Comment on ``Duality relations for Potts correlation functions''

In a recent paper by Wu (Phys. Lett. A 228, 43-47 (1997)) the three-point correlation of the q-state Potts model on a planar graph was related to ratios of dual partition functions under fixed boundary conditions. It was claimed that the method employed could straightforwardly be applied to higher correlations as well; this is however not true. By explicitly considering the four-point correlation we demonstrate how the appearence of non-well-nested connectivities invalidates the method.