OX THE CONSERVATION OF FORCE. 305 



vessel was substituted for water in a brass one, gave 425 and 

 426-3 metres. 



3. Two series of experiments, in which a conical ring rubbed 

 against another, both surrounded by mercury, gave 426'7 and 

 425-6 metres. 



Exactly the same relations between heat and work were also 

 found in the reverse process that is, when work was produced 

 by heat. In order to execute this process under physical con- 

 ditions that could be controlled as perfectly as possible, per- 

 manent gases and not vapours were used, although the latter 

 are, in practice, more convenient for producing large quantities 

 of work, as in the case of the steam-engine. A gas which is 

 allowed to expand with moderate velocity becomes cooled. 

 Joule was the first to show the reason of this cooling. For the 

 gas has, in expanding, to overcome the resistance which the 

 pressure of the atmosphere and the slowly yielding side of the 

 vessel oppose to it : or, if it cannot of itself overcome this 

 resistance, it supports the arm of the observer which does it. 

 Gas thus performs work, and this work is produced at the cost 

 of its heat. Hence the cooling. If, on the contrary, the gas is 

 suddenly allowed to issue into a perfectly exhausted space where 

 it finds no resistance, it does not become cool, as Joule has 

 shown ; or if individual parts of it become cool, others become 

 warm; and, after the temperature has become equalised, this 

 is exactly as much as before the sudden expansion of the 

 gaseous mass. 



How much heat the various gases disengage when they are 

 compressed, and how much work is necessary for their compres- 

 sion; or, conversely, how much heat disappears when they ex- 

 pand under a pressure equal to their own counterpressure, and 

 how much work they thereby effect in overcoming this counter- 

 pressure, was partly known from the older physical experiments, 

 and has partly been determined by the recent experiments of 

 Regnault by extremely perfect methods. Calculations with the 

 best data of this kind give us the value of the thermal equiva- 

 lent from experiments : 



