252 ORCHIDEA. 
The constant correspondence of the most perfect stamen, 
with the most perfect stigma, that is so far as the rostellum 
may guide us, argues strongly in favour of their true oppo- 
sition ?. 
This instance proves that the most permanent parts of Or- 
chideous structure, are the union of the filaments and style 
into one column, and the cohesion of the pollen grains into 
threes and fours. And it proves that in Vandea at least, 
an equally permanent character exists with the union of 
compound pollen grains into waxy masses of a definite 
number. 
It would be important however to determine the ‘vascular 
bundles of the column, their distribution, to explain the 
ovary, and to state what cases are known in monocotyledons 
of definite ovaria being more in number than the perianthial 
leaves. 
Of the relative amount of permanence of the above charac- 
ters, at present I can only speculate. If it be found that the 
irregularity of the present flower, which chiefly relates to the 
labellum, be of comparatively late date, we may assume that 
: sois be the soonest to disappear. And if the quaternary or 
rnary aggregation of the grains be the earliest in date, it will 
> the last to disappear. 
Besides this, the aggregation of the pollen grains is of 
almost universal occurrence in all families, and anomalies from 
arrest of development are, I am disposed to think, much more 
permanent than anomalies from excess of development? 
Has the occurrence of dimidial or unilocular anthers, even 
though abortive, any relation to Cannean anthers? 
In another instance in which two deficiently developed 
flowers had grown together, there was an apparent tendency 
in the labellum to assume the antheriform structure. 
However easily this deviation from the usual Orchideous 
form may appear of solution, and however easily, judging 
from it, Orchidez may be referred to a hexandrous type, the 
three innermost stamina being abortive, the hypothesis is 
