Feldman et al. have given an algorithm for inferring a regular grammer from a finite sample of strings. As the algorithm infers different grammers and of different sizes for a sample with different order of presentation of its strings, Feldman and his collaborators suggested the processing of sample's strings in the order of decreasing lengths. This is the order for which their algorithm infers equivalent grammers of small number of rewriting rules. Since the strings of the sample cannot be presented in a sufficiently large number of ways in this order, this may appear a serious problem to some users besides their interest of inferring a grammer of small number of rewriting rules for their sample.