264 PROCEEDINGS OF THE AMERICAN ACADEMY. 



in general much more self-consistent. Further, they now agreed very 

 definitely with the tests of Table IV in verifying the law of correspond- 

 ing states and lay close along the tentative curve previously drawn. 

 These facts, particularly the disappearance of the tendency to swoop 

 near 0.9, seem to show that this reasoning is not a "circular fallacy," 

 and that the values in Table V are a real corroboration of those in 

 Table IV. 



As a precaution against using these corrections too freely in cases 

 where they might, perhaps, not apply, it seemed best to include in 

 Table V only such of the 33 selected tests of the type in question as 

 resembled the tests from which the corrections were determined in 

 having comparatively large steam flow (more than 80 lbs. per hour). 

 Furthermore, all tests or parts of tests were rejected for which the 

 observed temperature drop was not as great as five times the correc- 

 tion, as the application of any correction amounting to more than 20 

 per cent of the quantity involved seemed unsafe. The 33 tests were 

 thus reduced to 19, and these, corrected as above, gave the 77 values 

 of the coefficient in Table V. 



Note on the Vertical Scale of Figure 1. 



The numerical values of the ordinates in Figure 1 are not the 

 "reduced " Joule-Thomson effect in the ordinary sense, because Buck- 

 ingham, in computing them, used 100 in. of mercury as his unit of 

 pressure, but nevertheless expressed his critical pressures in atmos- 

 pheres. The true reduced values of // are those indicated in the 

 upper corner of Figure 6. 



Jeffekson Physical Laboratory, 

 Cambridge, Mass., 

 December, 1909. 



