TJNIFOKMITIES AND COMPAEISONS AMONG COMPONENTS 363 



ernor. In other exchanges there may be no significant change of 

 standard difference with length of period. This is the case in total 

 heat losses (table 36) ; it may suggest the absence of a stabilizing 

 'inertia" for rate that outlasts the shortest period studied. Or, 

 there may be larger differences between short periods than between 

 longer ones, as in the cases of rates of ingestive water intake by 

 dog or man (table 12) and of superficial blood flow (Burton and 

 Taylor, '40). This feature characterizes some discontinuous proc- 

 ess, as drinking or vasoconstriction, which makes the gain of water 

 or loss of heat periodic. 



Further relations may be found for variabilities by comparing 

 the loads with the rates of turnover of the same component. A 

 variation of rectal temperature of ± 0.77 per cent of 37° C. (A, 

 table 39) is equivalent to a standard deviation of ± 0.29° C. or a 

 heat content of ± 0.24 Calorie/kilogram of body weight. In the 

 interval of one week between determinations of rectal temperature, 

 at least 200 Calories/kilogram have been gained and lost (turned 

 over) by the dog's body; for the whole interval of time, therefore, 

 gain equals loss with a precision of about 1 part in 1000. Or, con- 

 sidering the variation of the dog's body weight, I note that while 

 1.33% of Bo is gained or lost (net), about 40% of the total substance 

 Bo is gained and lost (total) in one week's time; an accuracy of 1 

 part in 30. Here are in figures the knowledge that many have in 

 the rough, that body temperature is more constant than body 

 weight, relative to the whole quantities of heat and of body sub- 

 stance (food and water) with which the metabolic processes deal. 

 Additional species may be compared in table 12. 



How much augmentation of rates occurs, on the average, at the 

 loads corresponding to borderline variations? The coefficients of 

 variation and of difference are by definition such that about two- 

 thirds of instances measured fall w^ithin their latitude. What 

 change in rate of exchange has on the average accompanied this 

 much load? Such comparison (fig. 47) serves to link the equilibra- 

 tion diagrams with the respective variabilities. For water and for 

 heat the net rates of exchange at the o of load are one-tenth to one- 

 fourth of the turnover rates. Should this ratio prove to be nearly 

 constant, it also follows that, when all net equilibration diagrams 

 are put into a uniform proportion between ordinates and abscissae, 

 the steeper the curves (or the greater the value of the net velocity 



