EXPERIMENTAL STUDY OF HEAT LEAKAGE. 771 



Mr. S. A. Moss has pointed out to the writer that if one may assume 

 that the total heat-leak per second, 8Q, is proportional to the drop of 

 temperature, AT, — certainly an assumption which appears more 

 plausible than that 8Q is proportional to the drop of pressure — then a 



plot of , vs. ~ should be straight, and the extrapolation of this plot 

 M / 



should lead to the correct value of . This appears on dividing 



M 

 equation (1) by fx-ST; this gives 



(3) i_i ae 1 



which represents a straight line in the variables -, and ,, provided 



M J 



8Q/i\T is constant. 



A few plots of — vs. - are shown in Fig. 14. 



A moment's consideration of the geometrical relations of the two 

 plots will show that if either is straight, with a slope not zero, the other 

 cannot be straight; that if either is straight and approximately hori- 

 zontal, the other will be nearly straight, and will also be approximately 

 horizontal ; and that if ' best representative straight lines ' are drawn 

 through corresponding experimental sets of points on the two plots, 



these lines will have intercepts on the axis of which will be, approxi- 

 mately, the reciprocals of each other, if the two lines are nearly straight 

 and nearly horizontal. The consequence of this is that where the 

 heat-leak effect varies so little within the range of experiment that 

 the extrapolation seems worthy of confidence, it makes little differ- 

 ence which plot is used. At the same time, the results of extrapolating 

 both lines, even in such a case as this, may be seriously in error; that 

 is, mere absence of large slopes in both plots does not signify necessarily 



that their intercepts are true values of /x or of ; see, for example, the 



curves Ul in Figs. 12, 13 and 14. 



Still another method of exhibiting graphically the results of Joule- 

 Thomson experiments is illustrated in Figs. 15 and 16, in which the 

 observed temperature drop is plotted as ordinate against the pressure 

 drop as abscissa. In the absence of heat-leak, all plotted points 

 should lie on a straight line passing through the origin, and that por- 



