An experiment was designed to study the combined effect of changes in moisture, dates of sampling, and numbers of colonies per plate on the estimated number of bacteria in the soil of two plots. The data represent 30 replicate estimates on a wheat plot and 30 on a fallow plot, each made from a different dilution on each of 19 dates of sampling from May 21 to September 27, together with moisture determinations for each plot on each date. The statistical treatment consists of a combination of correlation and analysis of variance studies. The results are presented graphically, and in the usual form of analysis of variance. They demonstrate the presence of a basic population in each plot which responds differently to moisture and time in different dilutions.