ABSORPTION BY SOIL CONSTITUENTS. 5 
natant liquid is practically the same as that of a check solution 
containing 10 cubic centimeters of the aforementioned dye solution 
in a total volume of 50 cubic centimeters. This is to insure the 
absorbed dye being in equilibrium with practically the same con- 
centration of nonabsorbed dye in each determination. The suspen- 
sions are then shaken one hour in an end-over-end shaker to insure 
complete absorption of the dye by the soil. This time is probably 
longer than necessary in most cases. Sufficient normal sodium 
cliloride solution to coagulate the colloid is then added to the sus- 
pensions; usually 5 cubic centimeters is sufficient. After a few 
minutes settling the suspensions are centrifuged in order to throw 
out any solid material and the supernatant liquid is read colori- 
metrically against the check solution. 
ESTIMATION OF NONCOLLOIDAL ABSORPTION FROM SOIL FRACTIONS. 
METHODS OF SEPARATING SOIL COLLOIDS. 
In separating the colloidal matter from the larger soil grains two 
difficulties are encountered that are inherent in all systems of me- 
chanical analysis: First, the difficulty of obtaining a complete defloc- 
culation of the ultimate soil particles: second, the difficulty of sorting 
out the particles of different size after they have been deflocculated. 
In the system of mechanical analysis heretofore practiced in the 
Bureau of Soils {10), treatment with a few drops of ammonia followed 
by long shaking have been relied on to bring about deflocculation. 
Williams (26) adopted long continued boiling supplemented by 
rubbing as the most efficient method of deflocculation. Atterberg 
(3) in deflocculating the soil for his system of mechanical analysis, 
utilizes rubbing and treatment with an alkaline solution of bromine, 
with hydrochloric acid and sodium hydroxide, or with concentrated 
nitric acid, according to the nature of the soil. Konig and Hasen- 
baumer (l^) suggested boiling for a short time and repeated rubbing. 
Wliile all these methods probably deflocculate and form good 
suspensions of part of the colloidal matter, it is not certain that any 
of them deflocculates all of it. Some colloidal matter adhering to 
mineral particles may be difficult of separation, and we have evidence 
in our work that the colloidal matter which is present in the dry soil 
as more or less indurated aggregates is especially resistant to dis- 
persion. In our attempts at separation we found the use of ammonia 
and rubbing with a rubber pestle to be especially effective. For 
instance, two soils when rubbed with distilled water yielded only 
5 per cent and 29 per cent of colloid; but when the residues were 
rubbed with distilled water containing a trace of ammonia further 
yields of colloid, 3.5 per cent and 13.5 per cent, respectively, were 
obtained. Other results show the effect of rubbing. A soil that 
had been agitated 15 times with water containing ammonia failed 
to give further colloidal suspensions. However, after the pasty 
residue had been gently rubbed, succeeding treatments with ammonia 
water yielded additional amounts of colloid. 
Separation of the finer particles from the coarser, after defloccula- 
tion has been produced, is brought about by subsidence in most 
methods of mechanical analysis. A much quicker separation, 
however, can be made by use of a centrifuge. In this particular 
investigation we made use of both subsidence and centrifuging. 
