32 BULLETIN 1193, U. S. DEPARTMENT OF AGRICULTURE. 
The low results afforded by the corrected dye and water ratios for 
the Cecil soil and the low result by the water ratio for the Huntington 
soil were probably due to the sampling error. Data given in Table 6 
support this view. It is apparent that the unextracted colloid of the 
Cecil soil had an exceptionally low adsorptive capacity for dye and 
water but not for ammonia as compared with the extracted colloid. 
A similar condition obtained for the water adsorption capacities of 
the extracted and unextracted Huntington soil colloids. 
It seems evident from a comparison of the values in columns 2 to 4 
with those in column 5 that the water adsorption ratio is more likely 
to show the true amount of colloid in the soil than the dye or ammo- 
nia adsorption ratios if no correction is applied for the altered adsorp- 
tive capacity of the extracted colloid, at least when the larger part of 
the extractable colloidal material is taken as a sample. 
It was pointed out in a previous paper (Robinson, 26) that the 
average water-adsorptive capacity of the colloids extracted from 34 
soils, including those given in Table 2, was 0.3 gram water per gram 
of colloidal material. Since variations in the adsorptive capacity 
from the mean value were usually not great, it was suggested that an 
approximate determination of the colloidal material could usually 
be obtained by dividing the water adsorptive capacity of the soil by 
the factor 0.3. This average figure for the water-adsorptive capacity 
of the colloid applies of course to the colloidal material as extracted 
and not as in the soil. Data given in Table 9, however, indicate that 
the average adsorptive capacity of the colloids as in the soil would 
probably be about the same as this factor. Moreover, the average 
value was that of colloidal material graded by a supercentrifuge 
which threw out most of the material between 0.3 micron and 1 
micron. In column 9 of Table 10, are given the percentages of 
colloid calculated by the use of this factor. In the case of four soils 
this method was fairly accurate, but in the case of the Huntington 
soil the result was considerably too low. 
PROCEDURE SUGGESTED. 
It would hardly be justifiable to recommend a procedure definite 
in all its details for determining the colloidal contents of soils by the 
adsorption method until some of the uncertain features connected 
with the method have been further investigated. It seems especially 
important to study further the possible alteration produced by the 
procedure of separating and concentrating a sample of colloidal 
material from the soil. Probably also the methods used for deter- 
mining the adsorptive capacities of the soils and colloids can be 
improved to some extent. The preservation of samples of soil and 
colloid so that they will maintain unchanged their adsorptive capaci- 
ties should be systematically investigated. Miscellaneous data 
obtained indicate that when the samples are kept in an air-dried con- 
dition the adsorptive capacity for malachite green may change in 
some cases. Also one or two samples of colloid preserved moist 
underwent a marked alteration in adsorptive capacity for malachite 
green within a few months' time. 
From results obtained thus far the best method of determining the 
colloidal content of the soil by the adsorption method appears to be 
as follows : The sample of colloidal material should be separated from 
the soil as described on pages 20 to 21. Grading of the particles into 
