“…Second-and especially third-order Markov chains were better predictors of enzymes that produced the fewest fragments, but they generally underestimated the average fragment size at least ten-fold (Tables 3 and 4). Our results are in general agreement with similar analyses of other bacterial sequences (Phillips et al, 1987a(Phillips et al, , 1987bMcClelland et al, 1987), and reiterate the usefulness of Markov chain analysis in showing higher order sequence asymmetry (Elton, 1975;Algamor, 1983;Blaisdell, 1985, Rogerson, 1989) and in predicting site frequencies (McClelland et al, 1987;Phillips et al, 1987aPhillips et al, , 1987b in genomic DNAs.…”