G. W. Robinson :]09 



will thus be conditioned not simply by the actual particle sizes but by 

 the magnitude of the gel coating. The writer hopes to return to this 

 point in a later paper. In the meantime it is suggested that the left 

 hand portion of the curve in the case of clay soils may relate to such 

 coated particles rather than to particles bounded by a sohd surface. 

 By using log v instead of log particle size, the necessity for a decision 

 as to the significance of these small velocities is postponed. 



It is of course obvious that the expression of the mechanical com- 

 position of a soil or clay by a curve of this type offers a way out of the 

 appalhng confusion created by the diversity of conventions used in 

 different countries. Since practically all the methods in use are based 

 on the principle of sedimentation, curves can easily be obtained if the 

 settling velocities are known. Whether the principle is applicable to 

 methods in which separation is effected by currents of water of varying 

 velocity, as in the Hilgard method, the writer is unable to decide. As 

 a first appro.ximation, it would appear that the method is applicable. 



We have assumed that the viscosity coefficient in the Stokes' equation 

 is constant. With varying temperatures this is of course not the case. 

 This difficulty can however easily be solved if a standard temperature 

 be adopted, say, 15° C. By putting the viscosity coefficient of water at 

 that temperature equal to unity and calculating the viscosities at other 

 temperatures'^ with reference to this, the correction can be applied. Thus 

 if results obtained at var3dng temperature are to be compared, it will 

 be necessary to use log (v x specific viscosity) instead of log v. 



Theory of New Method. 



A fundamental assumption underlying all methods of mechanical 

 analysis by sedimentation is that particles in a column settle inde- 

 pendently of each other. That there are Hmits to this assumption is 

 obvious. According to Oden this condition is fulfilled in suspensions of 

 concentration not greater than 1 per cent.'^ Wiegner, on the other hand, 

 brings evidence to show that concentrations of more than 5 per cent. 

 can be used without serious inaccuracy, which may be due to the 

 dangerous principle of compensating errors. 



Let us assume a suspension of soil or other granular material to 

 consist of a number of fractions, a, b, c, etc., each uniform in itself, 

 having hmiting velocities, i\, i\, v^, etc., respectively, and present in 



' For the effect of temperature on the viscosity coefficient of water, see Hosking, 

 Phil. Mat). 1907, 17. 509; ihid. 1909. 18, 260. 

 - Int. Mia. Bodenlninde, 1915, 5, 276. 



