ON THE MORPHOLOGY OF CRITCirER^. 7 
for in my specimen they were imperfect, there being only one gland at 
the base of each pair of long stamens, and in one instance I noticed it 
had become confluent with the glandular receptacle of one of the short 
stamens.* I \ymdd suggest that the two glands in Iheris are the same 
glandular structure as the glandular receptacle of a short stamen in 
Cheirantlius, because in Matlliiolathe shoiii stamen is inserted between 
two large glands. f 
I believe that if such a plant as I)icentra,% in which the parts of the 
flower regularly alternate with each other (the last two stamens, each 
of which is split in halves, being alternate the tw^o carpels) were to be- 
come polyandrous like PapaveracedB, the additional stamens would be 
4 
produced in the same manner as if a stem produced nnmerous addi- 
tional leaves in the space between two inteniodes^ which, in the ordi- 
naiy growth of the plant, is entirely destitute of leaves ; such stamens 
T regard as supernumerary, but whether they are always so admits of a 
question. Would not the persistence of the two carpels of a dicarpous 
ovary in the same position, in the same Natural Order, whether the sta- 
mens were of the same number and alternate with the petals or very 
numerous, render such an hypothesis probable ?§ The glands of Cru- 
* My specimen was gathered from old walls at Winchester, where it had pro- 
hably been w^ild for many years. The flowers are below the average size, so 
that a part of the glands are likely enough to be deficient. 
t The diagram of CheirantJitis represents the two carpels as alternate with 
the short stamens, which does not accord with my observations on other CWe- 
ciferab^ e. g. Matthiola, But as this neither alters nor modifies his plan of the 
flower^ probably it is a cii-cumstance he did not attend to. As far as can be seen 
from the dried plant, I shoidd expect it would agree in this character with 
other GrucifercB* 
X In Dicentra speciahilis the two external petals are alternate with the two 
sepals, and the two internal in consequence become opposite them. The two 
external stamens, having 2-celled anthers, are alternate with the two uiternal 
petals, and in consequence opposite the two external. And the two internal 
stamens, which are split in halves, each half having only a 1-celled anther, are 
alternate with the two external, and in consequence opposite the two internal 
petals. The split stamen occupies the same po^jition relatively to the two 
carpels as the pair of long stamens in Crucifer(B^^N\\\c\\ looks hke a tendency to 
unusual development at this point, especially as the half-filaments, if they were 
combined, would be broader at the base than the filament of a 2-celled anther. 
The two longer sides of two 1-celled anthers (which in their natural position 
when on the plant are turned towards each other), on heing brought together, 
form an anther precisely like one of the 2-ceUed anthers, it having the same 
4-lobed ajipearance with the two smaller ones external. 
§ As regards leaves, I believe this really occui^ in a new species of Mono- 
choria, or perhaps a new nearly allied genus as the placentation is parietal 
(Brazil, Gardner), which has very numerous leaves in dense whorls, and appa- 
