146 W. B. Crump. 
Put as an equation their result is 
M=0-13 C + 0-62 (D + E), 
where M is the Moisture Equivalent 
C=per cent of silt (0-05—0'005 mm.) 
D=per cent of clay (0'005—0* mm.) 
E=per cent of organic matter determined by the chromic 
acid combustion method. 
Neither M nor E is identical with the constants adopted in this 
paper, but there is a general correspondence. So it seems likely 
that the water-content of a soil depends primarily, and approximately 
to an equal extent upon the clay and humus-contents. In the case 
of a clay soil then it may be anticipated that the equation 
y=mx + nz 
will hold good; or if n proves to have the same value as m 
y=m (x + z) 
where z is the percentage of clay. 
Though the method of investigation and the types of soil are 
totally different in the two cases, the conclusions arrived at in the 
present paper receive independent confirmation in the results of 
Briggs and McLane’s study of agricultural soils. Both deduce a 
linear equation to express the equilibrium under stated conditions; 
both point to the soil-moisture under such conditions being an 
adequate and simple index of the soil. 
IX. —Ecological Considerations. 
In conclusion it seems desirable to glance at the broader aspect 
of the problems involved in the present investigation of the soil 
moisture in relation to the plant association. If the coefficient of 
soil humidity is to be of service in ecology it will only be by a 
thorough study of definite habitats, and the lines on which this can 
be prosecuted to advantage may be indicated as follows:— 
(1) . The mean value of the coefficient of soil-humidity m is 
the most important constant to be ascertained for every association 
whose distribution and existence are determined by soil moisture. 
This mean value, obtained by methods already described, may be 
indicated by M. 
(2) . But the coefficient for every such association will prove 
to have a considerable variation from its mean value. Given a 
sufficient number of estimations this variability would be best 
expressed by the “standard deviation,” but in actual practice it 
would be sufficient to ascertain the maximal and minimal values, 
say m x and ;// a , of a considerable number of estimations. These 
