In classical statistical mechanics we have to divide the partition function by a factor of $1/h^n$. In almost every calculation of a real quantity this cancels out and is thought to be a remnant of quantum mechanics.
However in the grand canonical ensemble we find in some rare equations the persistence of this term in the final result. For instance the Saha equation as derived from classical statistical mechanics has this factor remaining.
Perhaps related to this is the way we define a probability distribution on the phase space. How do we define (in terms of symplectic manifold theory) a phase space with variable number of particles (and hence dimensions)?
Does this lead to the weirdness that we are observing in some results of classical statistical mechanics retaining the $h$ factor? In essence does particle creation and disappearance violate formulation of mechanics on a manifold.