T. Dkighton 



2-23 



when this liad fallen to some point but shghtly above room temperature, 

 the same temperature being employed throughout any one series of 

 determinations. A sample was taken out from the middle of the mass 

 after each resistance reading and the moisture in it determined by drying 

 at 110° C. About 15 to '20 determinations were made in this way for 

 each sample. It was not, however, found possible to carry the moisture 

 content up to saturation since near this point the water tended to oose out 

 from the bottom of the bo.x. The results obtained with one soil are given 

 in Table VII. If plotted out they yield hyperbolas, but if, instead of 



Table VII. Result.s of Resistance Moisture Detenu 'mat ions on a 

 tijpical soil (Greensand). 



Hygroscopic moisture 1-4 % 



plotting directly, we plot the logarithm of the moisture expressed in 

 percentage dry weight, less the hygroscoi^ic moisture, against the loga- 

 rithm of the resistance observed we obtain the curves shown in Figs. 

 4, 5 and 6: whence it is clear that at low moisture contents the curve 

 is discontinuous at any rate at one point, the existence of the first point 

 of discontinuity is disputable owing to the small number of observations. 

 It will be observed that the type of curve is different for the artificial 

 mixtures to what it is for the natural soils and sands taken. In the 

 former case the resistance falls sharply at first, then less so and finally 

 sharply again, in the latter the reverse holds — slowly at first, then more 

 quickly, and slowly at the end. 



It would appear that there ought to be no sharp angle between the 

 different regions as shown in the curves. It was felt however that as 

 the accuracy of the measurements was not sufficiently great to justify 

 the use of them as a basis for the discussion of this matter greater 

 clearness would be attained by drawing the fines as shown. 



The facts seem explainable as follows : — In the case of pure sand and 

 artificial mixtures we may assume that there is no measurable amount 



Joum. of Affric. Sci. xii 16 



