1900 } MECHANISM OF ROOT CURVATURE : a 
error was reduced to a minimum. With smaller roots it would 
have been impossible to separate the different tissues from each 
other with a sufficient degree of accuracy to make the results 
reliable. The seeds were germinated in moist sawdust, and 
when the roots had attained a length of from 1-5 the seedlings. 
were removed from the sawdust, and placed with the roots pro 
jecting into water through holes in a thin board, the holes being 
of such a size that the upper part of the root fitted snugly, and 
the root was held in a vertical position pointing downward. The 
temperature of the water was regulated only by the temperature 
of the laboratory where the experiments were carried on, and 
was generally from 15°-20°C. No attempt was made to keep 
the temperature constant, and during the nights of the colder 
part of the winter it probably fell as low as 10°C. or lower; but 
as nearly all the observations used were made the same day the 
experiments were started, the fall of the temperature at night 
can have no effect on the results. The experiments were made 
with the roots in water rather than in moist air or sawdust, 
because the stage of the curvature could be seen in water with- 
out removing them and perhaps subjecting them to other stimuli ; 
and it was found by trial that the results were much more uni- 
form in water than in moist air. 
At the time of removing from the sawdust or after they had 
been in water for some time, the roots were branded on one side 
of the tip by means of a hot piece of metal. A warm stage was 
heated by means of an alcohol lamp, and the roots, held at an 
angle of 45° or less from the horizontal, were placed with their 
tips in contact with the hot stage for an instant. The roots were 
then put into water as described before. In from one to four 
hours after branding they begin to curve away from the side that 
was branded, and when the curve had progressed sufficiently they 
were ready for further observations or operations. Control roots 
were nearly always used, and in selecting them they were gen- 
erally paired off with the branded ones, so that the two sets 
should be as nearly alike as possible in point of size and stage of 
development. Roots that at first were inclined to curve 
