Large Energy Cumulants in the 2D Potts Model and their Effects in Finite Size Analysis

We develop an ansatz for expressing the free energy of the two dimensional $q$-states Potts model for $q > 4$ near its first order phase transition point. We notice that for the moderate values of $ q \lesssim 15 $, the energy profile at the phase transition is not expressible as a sum of gaussians. We discuss how this affects the traditional finite size analysis of this phase transition. In particular, the dominant length scale governing the finite size corrections turns out to be much (${} \sim 6$ ~times) larger than the largest correlation length in the problem.