Simulated B – R colors of 183 Centaurs, i.e., a sample size three times our observed sample size, constrained by the GMM model in Figure 2. Application of the dip test to this simulated data results in a p-value of 0.008, i.e., a value low enough to reject the null hypothesis of unimodality. Simulations like this one suggest that a sample size on the order of two or three times our current sample size is necessary to see if a broad red color group is substantially reducing the ability of the dip test to detect bimodality in the current sample of 61 Centaurs.