118 BULLETIN 1059, U. S. DEPARTMENT OF AGRICULTURE. 



by M'. and the excess over wilting coefficient by K, then in these 

 diagrams the condition is represented by 



rather than 



W = WG+ K 

 M'= WCxK, 



since it is readily seen that the osmotic equivalents are not propor- 

 tionate to the coefficients. 



Table 6. Osmotic equivalent of soils, in presence of solution at 20 atmospJieres, after 



four months exposure and other related properties. 



1 Average of 4 samples taken from each well, representing the surface and depths of 1, 2, and 3 feet , so 1 hat 

 mean value should be equivalent to that of soil as placed in the well. 



\ptiroximate. Test made on coarse sandy soil from depth of 4 feet, most nearlv approaching the quality 

 •if sand used in the well. 



Table 6 shows that K varies as between different groups of samples 

 from different sources, but that within a group of similar origin K 

 is essentially a constant. Thus, it has an average value of 8.13 per 

 cent for one group, 6.55 per cent for another, and 6.89 per cent for the 

 third, and this value seems not to have any constant relation to the 

 change which occurred in the samples during their period of ex- 

 posure, so that it may be accepted as representing something near a 

 final condition. In one sample representing a limestone soil. K is 

 found to be 17.12-15.33 per cent, or 1.79 per cent. In another soil 

 oi Lava origin, containing less of silt and clay, but a considerable 

 amount of sodium bicarbonate, K is found to be L2.23-4.44 per 

 cent, or 7.79 per cent. 



These findings compel the following conclusions: 



1. The wilting coefficient of a given soil is probably dependent 

 both on the solutes present and upon the colloids capable of ad- 

 sorbing both the solutes and the water, but more particularly upon 

 the latter; since only very rarely will the solutes be so abundant 

 to create an excessively strong solution before the disappearance of 

 the free water. 



