The main advantages of DCDC-specimens are their completely stable crack extension properties and very high path stability due to the strongly negative T-stress term. Unfortunately, problems of DCDC tests can be identified by comparing experimental results that show for different materials (silicon nitride, glass) deviations from the results to be expected by 2-dimensional FE modelling as usual done in literature. Experimental calibrations on silicon nitrides and mixtures of silicon nitride and silicon carbide resulted in modified relations deviating from FE-results in literature. As a possible source for the differences of measurements and 2-D-FE results, we identified the influence of Poisson’s number. This parameter obviously causes deviations between straight-crack assumption in FEmodelling and observable curved crack fronts in the experiments. In order to avoid specimen buckling we also used short specimens of roughly half length. This may slightly affect the stress intensity factors.