EXPERIMENT STATION BULLETINS. 599 



of supercooling and the heat of solidification of the solvent, which is 

 obviously equal to its heat of fusion. The greater the undercooling, 

 therefore, and the larger the amount of ice that will separate at the 

 moment of freezing, the greater will be the concentration of the remain- 

 ing solution, and, consequently, the larger will be the error on the 

 lowering of the freezing point. In order to compensate for this source 

 of error as much as possible, a correction may be introduced. Let u 

 be the undercooling of the solution in degrees, s, the specific heat of 

 the liquid, and 1, the latent heat of fusion of unit weight of the solvent ; 



su 

 then, — = f, where f is the amount by which the solution Avill be con 



1 

 centrated, due to the separation of ice. 



While this formula may be applicable in case of solutions in mass, 

 it was not considered to be so in soils, and consequently it was not 

 employed in the final determination of the lowering of the freezing 

 point of soils. The error arising from this source, however, is not 

 great enough to affect the results very appreciably considering the 

 amount of undercooling employed. 



The length of time at which the mercury column remained at the 

 proper freezing point seems to be inversely proportional to the con- 

 centration of the soil solution. This phenomenon is, of course, due to 

 the separation of tlie solid solvent which causes the solution to become 

 more concentrated and lowers its freezing point. When the soil solu- 

 tion is very dilute, the relative mass of pure water is large, and the 

 time required for all of it to freeze in the solid phase so that the freez- 

 ing point may sink, is very long. Hence, the real freezing point re- 

 mains at its place a long time. When the soil solution is concentrated, 

 however, the relative mass of the pure solvent is not large, and a small 

 quantity of ice separated, concentrates the mass of the solution and 

 lowers the freezing point, hence, the length of period during which the 

 actual freezing point remains at its place, is short. The time varies 

 from several minutes, in the soil with high moisture content, to less 

 than a minute in the soil with a very low water content. 



It has already been stated that the temperature of the cooling bath 

 was maintained at about — 4.5° C. It was found that if the temperature 

 was lower than this the rate of cooling of the soil mass would be too 

 rapid and solidification would commence before supercooling had oc- 

 curred. The temperature then would descend very slowly and the 

 proper lowering of the freezing point could not be determined ac- 

 curately. On the other hand, if the temperature was less than — 4.5° 

 the soil mass would cool very slowly, and consequently, would take 

 much longer to make a freezing point determination. 



From theoretical reasons, a too low temperature in the cooling bath 

 involves an error in the final reading owing to the radiation and con- 

 duction of heat from the soil. The greater the difference in tempera- 

 ture between that of the soil system and the cooling bath the greater 

 is the amount of radiation from the soil and consequently the lowering 

 of the freezing point will be slightly greater. The amount of error 

 arising from this source, hoAvever, is also very small. 



As previously mentioned, the lowering of the freezing point of soils 



