J. J. Simpson 
245 
one of the positions of the inner whorl of the androecium. This is the number 
in the accepted type, and thus there is no variability. What is the nature of the 
distribution of the stamens in the other position (Table XVIII) ? It is as follows : 
1 
2 
3 
4 
287 
1436 
26 
5 
1754 
The mean for this array is 1"8569 stamens, with a variability of •4116. Thus 
when there is no variability in one position of the inner whorl of the androecium 
there is a large variability in the other position. 
Similarly we find the following distribution for position 2 in the outer whorl 
of the androecium when position 1 is of the accepted type, i.e. shows no variability. 
0 
1 
2 
57 
1708 
1 
1766 
The mean for this array is 'OGSS with a variability of •1784. Again therefore, 
when there is no variability in position 1, there is a reduction of type in position 2 
with great variability. 
2. Study of the Correlation Coefficients. 
For the purposes of this study a number of correlation tables have been 
prepared and as the results of these will have to be considered under different 
groupings it seems advisable to tabulate them, and insert them consecutively. 
The system which has been adopted to facilitate reference is to commence with 
the outer whorl of the calyx and consider all its relations with the other whorls of 
the flower passing from the outside inwards ; following this comes the inner whorl 
of the calyx and its relations with the other constituents of the flower from the 
outside inwards and so on. 
The following table shows the characters studied and the correlation coefficients 
found. 
In order to make the comparison of the various correlations as complete as 
possible it will be necessary to consider each constituent or oi'gan with all the 
other constituents or organs and to avoid overlapping as far as possible. The 
most natural method would be to commence either with the outermost constituent, 
namely, the outer whorl of the calyx, or with the innermost constituent, namely, 
the inner whorl of the andrcecium. For reasons of a morphological character, which 
will be seen later, the inner whorl of the andr«cium has been chosen as the 
starting point. 
Biometrika x 32 
