Nearest neighbour graphs for both methods of determining the optimum ϵ for DBSCAN. To produce this plot, which is effectively an unnormalised log cumulative density function (CDF) of nearest neighbour distances, stars are sorted based on their kthNN distances and numbered from one to n. These labels as a function of kthNN distance are then plotted to form a continuous curve. The black line on both plots is the 9thNN distances of a 2.5° field around the nearby OC Blanco 1. On the upper plot, ϵ estimates are determined by re-sampling the field 30 times (shown in red) to smooth out the signature of clustered stars. On the lower plot, a model of the signature of the cluster (cyan, dashed) and the field (magenta, dashed) is summed (red, solid) to approximate the curve and produce four ϵ estimates.