788 TRUEBLOOD. 



rectilinear extrapolation of the /x' vs. plot for this plug, is the true 



value within | per cent., may be summarized as follows: 



1°. The results show that the set-up was to a large degree suc- 

 cessful in avoiding the sources of error it was designed to avoid. The 

 ordinary conductive heat leak is very small and, as already noted, 

 there is no evidence of regenerative leak. The actual percentage of 

 the true fi which the extrapolation is depended upon to supply is only 

 1.3 per cent. 



2°. It is very probable that the true value of fx lies somewhere 

 between the values given by plugs VI or V2, which are too high be- 

 cause of regeneration (or possibly because the fj.' vs. - plots for these 



plugs are really concave downward instead of rectilinear — and, if 

 this is so, it is doubtless itself due to regeneration), and the value as 

 given by the fx' extrapolation for plug A4, which is too low. (Plug A4 

 is the best of a series of axial flow plugs, all of the earlier of which give 

 extrapolated /i's which are certainly much too small.) The extra- 

 polated ju's for plugs VI and V2 are respectively 3.224 and 3.230 while 

 that for plug A4 is 3.149. Also, it has already been noted that the 

 mean /x as determined from the two A4 plots differs by only 0.008 — 

 about J per cent. — from the extrapolated fx of plug S. 



3°. Where the slopes of the AT vs. Ap plots reach, or approxi- 

 matel}' reach, a limit at or below the highest pressure drops, these 

 limiting slopes may be regarded as representing approximately the 

 true value of /x. If the calculations of slope are made by subtracting 

 one observed temperature drop from another and dividing the result 

 by the difference of the corresponding pressure drops, it is found that 

 none of the set-ups except VI, V2 and S show evidence of a limiting 

 value of the slope such as has just been mentioned. In the case of 

 these three plugs, the values of the slope, calculated as described from 

 the two highest pressure drops, are 3.20, 3.18, and 3.18 degrees C. per 

 kgm. per cm.^, respectively. These values would probably be slightly 

 changed if the smoothed curves, instead of observed points, were used 

 as a basis for the calculation ; but in either case, the method of calcula- 

 tion would necessarily be somewhat inaccurate, and the results are 

 offered only as confirmatory of the result (3°.182 C.cm.Vkgm.) ob- 

 tained by extrapolating the /jl' vs. - plot for plug S. 



4°. The /x' vs. plot for plug S has the least slope of all such plots, 



