AS INDICATED BY OSMOMETERS. 77 



our collodion one, backed by an osmotic solution) be placed against 

 the new soil surface thus formed, there should then be no alteration 

 in the surface tensions effective at the gas-liquid surfaces, for such 

 a membrane is to be thought of as saturated with imbibed water and 

 covered on its exposed side by a water film. Our system would remain 

 in static equilibrium if no movement of water were to occur through 

 the membrane. But water does migrate from soil to osmometer 

 solution (we need not here complicate the argument by attempting 

 to analyze this process in detail), and the result of such migration, in 

 its incipient stage, must be to decrease the thickness of the soil-moisture 

 films lying nearest to the absorbing surface and to enlarge the gas- 

 liquid surfaces. In this process the water-attracting force exerted by 

 our sugar solution is opposed by the tension of the water surfaces 

 abutting upon gas bubbles, the water being held to the soil grains by 

 the greater force of adhesion. Neither sugar nor soil particles can pass 

 the membrane, hence it is clear that the water movement is dynami- 

 cally dependent upon the relative magnitudes of the diffusion tension 

 of the solution, on the one hand, and of the gas-liquid surface tension 

 on the other. If the attraction of sugar solution for water were not as 

 great as that of the opposed force, then water should move in the 

 opposite direction, and the soil moisture films must thicken. It is not 

 probable that the osmometer ever reduces the adjacent soil j&hns to the 

 hygroscopic state. 



With the very inception of water absorption, then, the attraction 

 for water exhibited by the thin layer of soil next to the membrane is 

 more or less markedly increased. Before this occurrence, however, 

 our hypothetical system was supposed to be in static equUibrium, as 

 far as water movement is concerned, and now we find that the sUghtly 

 dried soil layer is attracting water more strongly than the more distant 

 portions of the soil mass. After such a disturbance equilibrium must 

 tend to recur, which must imply water movement from more distant 

 soil films to those which have been thinned. Thus the drying process 

 is extended outward from the absorbing surface, and our process of 

 absorption has now come to involve not only movement from soil to 

 osmometer, but also movement from soil film to soil film. The latter 

 phenomenon and its dynamics are apparently of fundamental impor- 

 tance to an appreciation of the water relation between plant and soil. 



The migration of water from one part of the soil to another — from 

 soil film to soil film — ^must of course take place through films already 

 more or less thinned, and the thinner are the films involved the greater 

 should be the force requisite to produce such movement at any given 

 rate. This means that such movement must be exceedingly slow when 

 the films are much thinned and, in any event, when its path becomes 

 of considerable length. It must thus soon come about (usually within 

 the first one or two short periods in our tests, apparently) that the 



