J. J. Simpson 221 
the chorised stamens to the peduncle and chorisis. The former would suggest no 
correlation, whereas the degree of correlation hinted at by the latter would depend 
on the frequency of the occurrence. 
We have so far considered only two possible positions, viz. a non-chorised 
stamen adjacent to the peduncle, i.e. in the proximal half of the flower, and a 
non-chorised stamen opposite to the peduncle (i.e. in the distal half of the flower 
with reference to the peduncle), but the question naturally arises " Are these the 
only two possible relative positions which might occur ? " Might the petiole not 
twist so as to bring the hypothetic non-chorised stamen into any position varying 
from 0° to 180° with reference to the original plane ? 
Let us illustrate this by means of the Figure 6, Plate I. 
Taking the position of the peduncle as our fixed point the non-choi'ised stamen 
might occupy the " adjacent " position a or the " opposite " position al. A rotation 
of the petiole, however, might cause this stamen to occupy any of the positions 
marked a2, a3 or a-i or even any intermediate position between a and al on either 
side of the vertical plane A — B, in the horizontal plane a, a4, a2, a3, al. 
In a study of the variations in this flower, this is precisely what was found to 
occur, i.e. the distribution was equal round a fixed point so that we are unable to 
say whether there is any connection between the proximity of the non-chorised 
stamen to the peduncle and chorisis or not. 
But the full bearing of this consideration does not end here. The orientation 
of the flower is of practical importance in fixing a basis on which to establish 
a grouping of the different variations. Any analysis of the data is impossible 
unless some definite part of the flower be agreed upon as a starting point. 
Now we have seen that the position of the peduncle with respect to any 
definite stamen does not require to be taken into consideration. Consequently 
we may take either of the two stamens of the outer whorl, which correspond in 
position to the outer sepals and which are "normally" non-chorised, as our fixed 
point and call it 1 ; the stamen opposite, i.e. in the same whorl, we shall call 2 ; 
the chorised pair of the inner whorl to the left (or in the floral diagram above) 
may be termed 3 and 4; while the corresponding pair to the right (or in the floral 
diagram below) would thus be 5 and 6 (Plate I, fig. 7). 
Where variations occur in any of these stamens we shall hereafter refer to 
those as occurring in " position " 1, 2, 3, 4 and 5, 6 respectively. 
On this basis of symmetry, it will simplify matters considerably if we regard 
as 1, in flowers in which either of the two outer stamens is modified, that one 
which still maintains its original character while, on the other hand, if both are 
modified, that one which retains the greatest approximation to normality, e.g. if 
one be chorised while the other is not, the latter would be in position 1; or if one 
Biometrika x '^9 
