RGB selection adopted. The red polygon in each panel shows a broadened Dartmouth isochrone model with [Fe/H] = −1.8, [α/Fe] = +0.6 and age 11.5 Gyr; we stress that this model simply provides a convenient visual fit to the RGB, and we are not using it to make any claims about the physical properties of the cluster. We chose the width of the RGB to be quadratic as a function of magnitude to encompass the obvious RGB seen in (a). That panel displays the targeted stars in the FLAMES survey that are located near the center of the cluster remnant (within ﹩10^{\prime} ﹩), where the contamination is least problematic. The filled circles show stars that have velocities within ﹩10\,\mathrm{km}\,{{\rm{s}}}^{-1}﹩ of the cluster, which we consider to be velocity members, while the open circles have velocities beyond that range. The chosen RGB polygon can be seen to encompass the majority of the member stars. The color-coding assigned to the points corresponds to a probability of being within the RGB region, assuming 0.01 magnitude errors in ﹩g﹩ and ﹩r﹩ (these are taken to approximate the photometric zero-point errors between fields, as well as allowing for some uncertainty in the RGB model). Panel (b) shows the same information for the AAOmega survey. Note that we included many blue horizontal branch candidate stars among the AAOmega targets. Panels (c) and (d) show the same information as the upper panels, but for the entire survey region. In calculating the RGB membership probability, we have taken into account the distance gradient as a function of position along the stream (which translates into small magnitude shifts).