WATER-RELATION BETWEEN PLANT AND SOIL. 31 



of actually disconnecting the reservoir at each weighing, there occur 

 in the columns for absolute transpiration several cases where there 

 was an apparent negative transpiration, as though the plant had 

 absorbed water out of the air. These are to be regarded as resulting 

 from accumulated errors of weighing, changes in the rubber tube, etc. 

 The discrepancies which would thus be introduced in the derived series 

 have been avoided by considering these values as zero, such assumed 

 values being indicated in the tables by italics. Thus, for example, in 

 table 2 the observed value of the transpiration rate for hour 24, Feb- 

 ruary 23, is —0.9, so that no satisfactory data are available for this 



T 

 hour in the columns for T and -^, and these latter data are assumed to 



Hi 



be 0.0. Since the rates with which we are dealing are all low during 

 the night hours when such discrepancies occur, and since only approxi- 

 mate accuracy is here attempted, the errors thus introduced are 

 regarded as negligible in the present preliminary study. The minima 

 and maxima, for the 24-hour period, in the derived columns are indi- 

 cated by full-face type. At the end of each series are given the respec- 

 tive average deduced values for the 24-hour period beginning with 

 hour 10 and ending with hour 9 of the next day. 



DISCUSSION. 



Since attention is to be directed to the daily fluctuations in the var- 

 ious values given in the preceding tables, it will be advantageous to 

 study the graphs^ of relative and derived data, which are shown in 

 figures 1, 2, and 3, instead of studying the tables 2, 3, and 4 themselves. 

 These figures show the five different double graphs, placed one above 

 another, it being obviously impossible clearly to present them all as 

 plotted from the same horizontal line. It is to be borne in mind, how- 

 ever, that all five double graphs for the same plant are comparable as 

 to their relative forms. The horizontal hne below each double graph 

 represents the common axis of abscissas for both component graphs. 

 The vertical lines continuing through all five graphs in each figure 

 represent hourly increments of time (marking the middle of the hour 

 represented), and on them have been laid off (upward from the respec- 

 tive horizontal axes) the various ordinate values, taken directly from 

 the tables. Thus, each graph represents the rate of increase or decrease 

 of the particular power or rate to which it refers, with respect to time. 

 The time period is 24 hours in each case, beginning with hour 10 of 

 one day and ending with hour 9 of the day following, the hours of the 

 day being simply numbered from 1 to 24, without resorting to the con- 

 ventional repetition after hour 12. Each double graph consists of two 

 single ones, one for the February period of study and the other for the 



^These graphs have been prepared by Mrs. Grace J. Livingston. 



