200 
Journal of Agricultural Research 
Vol. VIII, No. 6 
In performing tests to ascertain how closely duplicate determinations 
would agree by following the same procedure as closely as possible, it 
was found that in the majority of cases the duplicate determinations of 
the same kind of soil would agree very closely. Occasionally, however, 
the disagreement would be quite appreciable. This disagreement may, 
be attributed partly to three factors. 
In the first place, it appears that the phenomenon of solidification or 
crystallization does not start always the same; sometimes it starts with 
difficulty and its velocity is small, while at other times it starts very 
readily and its velocity is very high, even though the amount of super¬ 
cooling in both cases is the same. 
In the second place, the force of expansion may not be evenly dis¬ 
tributed throughout the bulb, and consequently the entire expansion 
may not be indicated in the stem. It is perhaps due to this factor that 
stirring or moving the bulb in the cooling mixture hastens the equi¬ 
librium. This equilibrium would doubtless be attained in a shorter 
time, and the results would be much more accurate if the stirring was 
performed in the bulb. A dilatometer was devised in which the stirring 
could be performed in this manner. This dilatometer consisted of a 
long tube having a ground joint, and on this joint were attached the 
thermometer and the measuring stem. The contents of the tube were 
stirred by an electrical arrangement similar to that employed in the 
Beckman apparatus for performing freezing-point-lowering determina¬ 
tions. Unfortunately tubes with ground joints such as were necessary 
could not be obtained now in this country; and since it was necessary to 
have several of them, this form of dilatometer had to be abandoned. 
In the third place, the water which the soils cause to become inactive 
or unfree is not in an absolute unchangeable condition, but is made free 
by various factors. Hence, if the empirical procedure of the method 
is not exactly the same, the duplicate determinations would naturally 
not agree. 
In order to obtain the correct factor or value for converting the expan¬ 
sion due to ice formation into the corresponding weight, the expansion 
of 5 c. c. of distilled water was determined in the dilatometer employed. 
It was found that these 5 c. c. of water gave an expansion of about 
0.5 c. c. upon freezing. According to these figures, 1 c. c. of water 
expands to about 0.1 c. c. upon freezing. This value is somewhat 
higher than that obtained by other investigators. Bunsen's data, for 
instance, show that 1 gm. of water at o° C. expands 0.09070 c. c. upon 
freezing. In the computation of the data presented in this paper, 
however, the former value was employed. It is believed that for the 
writer's purpose this value is sufficiently correct; it is also very conven¬ 
ient. 
