6o6 SCIENCE PROGRESS 



be transformed into work while the whole system is what it 

 is. We then call upon the experimental philosopher to convert 

 it into something that at present it is not, but without initially 

 altering its entropy. This he can do by substituting for the 

 partition separating the gases a piston porous to one, but not, 

 or not in the same degree, porous to the other. We might 

 say paradoxically that, while the separated gases are separated, 

 the available energy is not available ; but when by using the 

 porous piston the separated gases are not separated, the avail- 

 able energy, or some part of it, is available. A dinner is 

 placed before Tantalus, and it is mathematically proved to 

 him that it is an available dinner, but he is unable to eat 

 or drink. 



17. There may be other transformations of the system which 

 would admit of work being obtained. One other method has 

 been suggested — namely, a liquid which absorbs one of the two 

 gases, but not the other. By such a liquid we might perhaps 

 reduce one of the two gases to lower pressure than the other, 

 and so render the performance of work possible. But that 

 method might be applied with equal success to two separated 

 portions of a single gas. 



18. Bryan uses instead of the porous piston a membrane 

 (p. 125). And he suggests the ideal case of two membranes, one 

 of which is porous to gas (i), but wholly impermeable to gas (2), 

 the other membrane vice versa. But should not this ideal 

 combination of gases and membranes be required to prove its 

 own existence before it is used to prove anything else ? If 

 Bryan says it does exist, I accept his authority. 



19. Another line of argument to prove the Rayleigh-Bryan 

 law, not expressly relied upon by Bryan, is as follows: If any 

 irreversible process takes place in the interior of a system, the 

 unavailable energy of the system, and therefore the entropy, is 

 increased. And the mixture by diffusion of two gases originally 

 separated is stated by Bryan to be an irreversible process. I 

 admit the argument, if the diffusion is an irreversible process. 

 But there is no proof that it is not a periodic process. 



20. It is, of course, a familiar experience that two separated 

 gases, each at the same pressure and temperature, if the partition 

 be removed, at once begin to mix — or, let us say, if they are 

 originally mixed in different proportions on the two sides of the 

 partition, at once begin to assume a more uniform mixture. But 



