1885.] 0-n the transpiration-stream in cut branches. 



.:',: 



will be introduced. The same considerations mutatis mutandis 

 apply to barometric changes. 



It must further be remembered that the use to which we have 

 put the potometer is especially to test the effect of sudden changes 

 in the conditions ; and that where we record changes in rate of 

 absorption, which are observable from minute to minute, errors of 

 this sort need not be considered. 



Tests applied to the apparatus. 



We were not able to discover any serious sources of error in 

 the apparatus, but we were anxious to prove that the readings 

 of the stop-watch do represent the flow through the potometer. 

 We therefore substituted a siphon at a in place of the plant, and 

 proceeded to compare the different rates of flow of water through 

 the siphon with the corresponding readings of the potometer. It 

 was found that low pressures of the siphon corresponded to what we 

 were accustomed to consider high readings of the potometer. And 

 as we further found that the liability to error is greater with 

 rapidly moving bubbles, we may assume that our experimental 

 readings are more and not less trustworthy than those obtained in 

 the testing experiments. 



The siphon was so arranged that the free end could be raised 

 and lowered, and the experiment here given begins with a pressure 

 of 100 mm. of water: — that is to say, the efficient column of water 

 was 100 mm. in height; as the column was increased by regular 

 increments to 200 mm., the reading of the stop-watch decreased 

 from 7-3" to 3 5". 



In the following table A gives the reading of the stop-watch ; 

 B gives a series of figures proportional to the reciprocals of the 

 figures in A ; C gives figures proportional to the pressure of the 

 siphon, — figures which therefore approximately represent the 

 outflow from the siphon : 



A 



B C 



100 

 120 

 140 

 160 

 180 

 200 



It will be seen that the difference between B and C never 

 reaches 5°/ . 



With still higher pressures and with a correspondingly rapid 

 rate in the passage of the bubble the error increases from 15 to 



