A NEW METHOD FOR THE MECHAXTCAL 



ANALYSIS OF SOILS AND OTHER 



DISPERSIONS. 



By gilbert WOODING ROBINSON, M.A. 



Adviser in Agricultural Chemislry, University College 

 of North Wales, Bangor. 



Introduction. 



Attention has been given of recent years to the possibility of devising 

 methods of mechanical analysis which shall express the mechanical 

 composition of a soil or clay by a continuous curve. The standard 

 methods of mechanical analysis, as for example that followed in England, 

 can of course be used to obtain such curves, but with the comparatively 

 small number of fractious separated, very little detail can be inserted. 

 Any multiplication in the number of fractions nnist ine\'itably be 

 attended by an increase of experimental errors, and the problem must 

 therefore be attacked by other methods. 



Od^n has devised an elegant method whereby the mechanical analysis 

 of a soil or clay can be derived from an experimental curve showing the 

 rate of accumulation of sediment from a column of material in sus- 

 pension^. The weight of material is automatically registered from time 

 to time and from the curve obtained a mass distribution curve is derived 

 which gives a detailed representation of the composition of the material 

 under examination. Wiegner'" has applied the theoretical principles of 

 the Oden method in a very simple apparatus consisting essentially of 

 a U-tube system in which a column of sedimenting suspension is balanced 

 against a column of pure water. As the material falls below the point 

 at which the water tube joins the fall tube, the specific gravity of the 

 suspension decreases, resulting in a corresponding fall in the water 

 column. The lieight of the water column is read from time to time and 



» Int. Mitt. Bodenkunde, 191"). 5, 2.57-311 : Koll. Zeit. 19U), 18, S.-J-^tS; Tram. Faraday 

 Soc. 1922, 17. 327-348: Nefedof, ./. Exp. Lanilw. 1902. 3, 421-449, (nitlin.s a method 

 similar in prinriple to that of Oden. but apparently purely empirical. 



• Lantlio. Vrrstirh.i SIfnf. 1918,91,41. 



