In mixture experiments, optimal designs for the estimation of parameters, both linear and non-linear, have been discussed by several authors. In this paper, we attempt to find the optimum designs for testing the presence of synergistic effects in a mixture model using the maximin criterion.