INTERRELATIONS AMONG COMPONENTS 409 



Physical exercise is, in brief, an example of physiological ac- 

 tivity that involves numerous components. Two groups of them, 

 examined with respect to loads and net rates during loading and 

 during recovery, are found to be characteristically intercorrelated 

 both in cases where nexuses are recognized and in others. Eecov- 

 ery in each is relative to recovery in the remainder. 



§ 148. Some other components 



In the past the physiological interrelations among components 

 have been studied chiefly: (a) where one chemical compound is 

 formed from others, (b) where cations and anions balance, or di- 

 verse cations replace one another, (c) where energy is transformed 

 from one kind to another. The studies may equally well be ex- 

 tended to components where ' ' there is no reason to expect ' ' mutual 

 associations. As soon as "empirical" correlations have been 

 found, plenty of "reason" will usually be forthcoming. Many 

 physiologists object to the correlation of data except in the light 

 of some theory. To write A = f(B) is, they say, useless unless f 

 is either a constant or a believed causal connection. But it is prob- 

 able that any one relation, usually a partial correlation, between 

 any A and any B, is eventually worthwhile for someone to investi- 

 gate. When he feels the urge to find how A is related to B, that 

 urge can be legitimately satisfied. Indeed, the manners and forces 

 of interactions in organisms seem to be so various that I believe 

 physiology cannot afford to wait for non-empirical hypotheses or 

 visible "connections" to precede measurement. For, meanwhile 

 physiology would be limited by the contents of other sciences. As 

 an instance, relying on the knowledge of chemistry, no one has 

 been able to guess yet how much water is retained in mammalian 

 livers of any species for each gram of carbohydrate (glycogen) 

 loaded. 



All correlations, then, are equally valid, according to any gen- 

 eral criterion that has been proposed, other than opinion. 

 A = f(C) has no more statistical significance than A = f(B), 

 though it may have a higher coefiicient of correlation, or carry 

 smaller probable error, or be known to occur in more species. 

 Only by personal judgments and attractive hypotheses is a greater 

 intimacy assigned to one relation than to another. I doubt whether 

 it is any contribution to physiology to say that ' ' all measurements 

 of the state of the body with respect to water must be relatively 



