54 
Edith R. Saunders. 
II.—Results of Breeding Experiments. 
From the results obtained it appears that D. purpurea heptandra 
breeds true in that from heptandra individuals only heptandra forms 
have been obtained. As stated above, however, the extent to which 
the characteristic features of dialysis and staminody are exhibited 
by any individual is not generally constant throughout the flowering 
period, but varies in the different flowers according to their position 
(level) on the spikes (fig. 12). As a rule the flowers in order on an axis 
form a graduated series, but the particular grade with which the 
series begins and ends, and the number of grades exhibited vary in 
different individuals. Further, the series may be either wholly up¬ 
grade, approximating more and more to the normal as we approach 
the top of the spike, or the culminating point in the reversionary 
series may be reached below the apex, and may be followed by a 
series of down-grade stages. We have as yet no very definite 
knowledge of the forces regulating this wave-like tendency, but it 
can hardly be doubted that it is not assignable to any single cause, 
but is due to the combined action of a number of factors, of which 
some exert a preponderating influence at one time, some at another, 
with the result that the balance is inclined at first towards, and 
later, perhaps, away from the condition of normality. We can thus 
understand how it is that the phase exhibited by the open flowers 
on a lateral branch may not correspond with that of those opening 
simultaneously on the main axis; or again that the condition of 
fasciation is often accompanied by a sudden break in the regularity 
of the series. It is not surprising that the state of vigour of a 
lateral branch beginning to flower should be different from that of 
the main axis which is already probably well advanced in its 
flowering period; hence on the lateral branch the first flowers may 
start from a point in the series behind that already reached by the 
open flowers on the main axis, but yet not so far back as to recapi¬ 
tulate the whole series. In the case of fasciation, flowers borne at 
the level where fasciation begins to occur may show a sudden jump as 
regards increased size of the petaloid structures, as compared with 
the flowers borne immediately below the fasciated region, or with 
those opening simultaneously on a normal stem of the same 
individual. The heptandra individuals are, however, easily sorted 
