Left: LOSVD of the faint component (grey histogram with errorbars) and best-fit model pdf (black line), consisting of the halo (dashed blue line) and IGL (dotted red line), and scaled to the observed LOSVD. While the likelihood is fit with a bin-free method, we bin the LOSVD for visual convenience. For comparison, the best-fit single-Gaussian model is shown in orange. The dashed green line denotes the best-fit Gauss-Hermite expansion. Right: fraction of faint PNe associated with the IGL as a function of major-axis radius, based on the probabilities derived from the LOSVD decomposition. The number of PNe associated with the IGL is given on top of each bin.