igo4\ Mills — Molecular Attraction. 97 



The same explanation will hold for normal octane at 0° C. 



For ether at 180° C reference to the original data shows 

 that the Biot formula adopted differs from the observations 

 sufficiently to account for the divergence, and it is but proper 

 here to add that Prof. Young writes: "Above 180° the sub- 

 stance (ether) was heated with methyl salicylate which was 

 not quite satisfactory. In later work quinoline was used." 



I think it may be safely said that every error occurring in 

 these eleven substances could be entirely eliminated by 

 changes in Biot's formula, in no case affecting any vapor 

 pressure so much as five-tenths of one per cent., and I there- 

 fore emphasize the fact that the only divergences from the 

 theory shown by these eleven substances are not errors of obser- 

 vation, primarily, but of calculation. 



It may be well to mention that of the 217 observations for 

 these eleven substances, 152 are within 1 per cent, of the 

 average value, and the 63 that are within 2 per cent, of that 

 value group themselves more largely at those points where 

 errors of calculation and observation would be the greatest. 



These eleven substances, therefore, from 0° C to their criti- 

 cal temperature, show, it seems to the author, as perfect 

 accord with the theory as the method and observations will 

 permit. 



Of the other substances, the following observations are not 

 within two per cent, of the mean values given. 



Di-isobutyl, 120° to 190° C, inclusive, shows a divergence 

 amounting at the greatest to five per cent, from the mean 

 value. This may be caused by an error in the vapor density, 

 because PAV does not show a maximum value at 2000 mms 

 pressure, and the same value at 5000 to 6000 mms as at 500 to 

 600 mms, as is the case with similar substances. 



Prof. Young has very kindly examined for me his original 

 notes and calculations upon di-isobutyl. The measurements 

 and calculations were abundantly checked, and there would 

 seem to be no chance for unusually large errors. He suggests 



