528 Boodle. — On the Trifoliolate and other Leaves of 
trifoliolate type. Hence these are included in the countings, and for con- 
venience the bi- and trifoliolate and lobed leaves will be spoken of together 
as compound leaves. Among the apparently simple leaves some have an 
articulation at a distance from the base. This again is an indication of 
a compound nature, and these leaves would also have been counted with 
the trifoliolate leaves, but the presence of the articulation was overlooked 
until many of the plants had been examined and thrown away. Thus one 
type of approach to the trifoliolate condition has been ignored, but it appears 
probable that the nature of the result has not been thereby altered. 
The experiment was started at the end of March, 1913, when seeds 
were sown in two boxes of soil, composed of loam, sand, and leaf-mould, 
and in four boxes of sand, two of the latter being kept drier than the others 
during part of the experiment. The first examination of seedlings was 
made in the middle of July. It was then found that, though there was 
a good crop of seedlings, spiny above, others were still young, and some 
seeds were only just beginning to germinate. The backward seedlings, 
which were much more numerous in the cultures on sand, were examined 
later at intervals when they had progressed far enough. When the experi- 
ment was stopped at the beginning of September, some seeds were again 
just germinating, but the experience gained by that time appeared to show 
that the rest of the plants could be discarded without altering the direction 
of the result. 1 
The main result of the experiment is that seedlings of gorse grown on 
good soil, under the conditions of the experiment, produce a somewhat 
larger average number of compound leaves than those grown on sand. The 
seedlings on soil gave an average of 10*79 compound leaves per plant, 
while those on sand gave an average of 8-27 ; the difference (2*52) amounts 
to about 23 per cent. 
The number of seedlings examined was 2,895, of which 1,094 were on 
soil, and 1,801 on sand. In counting the leaves, the different boxes were 
dealt with separately. The values for the average number of compound 
leaves given by the two boxes of soil came out very close, namely 10*73 
and 10-87. The figures for the boxes of sand were 8*6o and 8*43 for the 
two drier boxes, and 8*49 and 7*51 for the two damper boxes. 
The conclusion appears to be justified, that in this case the presentation 
of an ancestral character is influenced to some extent by the nature of the 
soil, though at the same time there is great independent diversity in respect 
of this character. 
The very backward seedlings which were not included in the countings 
from the boxes of soil were comparatively few, and w r ould not have greatly 
1 See below, p. 529. The soil and sand used in the experiment were not sterilized. Root- 
tubercles were present on some seedlings. The effect of these would probably be to slightly 
diminish the difference in nutrition on the two kinds of soil. 
