Considerable scattering of experimental wear results affects seriously the evaluation of repeatability and reproducibility of tribological measurements and hampers detecting, studying, and verifying tribological laws. An accurate characterization of the statistics of results, the detection of the influence of operating parameters, and the verification of equations describing tribological phenomena can be achieved only through the analysis of large datasets with wide variation of parameters.Taking advantage of more than 400 experiments performed with the same material pairing (100Cr6) under the same conditions (unlubricated, ball-on-disc configuration, reciprocating sliding) on four different tribometers, the repeatability and reproducibility of volumetric wear measurements has been evaluated. The effect of the tribometer used for the acquisition, of five operating parameters (frequency, number of cycles, stroke, velocity, and normal force), and of the friction coefficient could be characterised using Welch's unequal variances t-test.The proportionality between volumetric wear and the product of sliding distance and normal force (sFN) is described in Archard's law through the proportionality factor k, commonly known as wear coefficient. By testing Archard's law over seven orders of magnitude of sFN, it was found that this law is not universally valid and that, under certain conditions, the wear coefficient depends on sFN itself. An alternative equation for the dependence of the volumetric wear on sFN is presented. The limits of the validity of Archard's law were investigated by varying the normal force and the sphere radius, using among others experiments with a reused, worn ball.