456 
Selective Elimination in Staphylea 
even a small proportion of them. The inflorescence is long enough to allow the 
lower ovaries to be developing into young fruits while tke upper ones are 
still in the unopened buds. A collection of the flowers before any elimination 
has taken place cannot be made for comparison with the ripened fruits, for 
before the more distal flowers are mature enough for the ovules to be counted 
with safety, the more proximal ones may be falling from the plant or developing 
into young fruits. 
It is desirable that the sample represent not only all parts of the plant, but 
also all regions of the inflorescence, since, should the ovaries be correlated with 
their position on the axis, the constants from material carelessly collected might 
not at all represent the population from which it is drawn. 
The labour of gathering ovaries with a record of their position on the inflor- 
escence seemed to be prohibitive. After careful consideration it became apparent 
that the collection of the spring material must be carried out in a rather rough 
and ready manner. The following was adopted. 
When a shrub was well in flower it was shaken gently to dislodge the flowers 
which had ceased to develop and were ready to fall. Sometimes flowers were on 
the ground which had been dislodged by wind and rain, but these were never 
taken for fear that they had been removed by too great violence. After the tree 
had been shaken and the fallen sample collected, a gathering as nearly random as 
possible of the opened flowers which remained was taken. 
In a few cases only a single collection of fallen and apparently developing 
flowers was made from an individual, but usually the shrub was visited again 
after a few days and a second gathering was made in the same manner as 
the first. 
Finally a series of matured fruits from these individuals was taken in the 
autumn. 
In the two spring series only ovaries dissected from well-opened flowers; 
were examined. All buds were discarded for fear that the ovules might not yet 
all be laid down. By discarding any young fruits which fell we avoided one danger 
of overweighting the proximal region of the inflorescence. 
From the foregoing remarks it appears that the 1908 ovaries fall into three 
classes : 
A. Ovaries from opened flowers which were eliminated from the shrub. 
B. Ovaries from opened flowers which were apparently continuing their 
development at the time of sampling. 
C. Ovaries which had completed their development to mature fruits. 
Series B should represent the original population of ovaries more nearly 
than A, since it contains many which would be eliminated later. Because of 
the methods necessarily employed in collecting the samples, I do not look upon 
the comparison of Series A and Series B as very critical. Only a relatively small 
