CORRELATIVES OF WATER CONTENT AND EXCHANGES 



277 



conditions (table 31). It turns out that two species of deer-mice 

 differ significantly in voluntary water intake, though reared in the 

 laboratory upon constant diet. The difference was 32 per cent of 

 the mean water intake, and thrice the probable error of each mean. 

 Two subspecies of either species, however, drank like amounts of 

 water. Such a difference between species calls attention to but one 

 of the large number of ''constitutional" factors related to water 

 exchanges in any class of animals. So far as the meager facts indi- 



TABLE 31 



Rates of intake of free water compared in five subspecies of deer mice (Peromyscus) . 



Paired tests between and within species. All individuals were bred in 



captivity. Data of Boss ( '<30) 



cate, races and subspecies (as classified on the usual criteria) are 

 not sufficiently different physiologically to exchange diverse 

 amounts of water in turnover. 



<§. 99. Summary 



It turns out to be possible to deal with many variables either 

 separately or together, when all relate to some one or two variables 

 (SW, AW/At) that are kept continually in view. The procedure is 

 to classify correlatives according to dimensions and in other arbi- 

 trary ways. Further, within the classes certain factors are 

 mutually exclusive {e.g., two diets, two atmospheres), and hence 

 each of those could be considered independently of its alternatives. 



No basis is found for believing that variables not given special 

 consideration are qualitatively different from those mentioned. 

 For the most part the variables studied were thought of earlier in 

 the development of physiological science, or were recorded handier. 

 So far as I can judge, size of body, or rate of nitrogen turnover, is 

 just as ''intimately" related to rate of water exchange as is refrac- 

 tive index of blood plasma, or body water content itself. But the 



