112 TIMBER AND SOME OF ITS DISEASES, [chap. 



side, and since the friction between the wood-wall and 

 the water, as well as the resistance to filtration, are 

 almost infinitely small, the water in the columns, 

 supported in the long chains by the air-bubbles, 

 moves. If no more water is being transpired than is 

 absorbed, the air-bubbles themselves do not act in the 

 process ; but now suppose more water passes off at 

 the leaf-surfaces than is supplied at the roots. In this 

 case the air-bubbles must expand (explaining Hartig's 

 discovery that the air in the upper parts of the tree is 

 rarer) and a " suction " is started, and propagated 

 downwards, accelerating the flow above, and extending 

 its action — which becomes more and more feeble 

 downwards — till it splits into smaller currents at the 

 roots. 



Hence, according to Elfving, the opponents of the 

 imbibition theories are right in saying that (i) the 

 water filters from element to element, and (2) that the 

 tension of the air-bubbles co-operates ; but they are 

 wrong in supposing (i) that the pressure of the atmo- 

 sphere can be effective, or (2) that the tension of the 

 air-bubbles only makes the pit-membranes permeable. 



In Dufour's experiments with sawn bi-anches, he 

 ignores the effects of cutting in air — the more rapid 

 the transpiration the more quickly air will pass in and 

 prevent water under pressure from passing through, 



