When is entropy measurable? (was: Re: Relativity question)

[pvalev <pvalev@BAS.BG>]

--- "John S. Denker" <jsd@MONMOUTH.COM> wrote:

> Similarly logically, entropy should be measured
> in bits, but that is often inconvenient, so people
> use Joules per Kelvin instead.

A few questions please.

1. Does entropy need to be a state function in order to be
measurable? If yes, do we need a PROOF that it is a state function?
If yes, please sketch the proof (for the classical entropy defined by
dS=dQrev/T).

2. In case entropy is measurable, is it true that it always increases
(in an isolated or adiabatically isolated system)? Let me remind you
that the famous principle "Entropy always increases" was initially
deduced from two premises raised by Clausius:

(A) Closed integral of dQ/T =< 0

(B) Any irreversible process can be closed into a cycle by a
reversible process.

In his late writings, Clausius abandoned both (B) and the
principle "Entropy always increases". Too late - the principle
had already become famous. The curious thing is that (A) is also
incorrect - there are counterexamples. Now nobody except perhaps a
few people know that "Entropy always increases" is a CONCLUSION and
should be DEDUCED from falsifiable premises rather than just stated.
And nobody cares of course. Please say whether you believe in the
principle "Entropy always increases" and why.

Pentcho