150 Professor Dewar [June 13, 



the respective temperatures of -115° C. and -125° C. In order to 

 get liquefaction conveniently with either of these agents, it is neces- 

 sary to work at a pressure of oxygen gas from 80 to 100 atmospheres, 

 and to have the means of producing a sudden expansion when the 

 compressed gas is cooled to the above-mentioned temperatures. This 

 is brought about by the use of an additional stopcock, represented in 

 the fif^ure at B. During the expansion the stopcock at A is closed 

 and the pressure-manometer carefully observed. No doubt liquid 

 nitrous oxide is the most convenient substance to use as a cooling 

 agent ; but as it is apt to get superheated during the reduction of 

 pressure and boil over with explosive bursts of vapour, it is well to 

 collect the fluid in a small flask of about 250 cub. centim. capacity, 

 and to change it into the solid state by connecting the flask with the 

 air-pump, and then to use the substance in this form. The addition 

 of alcohol or ether to the solid nitrous oxide makes the body more 

 transparent, and thereby favours the observations. 



It is evident that this apparatus enables the observer to determine 

 the density of the fluid gases condensed in tbe tube F ; since he has 

 only to measure the volume of fluid in F, and to collect, by means of 

 the stopcock B, the whole volume of gas given by the fluid and 

 condensed vapour, which gives an accurate determination of the total 

 weight of substance distributed between fluid and vapour in the whole 

 apparatus. The amount of substance which is required to produce 

 the vapour is easily found by observing the vapour-pressure of the 

 liquid gas before expanding it into gas for the volume measurement ; 

 and while keeping shut the stopcock B, by opening A suddenly until 

 tliis pressure is just reached and then instantly shutting off the 

 receiver. If this volume of gas is now measured by opening B as 

 before, the difference between the two volumes thus collected will 

 correspond to the real w^eight of substance in the liquid state. A 

 rough experiment with oxygen near the critical point gave the 

 density 0*65. 



As to the most convenient substance for use as a cooling agent, I 

 am still of opinion that marsh-gas would be the best ; and I may 

 take the opportunity of pointing out that the employment of this 

 body was suggested by me in a communication made to the Chemical 

 Section of the British Association in 1883. The following extract 

 from 'Nature,' of October 4, 1883, will prove that my experiments 

 with liquid marsh-gas were made a year in advance of those made 

 recently by M. Cailletet * and M. Wroblewski f : — 



" Professor Dewar pointed out an important relation between the 

 critical temperatures and pressures of volatile liquids and their 

 molecular volumes. The ratio of the critical temperature to the 



* " Sur I'emploi du Formeiic pour la production de3 trijs basses temperatures," 

 C'imptes Rendus, June 30, 1884. 



t " Sur les proprietes du gaz des niarais liquide, et sur son emploi commo 

 rdfrigeraut," Conqdes liendus, July 21, 1884. 



