% Journal of the Mitchell Society. [Nov. 



often be multiplied and are always compounded in their effect 

 upon this constant. To cover this range of temperature, the 

 number of substances, and the variously compounded errors 

 of observation in vapor pressure, vapor density, and density 

 of the liquid, errors which may be multiplied in the calcula- 

 tion, it has seemed to me reasonable to pass without discus- 

 sion all variations in the constant less than two per cent, 

 from the mean values chosen and given at the top of each 

 column. An agreement within two per cent, of this mean 

 value we therefore consider satisfactory. All values not 

 showing this agreement are in black faced type. 



Excepting values within ten degrees of the critical temper- 

 ature the eleven substances, ethyl, oxide, di-isopropyl, iso- 

 pentane, normal pentane, normal hexane, normal heptane, 

 normal octane, benzene, hexamethylene, fluo-benzene, and 

 carbon tetrachloride, show out of two hundred and seventeen 

 observations only two that are not within two per cent, of the 

 mean value adopted for that particular substance. The two 

 exceptions are normal octane at 0° C and ethyl oxide at 180° 

 C, and both of these divergences, as well as those occurring 

 within ten degrees of the critical temperature, are, as shown 

 below, easily explained. 



In six of the eleven substances above mentioned, the obser- 

 vations allowed the tests of the formula to be carried to 

 within one degree of the critical temperature, and in the case 

 of normal pentane to within 0.05 of that point, yet in no case 

 is the divergence greater than ten per cent, from the mean 

 value adopted for that substance. Nearing the critical tem- 

 perature an inspection will show that an error in the vapor 

 pressure will be multiplied proportionately some seventy times 



8 P 

 in the constant if the error likewise affects the -^- and, as 



already explained (p. 93), at the end points of the Biot for- 

 mula curve these errors do affect, to a considerable degree, 



the W 



