Observations only partially constrain dynamical models of globular clusters and similar systems, in large part because distances and velocities are seen only in projection, and only at a single (unknown) orbital phase. As a result, some dynamical assumptions are necessary if the cluster dynamics and mass distribution are to be inferred from kinematic data — typically that mass follows light, or that the stars observed are the sole source of the gravitational field in which they move, or that orbital energies are completely thermalized, or that orbital angular momenta are exponentially distributed — most often some combination of these is invoked. Such assumptions, reasonable as they may seem, are almost never justified by the data per se for the cluster in question, and thus diminish the credibility of resulting estimates of cluster dynamics and, especially, of the presence and extent of dark matter.