CONDENSATION AND CRITICAL PHENOMENA. 207 



density is smaller or higher. The result is that the liquid 

 surface may disappear at any height in the tube, depending 

 on the total volume occupied by the substance, and the 

 critical phenomenon will not take place at one volume only, 

 though always at the same temperature. This fact added 

 to the influence of impurities seems to me to explain all ab- 

 normal phenomena observed, in so far as uncertain and 

 often contradictory phenomena of that kind can be ex- 

 plained. The theory of the association of molecules in the 

 liquid state seems superfluous and inadequate to explain the 

 abnormalities. 



From the above it would appear that pure observations 

 are almost impossible to make at the critical point. In 

 studying the more recent literature on the subject it is 

 found, however, that the more trouble is taken to obtain 

 pure substances the more normal the critical phenomena 

 come out. This is, for instance, so for an interesting experi- 

 ment devised by Cailletet and Colardeau. In an O tube 

 the under half of which is filled with mercury, two unequal 

 quantities of liquid carbonic acid are condensed in the two 

 limbs. The mercury naturally shows a difference in level ; 

 in heating the tube to and above the critical temperature 

 the difference does not disappear at once, and coolinjj;- the 

 tube again gives more liquid in the limb where there was 

 an excess of liquid before. This experiment has been 

 repeated probably with purer gas, and the difference dis- 

 appeared only quite a little above 31°, and equal quantities 

 condensed at the two sides in cooling. The presence of 

 some impurity in the liquid or gaseous CO2 explains the 

 former result readily. Even in cases where impurities 

 have not been entirely removed (and that would be in 99 

 out of 100) very trustworthy experiments may be obtained 

 by thoroughly stirring the substance. A little iron rod 

 moved up and down in the tube by a strong electro-magnet 

 outside serves the purpose admirably and is astonishingly 

 effective. The equilibrium between the two phases, which, 

 according to some very interesting time experiments of 

 Gouy, would take days to establish itself, only takes a few 

 moments when the stirrer is set in action ; a result at once 



