EXPERIMENTAL STUDY OF HEAT LEAKAGE. 775 



plots (Fig. 14) also represent least-square adjustments. The dotted 



curves on this diagram represent the corresponding least-square n' 



vs. - lines : where no dotted curve is drawn, the two loci are so close 



together as to be barely distinguishable. The numbers adjoined to 

 each plotted point represent the number of runs involved, and, as a 

 rule, also the weight attached to the point in making the least square 

 adjustment. 



The extrapolated value of n (3°. 182 C. cm.Vkgm.) as given by the 

 set-up 8 is believed to be the true value of ix within |%, for reasons 

 which will appear in the following discussion. This ^'alue of // is indi- 

 cated on Figs. 15 and 16 by the slope of the straight dotted line passing 

 through the origin. Thi'oughout the following discussion, to avoid 

 circumlocution, the value of /x with which these lines are associated 

 will be spoken of as the true ^•alue of /x. 



The actual mean temperatures of the various experiments range from 

 162°.6 C. to 168°.9. In reducing the observed /x's to the standard 

 temperature of 165° C, the value —0.031 cm.Vkgm., taken from 

 Fig. 7 of Davis' paper, ' On the Applicability of the Law of Correspond- 

 ing States to the Joule-Thomson Effect in Water and Carbon Dioxide,' ^ 

 was used for djx/dT. This value is undoubtedly sufficiently accurate 

 for the purpose. A few measurements of /x for steam at 220° C. made 

 by the writer verify the trend of Davis' curve. Moreover, the devia- 

 tions of the temperatures at which the individual fx' 's belonging to 

 a given curve were experimentally determined, from the mean of 

 these temperatures for this set of tx' 's, are what are chiefly significant 

 as regards error due to an erroneous value of dji/clT. This is because 

 any error in the reduced values of these ix 's will be magnified in the 

 extrapolation. In the case of every curve, these deviations were much 

 smaller than the 2.4 to 3.9 degrees C. by which the temperatures of 

 the extreme ^' 's depart from 165° C. 



In general, it is the high-side pressure rather than the mean pres- 

 sure which was held approximately constant in the experimental work, 

 so that, strictly, a reduction for pressure should also have been made 

 before plotting the observations. No such reduction has been made. 

 Its effect would certainly be almost negligible and would probably 

 be to increase slightly the slopes of all the plotted straight lines of 

 Figs. 12, 13 and 14, and thus also the /x's obtained by extrapolation. 



8 Davis, Proc. Am. Acad., 45, 243-264 (1910). 



