METHODS. 



21 



the only part of an experiment of this type which has been used, and upon 

 which conclusions have been based, has been the latter half. 



That the error found in the above experiment is properly explained appears 

 true from the data derived from a similar experiment carried on at the same 

 time, in which the rate curve between 9 p. m. and 3 a. m. follows the normal 

 course, as indicated by my experiments in general. 



TABLE 12. Ocotillo Error due to use 0} a potometer. 



In the experiment displayed in table 12, approximately the same error 

 occurs, 2.3 grams of water having been held back by the plant in the first 8 

 hours. After 10 a. m. it was noted that the piece commenced to wilt rapidly, 

 the jate falling hourly as follows: 11.2, 8.3, 7.3, 5.6, 4.1 grams, and evidently 

 lost more water than was taken up. The total error, however, was within 

 3 per cent and was distributed over several hours. 



TABLE 13. Tests made with Verbena ciliata, June 14 and 16, 1905. 



Table 13 embodies the data derived from a similar test made with Verbena 

 ciliata, in which plus and minus differences were observed. Doubtless such 

 errors of observation creep in when a single observer tries to make as nearly 

 as possible simultaneous readings and weighings. Nevertheless the error 



