50 
WATER REQUIREMENT OF PLANTS. 
kiesselbach's experiments. 
Kiesselbach (1910) measured the water requirement of corn 
plants grown in river sand which was kept saturated with Schim- 
pers's normal nutrient solution. Four concentrations, 0.6, 0.4, 0.2, 
and 0.1 per cent, were employed. Five pots, each holding about 15 
kilograms and containing a single plant, were used for each concen- 
tration. The method followed is the same as that described on page 
20. With the exception of the highest concentration, which reduced 
the growth of the plants, the results (Table XXXVIII) show a slight 
increase in the water requirement as the concentration of the solu- 
tion is diminished. 
Table XXXVIII. — Influence of concentration of soil nutrient solution on the water 
requirement of corn, according to Kiesselbach (1910). 
Strength of 
solution. 
Dry weight 
including 
roots. 
Water re- 
quirement. 
Per cent. 
0.6 
.4 
.2 
.1 
Grams. 
47.8 
74.9 
65.3 
73.4 
245±5 
226 ±5 
238 ±4 
260±4 
LEATHERS EXPERIMENTS. 
Leather (1910, 1911) has investigated the use of nitrogen, potas- 
sium, and phosphorus as fertilizers in their relation to the water 
requirement of a number of crops in India. These experiments were 
conducted with four soils, which may be characterized as follows: 
The Pusa soil, where the Pusa Institute is situated, consists of Gan- 
getic alluvium and contains a small proportion of clay. It is highly 
calcareous (30 to 40 per cent CaC0 3 ), low in organic matter, and very 
poor in available phosphate. Potash salts have no effect on the crop. 
The Akola soil is a typical " black cotton" soil, with a high clay con- 
tent and from 1 to 3 per cent of calcium carbonate. The Shillong 
soil is highly ferruginous and contains much organic matter. The 
Palur soil is sandy in character. 
The water requirements of crops grown in these soils, showing the 
effect of various fertilizers, are given in Table XL. Each determina- 
tion refers to the mean value of the water requirement obtained from 
two pots, except in the case of the earlier measurements for wheat and 
corn, which represent the mean of the values given in Table XXXIX. 
The soil water content, except in the earlier determinations, was main- 
tained as nearly as possible at 20 per cent of the saturation content. 
The writers have added the probable error of the mean in each case 
as a guide in making comparisons. 1 
i The probable error, when only two determinations are available, has little value other than that of 
serving to show how closely the duplicate determinations agree. Usually in comparing mean values to 
determine whether a real difference exists, a difference equal to three times the square root of the sum of the 
squares of the two probable errors is considered the minimum significant difference. The probable error 
of the probable error of the mean of two determinations is one-half that of the probable error of the mean, 
285 
