We also measured Rsequence for sites recognized by E. coli RNA polymerase. Hawley and McClure (1983) compiled data on 112 well characterized E. coli promoters. For these promoters aligned by the -35 and -10 regions and using the range given by Hawley and McClure, Rsequence is only 11.1 bits. There are two difficulties with this analysis. First, a variable gap was introduced between the two regions, which will increase the uncertainty Hsand decrease Rsequence substantially, perhaps as much as 2 bits (unpublished observation). The other difficulty is that a reasonable estimate for the number of promoters in E. coli does not exist, so Rfrequency cannot be estimated. Nevertheless, promoters may be more frequent in E. coli (one per 500 bp) than is commonly assumed (see Discussion).