Or just design the apparatus as others have suggested (dewars
etc.), so that heat leaks are negigible in comparison to what
you're trying to measure.
================
This whole thread seems to revolve (implicitly or perhaps
explicitly) around the hypothesis that energy might not be
conserved.
-- If you end up with a null result, nobody is going to be
surprised. Nobody is going to care.
-- If you end up with a non-null result, you are going to
need a reeeeeally well-designed experiment if you expect
to convince anybody. As James Randi likes to say, extraordinary
claims require extraordinary levels of proof.