Studies in Sou. Piiysics, IV. 165 



defiinition altogether, is unimportant. Physical character 

 remains something which cannot be measured directly and 

 which we must examine, if we examine it at all, by devious 

 and indirect paths. Nevertheless some quantitative express- 

 sion or indication of physical character seems quite necessary 

 to further advance of the science. The actual, momentary 

 phvsical condition of a soil is indeed susceptible of some 

 quantitative examination and perhaps of sufficiently accurate 

 measurement, but this does not touch the core of the matter. 

 It is necessary to know, not what the soil is temporarily like, 

 but what it can belike — not its exhibited properties but its 

 potential properties. Within limits physical condition is un- 

 der human control; physical character is very seldom so, and it 

 is phvsical character that we most need to learn to know. This 

 iG— phvsically — the fundamental property of soils. 



But physical character is not even accurately decipher- 

 able let alone measurable. It is probably far too complex ever 

 to be measured by one process or in one unit. We must fall 

 back upon a subterfuge. If we cannot measure ths thin^ itself 

 we must select something dependent upon it and measure that. 

 We want some measurement which will indicate, if it does not 

 completly express, the more fundamental physical character. 

 The electrical engineer has certain curves which express the 

 behavior and cabapilities of a dynamo, Avhich curves he calls 

 the "characteristics" cf the machine. We want a physical 

 "characteristic" of the soil. I have spoken, and shall speak, 

 of this "characteristic" as a "physical constant"for it must 

 bear a constant relation to physical character and must be, 

 for any one soil, fully constant and reproducible. 



Of the qualities \\ hich must be possessed by this constant 

 the first is simply this constancy of relation to the ultimate 

 physical character of the soil, regardless cf the condition it may 

 happen to be in at the moment. Our constant must not be 

 affected, for instance, by accidental variations in the treat- 

 ment of the sample or by its recent history. Secondly, the 

 constant must vary closely and in rigid correspondence v^ith 

 changes in the physical character and its variations must be of 

 sufficient magnitude to enable the detection, through them, of 

 important differences in this physical character. Thirdly the 



