The wedge test, as used for the evaluation of adhesive fracture energy, is usually considered to be a 2D geometry: its simple analysis implies independence of the width of the adhesive joint, b. Recent work has shown this to be an over-simplification, at least in some circumstances, with (hypothesised) anticlastic bending giving rise to curvature of the crack front. As a result, crack front length, a, varies across the joint width leading to ambiguity in the interpretation of results to obtain fracture energy, Gc. This contribution constitutes a more detailed analysis of the geometry of the wedge test (in the particular case of one bending and one rigid substrate), treating the bent member as a plate, rather than as a simple beam. The Kirchhoff-Love plate theory is applied and solved by a perturbation method. Secondary curvature of the beam in the direction normal to the principal curvature results directly from the treatment, and this, in turn, leads to a concave crack front, corroborating the above-mentioned experimental observation.