A case involving glass fragments where the recovered glass 'just fails
' to 'match' the control is discussed. A Bayesian approach to interpre
ting this evidence is offered, and the authors advocate a move to cont
inuous models of this type for all cases involving glass evidence.