These data were used to estimate the probability of randomly selecting, from a general population, a glass specimen that is indistinguishable in all measured characteristics from a given glass specimen. For each of the 20,706 pairwise comparisons of the 204 specimens, the two specimens were analytically distinguishable. The use of highly discriminating analytical methods for comparing trace evidence and the corresponding low probability of two unrelated glass specimens being indistinguishable eliminates the need to collect extensive databases for the purpose of making exact probability calculations. The approaches in this paper were applied to glass evidence because robust data were available for that material. However, the same approaches can be used for other types of trace evidence. Figures, tables, references