( 783 ) 
AB is moved parallel to itself, and it is easy to see that this will 
not be the case if simply ¢,—/(z,a), but it will if «, = /(a—a). 
This may be assumed here, so the reduction is permissible. 
When there are more molecules the reasoning continues to be 
valid if €, = f(a—a,,a—a, etc.), at least for so far as only the 
piston is considered. 
We meet with a second difficulty, if we take also «, and ¢, into 
account; now we may ee 
Syl Eg 
= de nn OF 
he i = dh = —Ce de ES da,..dz, = 
da 
== £q=E3 
p é es 
i S(e,te,te Oe Lax ape 
fe Beren den Ee) a id ON hi a e : Oat, isan 
A da. da 
Now we may take account of «, and ¢, by regulating the limits 
in accordance with them in the way discussed, and further by 
a 
putting «, and «, =O within these limits, while athe may be put 
for ¢,. It seems, however, to me, that this method comes to this 
that for e, and ¢, not /(#,...2%,) is taken, as should have been done, 
