G. W. Robinson 307 



by appropriate calculation the experimental results can be thrown into 

 the form of curves similar to those obtained by Oden's instrument. Both 

 Oden's and Wiegner's methods have found apphcations in the study of 

 dispersions 1. 



A serious drawback to Oden's apjaaratus is its expense. While it is 

 of the highest value for the critical investigation of comparatively small 

 numbers of samples, it can hardly come into use for routine work as 

 the resources of an ordinary provincial institution would scarcely be 

 equal to the outlay involved in setting up more than one such instru- 

 ment and only comparatively small numbers of soils could be dealt with. 

 Wiegner's apparatus is cheaper and could b}' modifications^ be made to 

 give results of considerable accuracy. But here the time factor is serious. 

 Using, as is necessary, a column about a metre long, several days would 

 be required to obtain information as to the finer fractions which exert 

 such an important effect on the properties of the soil. 



In the present work the writer has attempted to devise a method 

 capable of giving more detailed data than the ordinary sedimentation 

 methods and which, though not giving continuous curves, avoids the 

 drawbacks of the Oden and Wiegner methods and can be used for 

 standard mechanical analysis with great saving of time. 



Before discussing the new method of mechanical analysis a few 

 remarks may be made on the expression of mechanical composition by 

 means of curves. The simplest method of expressing the results of a 

 mechanical analysis is to plot summation percentages* against particle 

 sizes. Such a method is however almost useless in the case of highly 

 dispersed substances, because in order to show the complete range the 

 most characteristic particle sizes are cramped together near the «/-axis. 

 A better distribution will be obtained by using the logarithms of particle 

 sizes. The end of the curve corresponding to zero size is of course at 

 — 00 but in practice a very manageable type of curve will be obtained^. 



Since the separation of particles of diameter smaller than -2 mm. is 

 universally based on the principle of subsidence, whereby fractions are 

 distinguished by their different setthng velocities in water, there are 

 good reasons for using logarithms of setthng velocities instead of 

 logarithms of particle sizes. In the ensuing treatment this method is 

 used, the velocities being calculated in ordinary c.G.s. units. The same 



1 Roll. Zeit. 1920,26. 100-121; ibul. 1920,26, 121^138. J. Landw. 1921, 69. 5-32. 



- As for instance in the method of reading the lieight of the water column. 

 ' i.e. percentages of material of a given particle size or smaller. 

 * Ct. Whittles, /. Agric. Sci. 1922. 12, 166-181. 



