72 TRANSPIRATION AND ASCENT OF SAP ch. 



from 0*450 gram to 0*500 gram per 10 min. If now one 

 branch be killed, the vital lifting-force, if present, will be 

 removed, and we should expect the amount transmitted 

 by the killed branch to increase correspondingly. This 

 increase, even though small, would be easily seen by com- 

 paring the flow through the dead and the living branches, 

 when both have been again brought to the same conditions. 

 For killing the branch I used either a jacket of steam or 

 an injection of poison. 



When it was desired to kill the experimental branch 

 with steam, the water in the cistern was run off through 

 a small side-tubulure (c), and, when the cistern was empty, 

 steam was passed by the same tubulure into a wide tube 

 (d), now placed round the experimental branch, and fitting 

 tightly into a socket made for its reception in the bottom 

 of the cistern. The space round the top of this branch, 

 and between it and the tube, was packed with cotton- 

 wool. The supply of steam was kept up for 20 min. ; the 

 tube (d) was then removed ; and the cistern was filled 

 with water through the tubulure (c). After some time, 

 during which the water in the cistern was kept stirred, 

 when it was judged that the experimental and control 

 branches had come to sensibly the same temperature, 

 measurements of the amounts transmitted by each were 

 resumed. In this way it was easy not only to compare 

 dead and living branches under the same conditions, but 

 also rough manipulation and shaking were avoided. These 

 latter are known to cause irregularities in the amounts of 

 water transmitted by cut branches, probably owing to the 

 displacement of, or compacting of, some clogging material 

 on their upper surfaces. 



Observations made in this way showed in each case that 

 the amounts transmitted by a branch before and after 

 killing by steam were sensibly the same ; or, if they 

 differed in the experimental branch, the same difference 

 was observed in the amounts transmitted by the control 



