492 A. B. BEAUMONT 
finally used. As some of the data obtained may be of interest, they are 
here presented. 
Effect of temperature on adsorption of water vapor.— The effect of tem- 
perature on adsorption of water vapor is a point that has been disputed. 
Patten and Gallagher (1908) and some other investigators (Taylor, 1915) 
have found a decrease in adsorption with an increase of temperature. 
This is what one would expect from a consideration of Le Chatelier’s 
theorem, for adsorption is accompanied by a liberation of heat, a rise 
in temperature. Therefore, raising the temperature would tend to reduce 
adsorption rather than to increase it. 
Hilgard (1911), Lipman and Sharp (1911), and recently Alway, Klein, 
and McDole (1917), on the other hand, think that the amount of water 
vapor adsorbed increases with the rise in temperature. The experimental 
work of hygroscopicity determinations has a high probable error. This 
should be taken into consideration in drawing conclusions from the data, 
which was not done by any of these investigators. 
In order to learn whether any great differences were obtainable with 
slight variations in temperature, adsorptions were run at various tempera- 
tures. The results are given in table 1. The results show differences, 
but in no case is the difference sufficiently greater than its probable error 4 
TABLE 1. Errecr or TEMPERATURE CHANGES ON THE ADSORPTION OF WATER VAPOR 
BY DUNKIRK SURFACE SOIL 
Per cent* of 
water vapor Difference 
adsorbed 
UNE ND NY Opgminis Mencia Ay aired sia Aone nba s tee gaat aes 5 tue 3.450-£0. 055. 
0.1750. 066 
y Niner am OF apie nr Aaa Ree) Lire eae at ote i te 3.275+0.037 
0.095--0.066 
Dara Ut OLSEN NG Sopa sat emia Ved Lei A SR Ga i Wee 3.180+0.055 
0.150+0.097 
Nt AO RIC Sale Pea es ir terre te Cate n eID e ai ean aay Rp eam 3.330+0.080 
* Unless otherwise stated, all percentage calculations in this article are based on oven-dried weights 
of soil. 
1 The probable error of the mean in this and all other experiments cited in this article was calculated 
by means of Peter’s approximation formula, given by Mellor (1913). The probable error of the mean 1s 
0.8453 ¥' (+v) 
nV (n-1) . : 
out regard to the sign, and n is the number of individuals. If the difference between means is 3.8 times 
its probable error, the chance is 30 to 1 (Wood and Stratton, 1910) that the difference is significant. 
, in which ¥ (-|v) is the sum of the deviations of all the individuals from the mean, with- 
