(163 ) 
for the sake of comparison with the other species to take the 
values represented in Table XIX, as it will be seen later that 
there is some correlation between change of position and 
change of length. In“ 6-9 we have a representation of the 
changes in length, in each of the four cases, together with 
the other features mentioned before. 
Comparing the three species as regards the change in 
length we get an average decrease of 6.8 per cent. in Par- 
thenocissus, of 6.7 per cent. in Zr7foliwm and of 6.9 per 
cent. in Luprnus (7.0 per cent. in the last named if we con- 
sider all the cases). This gives us an average decrease of 
.8 per cent.as the result of the removal. When we take 
into consideration the striking uniformity in the results, the 
fact that in each case the difference representing the change 
is greater than the sum of the probable errors of the averages, 
and above all the care exercised in securing uniformity in the 
two series of one species, we must conclude that the removal 
of one of the leaflets tn the compound leaves of the three spe- 
cies mentioned directly inhibits the growth of the remaining 
leaftets. 
As was hinted above there is yet another matter con- 
cerning the changes in length which seems to be of great 
interest. This concerns again the operated series of Lupinus 
albus as givenin Table VI. The great variety of rotations 
furnishes a basis for a consideration of the relation between 
the length of a leaflet and its position. We have shown in 
Table XVIII that the general averages of Lupinus albus ex- 
hibit a very uniform decrease when we compare the whole 
operated series with the normal one. What will be the re- 
sult if we separate the cases into groups as already done in 
the treatment of the intervals? If the length of a leaflet is at 
all correlated with its position in the leaf-system may we not 
find some indication of it here? Obviously the differences 
with which we must deal in this case are dangerously near 
the sums of the probable errors of the averages, but if we get 
anything like uniformity in the different cases we should not 
hesitate to attach importance to the results. 
