484 Dr. J. E. Mills on the 



attraction. In these equations L— E e is the internal heat of 

 vaporization, and d and D are the densities of liquid and 

 vapour respectively. 



It is fortunately possible to test the correctness of 

 Kh 'email's statement by reference to the facts. Prof. Sydney 

 Young in 1903 wrote me that he considered benzene, fluor- 

 benzene, normal pentane, normal hexane, and isopentane as 

 probably yielding the best experimental results among the 

 substances measured. For testing Kleeman's equation I 

 therefore chose the three of the above substances of different 

 type, and in addition ethyl oxide and carbon tetrachloride as 

 representing other types of compounds. The data used are 

 completely given by Young *. I give the results of the test 

 of both my own and Kleeman's equation below (pp. 486-487). 

 The average of the constant was in each case taken to the line 

 drawn, and all divergences greater than 2 per cent, from this 

 average are marked by an asterisk above and to the right. 

 My equation gives for these five substances two divergences 

 from the average value of the constant greater than 2 per cent. 

 Kleeman's equation gives with isopentane 6 such diver- 

 gences, with normal pentane o, with ethyl ether 9, with 

 benzene 15, with carbon tetrachloride 12. But this is not 

 all. My equation gives for the entire 26 non-associated 

 substances measured by Young and his co-workers not a 

 single divergence f greater than 2 per cent, from the mean 

 value of the constant given, except as introduced by known 

 errors occurring at 0° C. (30° for bromo-benzene) near the 

 critical temperature, and for di-isobutyl, all as noted below. 

 From the average values of the constants there given my 

 equation shows fur these 2b' substances within the limits 

 stated divergences greater than 2 per cent, only for chloro- 

 benzene at 130° (2*15 per cent.), carbon tetrachloride at 

 250° and 260° (see above), and ethyl propionate at 100° 

 (2*1 per cent.). The last line on p. 91 of the article just 

 cited should read " more than 1 per cent, from the average 

 values are marked above and to the left with a." Kleeman's 

 equation, as shown more fully below, in no case investigated 

 gives results which can possibly be regarded as within 

 the limit of the experimental error of the measurements 

 involved. 



Near the commencement of my work upon molecular 

 attraction some eight years ago I discussed J the way in 

 which the various errors of measurement would affect the 

 constant. For a discussion of the errors of measurement 



* Sci. Proc. Rov. Dublin Soc. xii., June 1910. 



t Phil. Mag. July 1911, p. 84. 



\ Journ. Phvs. Cliem. viii. pp. 392 397 (1904). 



