52 Prof. J. Dewar. The Boiling Point of Liquid Hydrogen, 

 In the case of an error in P, a similar process gives 



rfA " (ap - spp 273 + <-iTi r (5) ; 



If x = 1/50, * = 13°, P = 760 mm., Ti - - 180°; dT = 0-3563 dP, 

 so that an error of 1 mm. in P would only alter the boiling point of 

 oxygen by a third of a degree. In the same circumstances at - 250°, 

 dT = 0*3516 dP, which is practically the same result at the boiling point 

 of hydrogen as at that of oxygen. 



For the second thermometer, these two equations become 



at - 180°, dT = 0-3575 dP, 

 at - 250°, dT = 0*3548 dP. 



In each of the last four results if P = - x 760 mm. the formulas 



n 



become respectively 



dT = % x 0-3563 dP, and dT = n x 0-3516 dP, 

 dT = n x 0-3575 dP, and <2T = n x 0*3548 dP ; 



in other words, any error in reading P is magnified in its effect on T 

 directly in proportion as P is diminished. This affords some expla- 

 nation of the weakness of the results in Experiment (No. 3). 

 In like manner, from an error in P , we get 



dT --p S^ <•>■ 



Here if x = 1/50, t = 13°, P = 760 mm., T x - - 180° ; 



dT = - 0-1188 dP 0y 



or an error of 1 mm. in P would only alter the boiling point of oxygen by 

 a ninth of a degree ; but with the same data at - 250°, dT = - 0*0264 dP 0> 

 so that the boiling point of hydrogen would only be altered by a tenth 

 of a degree for a change of 4 mm. on an initial pressure of about one 

 atmosphere. 



In this case also if P = - x 760 mm. we get similar results to those 



n 



in the case of P, namely, 



For x = 1/50, dT = - n x 0-1188 dP and dT = - n x 0*0264 dP ty 



For x = 1/115, dT = - n x 0*1192 dP and dT = -n x 0*0266 dP . 



The general result of an error in either P or P is, that the more 

 reliable experiments are those in which the initial pressure is as high 



