354 
Journal of Agricultural Research 
Vol. VII, No. 8 
The control soils H and S, exposed in the second set, contained 2.3 and 
7,4 per cent of hygroscopic moisture, respectively, thus giving ratios of 
2.44 and 2.98, respectively, with an average of 2.71. The hygroscopic 
coefficients calculated from the hygroscopic moisture, using this average 
ratio, and the departures of these from the directly determined values 
are reported in Table VII. In general, the departure is slight and on 
the average is negligible. However, it should be pointed out that the 
latter circumstance is due to the average of the ratios for the two control 
soils being the same as that for those of the 36 samples under considera¬ 
tion. If we had used only H as a control, the calculated values would 
have been one-tenth higher, while if S alone had been employed, they 
would have been one-tenth lower. In either of these two cases, however, 
the calculated values would have been much nearer the directly de¬ 
termined hygroscopic coefficients than would the values computed' from 
the moisture equivalents by the Briggs-Shantz formula (3, p. 839). 
We similarly exposed several other series of soils of which the hygro¬ 
scopic coefficients had already been determined. The concordance of 
the calculated with the found values was much alike in all. The first of 
these series, which was strictly typical, is reported in Table VIII. It 
consisted of 24 samples, part surface soils and part subsoils, but all 
derived from residual material in western Nebraska. The soils were ex¬ 
posed on metal trays on the shelves of an inclosed basement room for two 
weeks, April 29 to May 13. The maximum temperature recorded in the 
room during this period was 2 2 0 C., and the minimum 18°; observations in 
this room, extending over three years, had shown that there was but rarely 
a daily range exceeding one degree. In the table the soils are arranged 
in order of texture. The ratio for the two control soils exposed in triplicate 
was 2.9. The data on the determined hygroscopic coefficients are the 
means of duplicate determinations. The calculated values in nearly all 
cases agree satisfactorily with those directly determined. The greatest 
divergences are shown by soils 13 and 23, the calculated hygroscopic 
coefficient being one-sixth too low for the former and one-seventh too 
high for the latter. 
Finally we exposed on paper pie plates 145 soils of which the hygro¬ 
scopic coefficients had not been determined. Each soil was exposed in 
duplicate and there were also 15 plates of each of two control soils, 
La and S. In the case of both of the control soils triplicate samples 
were exposed in different parts of the room in order to determine whether 
the position in the small room exerted a marked influence upon the amount 
of hygroscopic moisture absorbed. During the eight days of exposure, 
June 5 to 13, the temperature ranged from 18 0 to 23 0 C. 
The position in the room wag found to have a slight but appreciable 
influence upon the amount of moisture absorbed, the extremes being 
shown by two shelves, on one of which the samples of S and La were 
found to have 9.3 and 3.8 per cent, respectively, and on the other 8.7 
