42 
Research Bullet in Xo. 3 
Very different from the above are the data on the three cylin- 
ders of semiarid soil. In the first foot alone were the conditions 
similar to those in the three cylinders of humid soil, the drying 
due to direct evaporation having extended to the bottom of the 
surface foot. Below the first foot conditions were very different. 
There was no regular rise in moisture content with increasing 
depth, it rising and falling independently of the depth. The 
moisture in the four 3-inch sections of each foot was uniform 
within the limits of experimental error. Here the statement of 
the free water is instructive. In I and IV, which had produced 
the heaviest crops, the moisture in the different foot sections 
below the surface foot had been reduced almost to the hygro- 
scopic coefficient, viz. to 1.4 per cent of free water in I and to 
0.8 per cent in IV. In II, which had not produced as heavy a 
crop, the moisture content was considerably (6.7 per cent) above 
the hygroscopic coefficient and the moisture rose slightly from 
the surface downward. In all three the roots had penetrated 
freely to the very bottom of the cylinders, being found even in the 
sand layer. When the soil moisture had run low the roots seem 
to have drawn quite uniformly on the free water in the different 
levels. 
The data given on line 12 of Table 6 on the loss of water per 
gram of dry matter of crop produced include the water lost by 
direct evaporation from the soil as well as that transpired. A 
satisfactory correction for the amount lost by direct evaporation 
cannot be made from the data on hand. It is certainly consider- 
ably below that lost from the implanted cylinders. The great 
difference between the two unplanted cylinders is to be at- 
tributed, at least in part, to the marked difference in water ca- 
pacity of the soils they contained, while the initial total moisture 
content was very similar in both. The water transpired per unit 
of dry matter produced lies between the very wide limits of .118 
and 934. The yields of grain calculated to an acre basis are 
interesting, those on the semiarid soil being 26. 30, and 41 bushels 
and on the humid soil 6, 7. and 10 bushels. Such a result could 
not be obtained in the field on a deep, well-drained semiarid soil 
similar to that used, unless water were added during the growth 
of the crop, for the reason that so much water would not be held 
in the surface six feet. 1 
On the H O Ranch, 200 yards from where the semiarid soil 
was obtained, there was a young orchard which had been kept in 
clean cultivation for 3 years. On November 21. 1907. the soil con- 
1 It should be emphasized in this connection that 6 feet is not the 
limit of root penetration for wheat on all semiarid soils where free water 
in considerable quantity extends to a greater depth. 
