358 
LOCK : STUDIES 
PLANT BREEDING 
partly announced, and to add to the number some further experi¬ 
ments which bear in the same direction. 
The object of these experiments was two-fold. They were designed 
in the first place to test the conclusions of Mendel and of those 
other botanists who in recent years have extended and confirmed 
his observations; and they were also undertaken with a view 
to applying these theoretical principles to the practical processes of 
plant breeding. To this attempt was’ added a special interest 
from the fact that it was to be made in a tropical country, under 
circumstances in which very little work of this kind had been pre¬ 
viously carried out. 
At the outset attempts were made to effect cross-fertilization 
between varieties of a considerable number of different species of 
plants, some of them natives of the tropics and others introduced 
from a temperate region. Many of these attempts met with only 
partial success, as indeed was to have been expected, and others 
were quite unsuccessful. Including all the cases of plants from 
which fertile offspring were obtained by crossing, a considerably 
larger number of experiments was begun than could have been 
completed if all had progressed without misfortune. But the 
progress of the work met with a variety of difficulties ; and 
furthermore, the surviving cross-bred plants underwent a kind of 
natural selection, the losses being determined by such factors as an 
unfavourable climate, a want of experience of the best conditions of 
growth, the attacks of fungi, insect pests, and the like. In the end 
the experiments with Indian com and with peas were found to be 
the only ones which yielded results of any importance. In the 
present instalment I shall confine myself to an account of the 
evidence derived from peas. 
In the case of experiments with peas, Mendel’s own results, as 
' uurnoer oi 1 
sent observations, could be put to a 
practrcal h*t. In work of this kind, if it is desired to pass beyond 
the stage of merely confirmatory experiments, it is often necessarv 
at t e outset to make crosses more or less at random between such 
forms as may be at hand, in the hope that some of the results will 
