Total proper‐motion difference between M4 and background/foreground objects as a function of magnitude for the outer field data. In this diagram the motion is zero‐pointed on M4, and the displacement is given in HST pixels over the 6 yr baseline of the observations. To separate the cluster from the field, all objects possessing motion within 0.5 pixel of that of M4 were assumed to be cluster members. Because we do not resolve the M4 internal motion, it is clear from the diagram that our uncertainties in the measured proper motions do not exceed 2 mas yr⁻¹ down to F606W = 27.5 and at F606W = 29.5 they are smaller than 7 mas yr⁻¹.