784 TRUEBLOOD. 



per gram of .fluid than the others; this is due to the better insulation 

 of the cross-channel in this plug. 



The n' vs. - plots for these three plugs indicate clearly that the 



range of observation would have to be much greater than that actually 

 covered by the experiments to justify any expectation of reliable 

 results, and that, if the range were extended, these loci would neces- 

 sarily be curved and hence that their extrapolation would be difficult 

 or impossible. The extrapolated yu's from these plots are from 6 to 9 

 per cent, too small. It is probable that the small value of the extra- 

 polated IX for A3 is due in part to an uncertainty regarding the pressure- 

 drops for some of the runs involved in this plot, due to a difficulty 

 with the zero correction to the manometer which has already been 

 mentioned (page 752). 



c. Axial flow plug A4. 



The results obtained with the set-up A4, and represented by the 

 lines so marked, are the best that have been secured with an axial 

 flow plug and are probably as good as can be obtained with apparatus 

 of the particular type used. The number of runs involved is excessive, 

 amounting in all to seventeen, for a single value of n which is about 

 1% too low. The very marked improvement, as regards heat-leak, 

 over the axial flow plugs Al, A2, and A3, is due chiefly to the larger 

 cross-channel, the kinetic energy depression of temperature being done 

 away with. It is probable also that' the improvement in thermal 

 insulation effected by the use of poplox lagging in the vertical tubes 

 T, T' of Fig. 1 is in part responsible for the reduction of heat-leak 

 errors. 



The AT vs. Ap plot probably shows to the best advantage the superi- 

 ority of this plug to the other axial flow plugs. The heat leak per 

 gram of steam is, generally speaking, less than half as great for a 

 given pressure-drop as it is for plugs A2 and A3, and increases much 

 less rapidly with increasing pressure drop than in these plugs. This 

 is partly due to the larger flow per unit A^ in plug A4. The slope of 

 the A T-Ap plot at its upper extremity approximates the true value of 

 H much more closely for plug A4 than for any of the other axial flow 

 plugs. 



All of the m' vs. plots for the axial flow plugs have large slopes, 



J 



