METHODS. 19 



to disturb the parts. Furthermore, it was difficult to get records of a large 

 enough series of experiments carried out through a long enough period without 

 the most laborious business of making very frequent readings at unequal 

 intervals, a burden quite out of proportion to the amount of data gained. 

 For my purpose it was quite necessary that I should have evidence that the 

 parts studied were behaving normally, and, under the circumstances, the 

 only way to be sure of this was to have a number of experiments going on 

 at the same time under the same conditions, so that each might act as a con- 

 trol on all the rest. I was led to meet this demand by substituting, in a 

 particular experiment, a vertically placed burette for the horizontal calibrated 

 tube when I wished it to run all night. Since the supply of burettes happened 

 to be limited, I substituted, for the time, ordinary glass tubes. L,ater on, I 

 used blue-line burettes which could be read quickly and accurately. The 

 arrangement used then was this : A glass tube or burette of suitable length 

 was attached to the cut end of a piece of ocotillo stem by means of rubber 

 tubing, always with the precaution that the cut surface of the stem was clean. 

 The possible stoppage of the pores of the wood was avoided, at least for the 

 time during which the piece was to be used, by a filter consisting of a firm 

 plug of absorbent cotton placed in the rubber tube in contact with the cut 

 surface. It was found also that it was important to trim off the green cortex, 

 which quickly decayed when exposed to the water, so that only the wood was 

 inserted into the rubber tubing. 



Undoubtedly, as Ewart (1905) has recently shown with more accuracy 

 than heretofore, the rate of flow of water through cut stems is gradually 

 reduced by clogging bacteria, probably both in the cut surface and throughout 

 the length of the piece. Having had due regard to this circumstance, there 

 is abundant reason for believing that potometer experiments conducted 

 within reasonable time limits (about 24 hours) and set up carefully, as above 

 described, give accurate enough results for my particular purpose. 



The piece of stem having been cut under water and adjusted to the poto- 

 meter, it may then be supported in a vertical position by tying it up to the 

 tube and the whole suspended from a rack ; or the piece might equally well be 

 allowed to hang down, as comparative readings showed no difference in the 

 behavior in the two positions. It will be seen that the number of separate 

 experiments which might be set up was limited only by the materials avail- 

 able and the time necessary to make the readings. It was found by experi- 

 ence that one can hardly attend profitably to more than from 8 to 10 such 

 (plate 3, fig. 2), hence I confined myself to the results of several series in these 

 numbers. 



The results obtained from a battery of potometers were in almost all cases 

 so uniform that I have little hesitation in saying that they depart but little 

 from those which would be obtained by any other method, at least so far 



