786 TRUEBLOOD. 



Both of the n' vs. - plots, on this rectihnear representation, extra- 

 polate to values of m which exceed the true value by about 1 per cent., 

 the result for the unlagged plug (V2) being slightly greater than that 

 for the other. (Both extrapolated values would of course be too large 

 if there is genuine negative curvature in the loci.) The slope is 

 greater for the unlagged plug than for the other. 



The AT vs. Ajj plots exhibit definite characteristics unlike those of 

 any other plug. Both are concave downward within the region of 

 observation, the heat-leak per gram of fluid decreasing with increas- 

 ing pressure drop, with a well-marked tendency to approach a limiting 

 value which is nearly zero for VI, the lagged plug, and which, for V2, 

 is small and of about the same magnitude as the corresponding leak 

 for plug S. Both plots would show points of inflexion if observations 

 had been taken at smaller pressure drops, since AT and A^ vanish 

 together; this is indicated by the dotted portions of the plots. 



This peculiarity is believed to be due to the operation of the regen- 

 eration efPect discussed under II, 2, a. As is there pointed out, this 

 effect produces an outward heat-leakage from the steam between the 

 thermometers, the effect of which is to oppose the normal inward 

 heat-leakage due to conduction. The external lagging used with 

 plug VI assists the operation of this regenerative action, since it inter- 

 feres with the flow of heat from the bath to those portions of the 

 passing fluid upon which the regenerative action takes place. It is 

 on this account that the net inward heat-leak is larger for the unlagged 

 than for the lagged plug. It seems probable that the total outward 

 heat leak due to regeneration Avould increase at about the same rate 

 as the pressure drop, and since the flow varies approximately as the 

 I power of the pressure drop, the outward heat leak per gram of fluid 

 due to regeneration would increase with the pressure drop; moreover, 

 since the inward leak per gram of fluid due to conduction, etc., 

 increases with pressure drop in the case of every plug in which this is 

 the only leak occurring (including plug S), it is evident that the net 

 result might be to produce a heat-leak per gram which would tend 

 toward some limiting value such as the experimental results actually 

 indicate. It is hardly possible to be more definite than this regarding 

 the effect which might reasonably be anticipated from the joint opera- 

 tion of the several agents at work, because of the complicated nature 

 of the phenomena concerned. 



The fji' vs. - plots also point, though less clearly, to the operation of 



