600 STATE BOARD OF AGRICULTURE. 



could be measured from any maximum water content down to a very 

 low water content, probably lower than at the wilting coefficient of 

 plants. Below this last lower limit, solidification could not be induced 

 very readily. At the high moisture content duplicate determinations of 

 the same sample would agree exactly with one another, or vary only a 

 few thousandths of a degree. At the low moisture content, however, the 

 duplicate determinations of the same sample would agree only with 

 some soils, such as the quartz sand and sandy soils, but with other soils, 

 such as the clay and loam, the variation would be considerable, but 

 always consistent, namely, the first lowering of the freezing point was 

 always greater than the second. The duplicate determination of differ- 

 ent samples of the same soil, however, would agree very closely, even 

 with the latter soils. The probable reason for the change or decrease 

 in the lowering of the freezing point with successive freezings in the 

 case of colloidal soils at low moisture content will be offered subse- 

 quently. 



Great care was always exercised to secure as uniform moisture con- 

 tent throughout the entire sample of soil under investigation, as possi- 

 ble. It was recognized that if some particles of the soil sample were 

 not moistened the lowering of the freezing point would be less, because 

 not all of the soluble material would be in solution. The soil sample 

 short of saturation, therefore, was prepared by passing it through a 

 sieve several times and mixing it thoroughly with the fingers. Even 

 then, it cannot be said that all of the soil grains were moistened exactly 

 to the same extent, but if they were all sufficiently moist so that their 

 film would freeze upon supercooling then the freezing point depression 

 observed would be an average of the lowering of the freezing of the 

 various particles. The soil samples which contained high moisture con- 

 tents and could not be passed through the sieve, especially if they con- 

 sisted of the colloidal types, were thoroughly worked with the fingers. 

 Great precaution was taken, of course, to have the latter thoroughly 

 clean. There is some danger of salts being added to the soil through 

 perspiration of the fingers, but the amount would be so insignificant that 

 it could never be detected by the method. 



It was feared at the outset that on .account of the compressability 

 of the mercury bulb great errors might be caused by the expansion of 

 the soil upon freezing. A thorough examination of this point, however, 

 proved that an error might arise from this source only when the soil 

 in the tube was extremely compacted, but otherwise there is none. To 

 guard against any possibility of error of this nature, however, the soil 

 in the tube was only slightly compacted and the bulb of the ther- 

 mometer was always moved in the soil at the beginning so that it would 

 be loose. 



In calibrating the thermometer with distilled water and attempting 

 to keep it at that set point for several days, it was found that it would 

 not remain the same, but would change. This change was not always 

 in the same direction, but varied in both directions. In examining 

 several treatises on thermometry, no mention was made of this fact, 

 and could not be accounted for at the beginning of the investigation. 

 It was soon discovered, however, that it was caused by the difference 

 in temperature of the room in which the thermometer was calibrated 



