Stidies in Soil. Physics, IV. 167 



to be essential for the needed constant, and the sooner we cease 

 believing that it does, and begin to examine critically its use- 

 fuhiessesand its limitations, the sooner shall we be able, not only 

 to find a better constant,hut to attain to a rational use of the 

 mechanical analysis itself. 



In the fisrt place it must be remembered that the whole 

 business is purely empirical. The mechanical composition of 

 a soil doubtless determines its physical properties but the laws 

 according to which the determination is affected are still un- 

 known and quantitative correlations, if they exist, remain 

 imdiscovered. We can do no more than to examine the external 

 properties of soils and determine the corresponding mechani- 

 cal analyses in order that, by the gradual accumulation of such 

 comparative data, we can come more and more to say that a soil 

 of a certain mechanical analysis will probably possess 

 certain properties — which properties we have learned b y e x- 

 perience usually belong to soils giving that, or nearly 

 that, analysis. 



However, this empirical character of the process is un- 

 fortunate mainly because it makes impossible exterpolation 

 or wide interpolation and hence requires that over the whole 

 field of possible mechanical compositions correlations with 

 observed properties be obtained by actual experiment. Except 

 for the extra labor thus required it makes no great difference 

 whether the process and its correlations be empirical or not. The 

 correlations of phosphorous content with breaking strength 

 in steels are useful enough for all that they are empirical. 



The main trouble with mechanical analysis and the main 

 reasonwhy it can never be satisfactory as a physical soil constant 

 lies in the complexity of the final ' 'constant" to which it leads. 

 The mechanical analysis must be expressed as a series of 

 unrelated quantities — the percentages of the particles in the 

 various groups. It can never be expressed in any one unit 

 but must always refer to several — one for each group in the 

 series. And there is no common denominator. It is impossi- 

 ble, for instance, to arrange soils in the "order of their mechani- 

 cal composition ' ' as one would arrange minerals in the order 

 of their specific gravity or wires in order of their electrical 

 conductivitv. This makes mechanical analvsis a verv hard 



