THE WATER OF SOILS. 2OQ 



can exist only in a very short (strictly speaking, an mfinitesimally 

 short) vertical column. The least practicable height for experimental 

 work being about i cm. (| in.), the writer has adopted for the purpose 

 of rapid determination of this factor, the use of a brass cylinder i cm. 

 high and of such width as to contain, for the sake of convenience, 25 

 or 50 cm. of soil. This cylinder has a finely perforated bottom, which 

 may be covered with filter paper; after being filled with soil which ha? 

 been struck level, and weighing, it is immersed to i mm. depth in dis- 

 tilled water and allowed to rest for an hour ; then quickly dried outside 

 and beneath with filter paper, and again weighed. The amount of 

 water found by difference should for all practical purposes be referred 

 to the volume, not to the weight, of the soil, so as to eliminate the 

 error arising from the varying specific gravity of the latter. 



In most cases the surface of the soil in the sieve cylinder remains 

 level after wetting ; but sometimes it swells so as to rise above its dry 

 level, even to the extent of nearly 30% (see chapter 7, p. 114). 

 This happens especially in strongly ferruginous soils. In the case of 

 "black alkali" soils, in wetting an enormous collapse sometimes takes 

 place (see chapter 22). 



If it be desired to determine also the minimum liquid absorption 

 (see below), the surface of the wet soil is first covered with air-dry soil, 

 to absorb the surplus moisture, and finally with soil previously saturated 

 with hygroscopic moisture ; the added soil being each time thrown off 

 and finally the surface " struck " level with a tense silk thread before 

 weighing. Corrections must be applied for the usual increase in 

 weight, from the addition of soil, and for the hygroscopic moisture. 



While the minimum of liquid absorption can thus be deter- 

 mined quickly, without awaiting the capillary ascent of a water 

 column, and if sufficient time is given can also be determined 

 in higher columns, as proposed by Mayer (Wollny's Forsch. 

 Vol. 3), the maximum cannot thus be determined without 

 gross inaccuracy. In determinations made by the writer it 

 was found that the figures for the minima of very different 

 soils (clayey and sandy) of the arid region, differ proportion- 

 ally much less than do the respective maxima. In few of 

 these soils it was found to exceed about 10 per cent, and it 

 scarcely fell below 4 per cent even in very sandy soils. A 

 very deep, sandy soil, which had been irrigated in May, and 



