J. J. Simpson 
229 
this example what leally occurs is : There are two individual stamens each equal 
to the original length and bearing a complete anther and separated from one 
another by secondary chorisis. 
(2) Fig. L. Divisions 5 and 6 are sub-divided to show that there has been 
secondary chorisis in both of these stamens. In the former, chorisis has been 
complete but has resulted in one being full-length and with a functioning anther 
while the other is only half-length with a functioning anther. In the latter, 
chorisis has not been complete inasmuch as only the anther has been chorised. 
(3) Fig. LIT. Divisions 6 and 6 are both sub-divided, consequently we may 
infer that both of these stamens have undergone some stage of chorisis. In the 
first column we see that all are full-length, in the third that all have functioning 
anthers, but the second tells us that chorisis has been only partial in each case. 
The symbol h between 5 and 6 indicates that between these two chorisis has also 
been partial. Thus we conclude a state of affairs as follows : In position 5 . 6 
(1) there arises a single filament which divides into two at some distance from 
the base and (2) that each of these again sub-divides and (3) that on the end of 
each of these four sub-filaments there arises a functioning anther. The others 
may be worked out in a similar manner but a reference to the diagrams will at 
once obviate any misrepresentation. 
From the foregoing table and illustrations it is evident that further classi- 
fication is possible but it would be well to point out here certain difficulties 
which arise. As an example let us consider such a case as (using our original 
terminology) that in which, in any of the positions (1 ; 2 ; 3 . 4 ; or 5 . 6), the stamens 
are represented thus (1 .1 .0) (0.0.0), thus (i.1.1) (i .1.1) or thus (i.O.l) (f 0.1). 
Which shall have precedence ? If we are to consider these variations as deviations 
from the usually accepted normal cruciferous flower, then we may safely assume 
that that flower which has the greatest number of functioning parts in a certain 
position is less aberrant than one in which any or all of the parts are altogether 
wanting; while, on the other hand, if in a position in which chorisis normally takes 
place, we have defective groups like those in cases 2 and 3 cited above, in one of 
which chorisis has taken place but not in the other, we must consider that group 
in which chorisis has occurred as being the one less removed from normal. On 
this basis then the above examples would be placed in the following order with 
regard to normality : 
(1) (i.l.l)(i.l.l); (2) (i.0.1)(i.0.1); (3) (1 . 1 . 0) (0 . 0 . 0). 
Similarly for any of the others. 
Consequently we are now in a position to classify the actual cases under 
observation. 
So far we have considered only those flowers in which there was the typical 
number of stamens, with their manifold variations in size and structure, but now 
Biometrika x 30 
