence and origin in the Ophrydoideae of a so-called ‘‘ros- 
tellum”’ in addition to the three fertile stigmas. Since 
in the other groups (Neottioideae and Kerospheroideae) 
the gland of the pollinia is attached to the tip of the ros- 
tellum, it is believed that Ophrydoideae must also possess 
such a structure. Vermeulen postulated that the rostel- 
lum in the Ophrydoideae has an independent origin when 
all three stigmas are fertile, while Hagerup would derive 
the glands of the pollinia from the aborted lateral stamens 
of the outer whorl. I am unable to find a rostellum in the 
Ophrydoideae (comparable to that of the other groups): 
the structure which is generally called ‘‘rostellum’’ is 
merely a connective tissue between the two thecae of the 
anthers. Swamy in his studies of vascularization, has 
[ 75 
