56 



Prof. H. H. Dixon. 



[Nov. 7 ; 



and the freshly cut inner end of the branch. A tongue of bibulous paper is 

 applied to the outer end to draw off the transmitted water. The head once 

 adjusted is kept constant ; but after each determination the branch is 

 shortened. Fig. 4 records such a series of experiments. The head through- 

 out these was 100 cm. The initial length was 25 cm. At that length the 

 flow was 1*18 mg. per second. Five centimetres were then cut off the outer 

 end and the flow rose to 1*66 mg. per second. A shortening to 15 cm. 

 increased the flow to 2*33 mg. When the branch was 10 cm. and 5 cm. long, 

 the flow was 3*79 mg. and 6*70 mg. respectively. The curve plotted in fig. 4 



7o 



( 



> 









6-0 























SO 





\ 









40 



i 













■\ < 



> 



















l-o 



Ox 





*'•< 



».... 























■■'•■• 



►-. 



1-0 



- 









o 



?vO 



%S 



Fig. 4. 



is a rectangular hyperbola in which M = K/l; M being the number of 

 milligrammes transmitted per second, I = length of wood, and K = average 

 value of the product Ml observed. The observations, it may be seen, 

 approximate fairly closely to this curve. When the weights of transmitted 

 water are plotted against units head the curve shown in fig. 5 is obtained. 

 Here the proportionality of flow to head, or, rather, the inverse proportionality 

 of flow to length, is immediately apparent. Up to 10 units the curve is 

 almost a straight line. The bending over which occurs after that point is to 

 be attributed to the clogging, which is practically unavoidable when the flow 

 is rapid. 



In conclusion, it seems to me that Ewart's objections to the cohesion- 

 theory, based as they are on an excessive estimate of the velocity of the 

 transpiration current and of the resistance to this current in the conduits, 



