VAEIABILITIES OF WATER RELATIONS 75 



points (Kermack and McKendrick, '37). Here the mean size of 

 run is 2.30 ± 0.18. 



(7) Numbers of points succeeding from one maximum to the 

 next may be observed. This succession is termed a gap, and usu- 

 ally consists of 4 points (Kermack and McKendrick). Here the 

 mean gap is 3.67 ± 0.37. 



Hence, of many possible tests, the five (3 to 7) that have been 

 selected for their wide applicability all indicate randomness in the 

 dog's daily body weights. Regulation, therefore, consists not in 

 steering the body weight from one day to the next, but in prevent- 

 ing wider fluctuations. From a somewhat different aspect this 

 means that one day is too long an interval for the steering to be 

 visible. 



Accordingly, fluctuations at shorter intervals of time may be 

 examined. At one-hour intervals, in which either no food is fur- 

 nished but water is allowed, or both food and water are allowed 

 ad libitum, the fluctuations are no longer random. If no attention 

 were paid to their succession, the non-randomness would not be 

 evident; only the parameter CA = 0.098% would be known. In- 

 stead it may be noted (figs. 45 and 46) that within 4 hours there are 

 no inversions, and the runs are the full length of the series. No 

 longer is the body weight fluctuating ; it is diminishing throughout. 

 That fact leaves two possibilities ; either regulation is absent for 

 several hours on end, or body weight is no longer a measure of 

 water content. Evidence to be presented below indicates that both 

 are factors. 



When observations are extended over a sufficient number of 

 hours, runs succeed one another. The fact that they are greater 

 than 4 hours in length (actually 4.5) implies that inversions of 

 body weight (when the dog drinks) are at such intervals. When 

 the dog is weighed every 0.25 hour, the same result is exaggerated ; 

 the mean run has 18 points. Such an event as drinking is, hence, 

 anything but random in occurrence; instead it comes at rather 

 regular intervals, and by it the dog obtains amounts of water that 

 carry the body through several hours. The fact of periodicity is 

 the clear expression of non-randomness, and indicates a charac- 

 teristic of regulation additional to ** restricted fluctuations." 

 Thus, over daily intervals only fluctuations are evident, but over 

 hourly intervals periodic factors are apparent. The latter reflect 

 the fact that the dog does not sip water every quarter-hour or even 



