144 PEINCIPLES OF PLASTT-I'EBATOLOGY. 



2. In cases where the labelhim is suppressed, or in 

 the form of a lateral petal, stamens A^ and A^ some- 

 times reappear. But this is far from proving that 

 they were ever fused with the labellum. Moreover, 

 in the second alternative the change is merely that 

 towards the primitive, regular flower, as in Aposta- 

 siese, and in the first alternative the change can be 

 explained along the lines of correlative development 

 of organs. 



3. Magnus describes a flower of Orchis papiiionacea 

 in which anther-formation occurred on either basal 

 edge of the labellum in most of the flowers of an 

 inflorescence. This case is certainly in favour of the 

 popular hypothesis. It would seem, however, to be 

 the only one known of the same nature. It can quite 

 well be interpreted in another way, viz., as a partial 

 reversion, such as we so often see in petals, to the 

 original staminal nature of the whole labellum. Both 

 His and R. Brown describe transformation of the 

 petals into stamens. 



4. In the flowers of Gi/pripedium insigne (PL XL, 

 fig. 5), recently seen at Kew, in which the two lateral 

 petals were changed into labella, it is true that the 

 normally fertile stamens a} and a^ of the inner whorl 

 were absent. But this does not prove that they have 

 entered into the construction of the extra lips. It is 

 probably simply a case of correlative development : 

 the extra lip-formation involving economization of 

 material in other, closely contiguous, parts of the 

 flower, and this has meant suppression of the two 

 stamens. Besides, on the Brownian and Darwinian 

 theory for the formation of each lip two stamens are 

 required, whereas there is only one stamen available for 

 fusion with each petal in the proper position. To 

 suppose that each of these stamens has become divided 

 into two for the purpose is an assumption for which 

 there is no evidence whatsoever. 



Points against the theory. — The facts and argu- 

 ments which decidedly seem to tend towards disprov- 



