516 Sir W. Ramsay ami Dr. B. D. Steele on the 



Now the greatest divergence between results with different 

 samples occurs in the case o£ toluene; and it is only 1 part 

 in 1900 ; it is clear, therefore, that the discrepancy cannot 

 be attributed to impurity of material, nor to inaccuracy in 

 experiment. To what cause, then, is it to be ascribed ? We 

 have given on p. 494 several hypotheses which may explain 

 the divergence between found and "theoretical"'^ results; 

 these shall now be considered. 



First, it is evident that Avogadro's law does not hold for 

 vapours, even under the ideal condition of zero pressure. 

 There would appear to be only three possible causes which 

 would explain the anomaly. 



We regard it as unlikely that the volume of the vapour is 

 decreased by the adherence of a film to the glass walls of the 

 volume-tube. It is to be expected that this phenomenon would 

 become more observable at high, than at low, pressures ; and 

 one of the authors, in conjunction with Prof. Young, has 

 investigated the behaviour of ether and methyl-alcohol at 

 very high pressures, and Prof. Young has himself examined 

 the hydrocarbons without finding any ground for entertaining 

 this supposition. On the other hand, evidence that it does 

 take place in the case of water has been obtained by them. 

 It is manifested by an apparently continuous, instead of an 

 abrupt, change on passing from the gaseous into the liquid 

 condition. It has an eff'ect in modifying what is often termed 

 the " Andrews " diagram in the same manner as if a per- 

 manent gas were present along with the substance under 

 experiment. To quote from the paper mentioned (Phil. Trans. 

 1892, A, p. 113): — "As the vapour-pressure is approached, 

 the curves, instead of cutting the vapour-pressure line so as 

 to form an angle, as is the case with the other liquids which 

 we have examined, gradually turn and run nearly parallel to 

 the vapour-pressure line at a somewhat lower pressure. When 

 a considerable amount of liquid has condensed the true vapour- 

 pressure is reached.'^ Now this phenomenon increases with 

 rise of pressure ; but in the experiments of which an account 

 has been given in this paper, the curvature is such as to show 

 that the rate of increase of j9v/T is less at low than at high 

 pressures. Not merely, then, is the phenomenon absent in 

 the class of liquids under experiment, but even if it should 

 be conceived to be present its variation with pressure is in 

 the wrong direction. We think, therefore, that it may be 

 dismissed from further consideration. 



That the formation of complex molecular groups is the 

 cause of the high density of these vapours is certainly a con- 

 ceivable supposition, but hardly a tenable one. For methyl- 

 alcohol, which is known to consist of complex groupings in 



