1900| MECHANISM OF ROOT CURVATURE 3F 
of the curve. In the table x stands for the strip of convex cor- 
tex, v for concave cortex, a for the axial cylinder. The first 
part of the table shows the relative length of x, v, and a in water, 
the second part in 10 per cent. KNO,. Sometimes the cortex 
curved away from the axis, and had to be held in place in order 
to see their relative length. In doing this care was taken not to 
push the axis out of the position it naturally took. Most of the 
roots were somewhat curved, and if they were forcibly straight- 
ened, x was always longer than a, and v nearly always shorter 
than a. The time of splitting varied from 1.5-6 hours after 
branding. 
From the table it is seen that in water 23.3 per cent. and in 
KNO, 35.5 per cent. of the roots showed both strips of cortex 
longer than axis and the convex side longer than the concave. 
In 13.3 per cent. of those in water and 9.6 per cent. of those in 
KNO,, x and v were equal but both longer than a. In water x 
was shorter than a in 3.3 per cent., in KNO, # was equal to a 
in 6.4 per cent. of the roots. 
In water x was longer than v in 86.6 per cent. and longer 
than a in 96.7 per cent.; v was longer. than a@ in 36.6 per cent., 
equal to a in 30 per cent., and shorter than @ in 33.3 per cent. of 
the roots. 
In KNO, # was longer than a in 80.5 per cent., and longer 
than v in 83.8 per cent.; v was longer than @ in 45.1 per cent., 
equal to @ in 16 per cent., and shorter than a in 38.7 per cent. 
of the roots. 
All the measurements in this table are merely relative and 
not absolute. It is not shown from these experiments that the 
tension is increased on the convex side, but itis shown beyond a 
doubt that it is decreased on the concave side to such an extent 
that in about one third of the cases it was reduced to zero, and 
in another third was reversed, the cortex of this side becoming 
negative with respect to the axis. This shortening cannot be 
merely by compression, for there is not sufficient lengthening 
of the convex side above the normal to account for so much com- 
pression. Hence the stimulus causes a change in the cells of 
