B. A. Keen 403 



very fair approximations to the amounts present in these three forms. 

 The actual amount of any one division varies considerably from soil to 

 soil, but generally speaking, the amount of free water decreases, while 

 the capillary-adsorbed and combined water increase as the soils pass 

 from non-colloidal to colloidal in type, although Bouyoucos finds many 

 exceptions to this rule. 



It will be observed that the dilatometer method as used gives 

 qualitative information only. The observed fact that the capillary- 

 adsorbed group passes gradually at one end to combined water, and at 

 the other to free, suggests that the variants controlling its relation to the 

 soil alter in a continuous manner over the whole range of this division, 

 a view confirmed by the present writer's experiments cited above. 

 This view is also supported by Bouyoucos' second set of experiments on 

 the lowering of the freezing-point of the soil solution. 



These experiments were done in the usual Beckmann apparatus, on 

 the moist soils and sands. It was found that solidification could be readily 

 induced, except when the moisture content was quite low. In quartz 

 sand for instance, the lowering of the freezing-point could be measured 

 when the moisture content was only 0-7 per cent. 



In quartz sand and some extreme types of sandy soil, the depression 

 of the freezing-point was found to be approximately inversely propor- 

 tional to the moisture content, i.e. 



MD = K, . . . . (1) 

 where M = moisture content, 



D = depression of freezing-point, 

 K = constant. 



With soils a different relationship holds, the depression of the freezing- 

 point increasing in geometrical progression as the moisture content 

 decreases in arithmetical progression. Bouyoucos interpreted this as 

 indicating that the soil solution increases in concentration at a greater 

 rate than would be accounted for by the known decrease in total moisture 

 content, and the general hypothesis is advanced that in soils some of 

 the water is rendered "unfree," and thus does not enter into the actual 

 soil solution, as determined by the freezing-point method. It is the 

 quantitative examination of this suggestion, which is supported by a 

 considerable amount of evidence in addition to that advanced by 

 Bouyoucos, with which the present paper is mainly concerned. 



