92 Dr. J. E. Mills on the Relation of 



+ or — to indicate the character o£ the divergence. The 

 average value of the unstarred results is shown at the bottom 

 of the table. 



The most casual examination of the table verifies at once 

 the substantial truth of the equation for most of the sub- 

 stances examined. A detailed discussion of the data and 

 results is given in the papers above cited. Here I would 

 only point out that in the first twenty-six substances shown 

 in the table, for which the very accurate measurements made 

 by Dr. Sydney Young* and his co-workers were available, 

 there is not a single divergence from the mean value of the 

 constant greater than 2 per cent, except at 0° 0. and near 

 the critical temperature and with di-isobutyl. Eight of the 

 substances show not a single divergence within the limits 

 stated greater than 1 per cent. Di-isobutyl was quite 

 possibly impure when measured. At 0° (30° for brom- 

 benzene and iodo-benzene) and near the critical temperature, 

 unusually large errors, for a discussion of which see the 

 papers cited, are unavoidable in obtaining the values of the 

 constants. The errors of measurement are always com- 

 pounded and often multiplied in their effect upon the con- 

 stant. The associated substances, water, the alcohols, and 

 acetic acid, should not give a constant and do not. For the 

 last seven substances shown in the table the data obtainable 

 are not always trustworthy. After careful consideration of 

 the facts upon which these statements are based (see the 

 papers cited) I have no hesitation in saying that the constant 

 is as near a constant as the errors of observation in the 

 measurements used will permit, and I regard the substantial 

 truth of the equation as established beyond question. 



It would be quite impossible within the limits of this paper 

 to discuss the data used in detail and the important con- 

 sequences that follow from the equation. Most of* the data 

 used were obtained from the measurements of Dr. Sydney 

 Young and his co-workers, and I am deeply indebted to 

 Dr. Young for extensive information regarding the data and 

 for obtaining in advance of their publication (see reference 

 cited) his extensive revision of much of the data used in my 

 earlier papers and by Kleeman. The series of papers cited 

 contain the details and the discussion and a complete re- 

 publication of all of that work is contemplated some time in 

 the future. Suffice it to say here that a comparison of the 

 results of equation 1 with the corresponding equation pro- 

 posed by Kleeman on the supposition of an inverse fifth power 



*. Sci. Proc. Royal Dublin Society, xii. p. 374 (1910). 



