The experimental study of joint mechanics has been limited in its effectiveness due to the high uncertainty associated with assemblies of sub-components. In particular, two categories of uncertainty are variability (the uncertainty in measurements of different, nominally identical parts) and repeatability (the uncertainty in measurements of the same set of parts). As a result, the uncertainty measured is often greater than the nonlinear characteristics being studied (such as amplitude dependent frequency and damping), which makes meaningful experimentation challenging. This paper analyzes the contributors to uncertainty in the form of variability and repeatability in order to make recommendations for methods to reduce the uncertainty and to redesign a joint to improve its dynamics. Experiments are summarized that investigate the role of experimental setup, interface roughness, settling versus wear, interface geometry (both meso-scale and macro-scale), and the structure surrounding the joint. From the results of these studies, recommendations for the measurement of nonlinearities in jointed structures are made.