Thermal Behaviour of Liquids. 217 



determined by one of us. Our observations have led us to the 

 following conclusions : — 



1. If temperature and volume be kept quite constant, striae 

 are never seen, but only a blue, misty appearance. 



2. It is absolutely essential that the whole of the sub- 

 stance should be equally heated, and this result cannot be 

 attained if the vapour of an impure liquid be employed as a 

 jacket. 



d. It is equally essential that the liquid examined should 

 be perfectly pure and free from air ; these precautions are 

 especially important in the determinations of volumes near 

 the critical point, since the smallest alteration in temperature 

 or pressure produces a great alteration in volume ; and to this 

 fact the erroneous conclusions referred to are in all probability 

 to be attributed. 



In certain cases, as for example with alcohols and acids, 

 there may be molecular complexes at the critical point ; but 

 we have no doubt whatever that the substance is physically 

 homogeneous, as much so indeed as a thorough mixture of 

 oxygen and trydrogen gases is ; and that the only difference in 

 density in different regions is that due to the weight of the 

 substance itself compressing the lower regions to a slightly 

 different density than that of the upper regions, just as the 

 density of the atmosphere at the level of the sea is greater 

 than that on a mountain (see Gouy, Compt. Rend. 1892, 

 p. 720). 



These considerations are sufficient, we think, to explain the 

 erroneous conclusions of the authors referred to. 



It is necessary next to draw attention to an incorrect ob- 

 servation of Battelli, regarding the vapour-pressures of liquids 

 and in particular of alcohol (Memorie della Reale Accademia 

 delle Scienze di Torino, serie ii. torn. xliv. p. 11). He there 

 states that he notices with alcohol what he has previously 

 noticed with other liquids, that the vapour-pressure at a given 

 temperature depends on the relative volumes of liquid and 

 gas, increasing as the volume of liquid increases and gas 

 diminishes, i. e. with decrease of volume. This observation 

 is absolutely opposed to our experience, and implies one of 

 two causes, — either that his sample of alcohol contained some 

 other liquid, or that it contained some permanent gas. His 

 data do not allow us to decide to which of these causes his 

 results are to be attributed. It is also to be noted that the 

 vapour-pressures which he records are in every case some- 

 what higher than those observed by Regnault and by ourselves. 



Such work as that referred to requires the utmost care, 

 especially in securing uniform known temperatures ; and it 



