2 the measure to obtain the divergence structure that fits the divergences one expects from diffeomorphism symmetry. This of course does not fully guarantee diffeomorphism symmetry or triangulation independence, as divergences might also arise due to other reasons.In Regge calculus, several measures have been proposed: in [6] Hamber and Williams propose a discretization of the formal continuum path integral, with a local discretization of the (DeWitt) measure [6,7], conflicting with the proposal by Menotti and Peirano [8], who mod out a subgroup of the continuum diffeomorphisms resulting in a highly non-local measure. A different discretization, also leading to a non-local measure due to discretizing first the DeWitt super metric [20] and then forming the determinant, was proposed in [9].In this work we will pick up the suggestion in [21], to choose a measure that, at least for the linearized (Regge) theory, leads to as much discretization invariance as possible. These considerations require to actually integrate out degrees of freedom and thus take the dynamics into account.The requirement of discretisation independence seems to be at odds with interacting theories, which possess local / propagating degrees of freedom. This apparent contradiction can be resolved by allowing a non-local action or nonlocal amplitudes for the quantum theory -which in fact are unavoidable if one coarse grains the theory. Non-local amplitudes are however difficult to deal with. We therefore ask in this paper the question, whether in the quantum theory we can retain as much symmetry as in the classical theory, with a choice of local measure. The classical 4D Regge action is known to be invariant under 5 − 1 moves and 4 − 2 moves, but not under 3 − 3 moves [21]. Here the non-invariance under the 3 − 3 moves -in fact the only move involving bulk curvature for the solution -allows the local Regge action to nevertheless lead to a theory with propagating degrees of freedom. We therefore ask whether it is possible to have a local measure for linearized Regge calculus that leads to invariance under 5 − 1 and / or 4 − 2 moves. Such a measure would therefore reproduce the symmetry properties of the Regge action. We will however show that such a local path integral measure does not exist.Our requirement of (maximal) discretisation independence is motivated by the 'perfect action / discretisation' approach [22] that targets to construct a discretisation, which 'perfectly' encodes the continuum dynamics and has a discrete remnant of the continuum diffeomorphism symmetry. Examples of such 'perfect discretisations' are 3D Regge calculus with and without a cosmological constant [23] and also 4D Regge calculus, if the boundary data impose a flat solution in the bulk. In these examples, the basic building blocks mimic the continuum dynamics, e.g. one takes constantly curved tetrahedra for 3D gravity with a cosmological constant. Such perfect discretisations can be constructed as the fixed point of a coarse graining scheme, see for instance [23][24][25...