EXPERIMENTAL STUDY OF HEAT LEAKAGE. 785 



and it is therefore to be expected that the — , vs. - plots will exhibit 



curvature. They all do, but the curvature is least distinct for plug 

 A4, and, owing to the way in which the experimental errors distribute 

 themselves, the curvature of the A4 plot might almost be considered 

 accidental. It is interesting to note, in this connection, that if either 



of the n' vs. -or the — vs. plots is straight, the other will be concave 

 J 1^ J 



upward ; and that a rectilinear extrapolation of a ijl' vs. . plot which 



is slightly concave upward will lead to too small a value of fx, while 



an extrapolation of a similar — vs. - plot will lead to too small a value 



M / 



of — i. e., to too large a value of /x. Hence if, in anv given case, both 



plots seem sensibly straight, it might be expected that the true fj. 

 would lie somewhere between the two values obtained by these extra- 

 polations. In the case of plug A4 the mean of these two extrapolated 

 values is 3°.190 C. cmVkgm, which is within 0°.008 C. cm.-/kgm. or 

 I of 1% of what has been taken as the true fjt. in this discussion. 



(1. Radial flow plugs Vl and V2. The regeneration effect. 



The plotted points on the ji' vs. - diagram for plug V2 (no lagging) 



and, possibly, also for plug VI (external poplox lagging) are so situ- 

 ated as to admit the possibility of some real curvature in these plots, 

 the curvature being negative (concavity downward). This curvature, 



reversed, is more evident in the — vs. - plots for these two plugs. 



It is probable, therefore, that rectilinear extrapolation of either plot 

 for either plug is of rather doubtful reliability, in spite of the com- 

 paratively small slope of the lines concerned. However, the devia- 

 tions of the plotted points from the least square straight lines are not 

 too large to be ascribable to accidental errors, at least in the case 



of the /x' vs. -plots, and even if they were, and if some better represen- 

 tation could be devised, the plots in their present form are useful for 

 comparison purposes. They are therefore shown. 



