EFFECT OF NUMBERS OF COLONIES PER PLATE ON THE ESTIMATE OF THE BACTERIAL POPULATION IN SOIL

Numbers of colonies of bacteria on plates from two series of 36 plots were found to vary widely on one date of sampling and from day to day. In an attempt to determine the effect of this variation and of introducing a dilution of different strength during the progress of an experiment on the accuracy of estimates, samples were plated in a series of replicate dilutions, with three parallel plates from each. In Experiment 1, four replicate dilutions of one strength were used. The experiment was repeated 120 times and shows that parallel dilutions, prepared by a slightly altered technique, add only very little to the error of the experiment beyond that of the interaction. In Experiment 2, three replicate dilutions of 1:300,000; 1:400,000, and 1:500,000 strengths, respectively, were used. This experiment was repeated 162 times, yielding a final result similar to that from dilutions of one strength. In the third experiment three replicate dilutions of 1:150,000; 1:250,000, and 1:400,000 strengths, respectively, were used. This experiment was repeated 360 times and shows that parallel dilutions, varying to this extent, introduce an appreciable error to the experiment.When counts from the 1:150,000 dilutions were correlated with those from the 1:400,000 dilutions of the same samples, it became obvious that the two systems of diluting do not provide estimates that are interchangeable. This proved true also with counts from five other pairs of dilutions, prepared in Experiments 2 and 3. In each case a curve was fitted to the data from the correlation surface, resulting in a sixth degree curve for the 1:150,000 and 1:400,000 dilutions, third degree curves for four other pairs of dilutions, and a linear relation for one pair.The results show that one dilution does not provide an estimate that agrees with that from another dilution of different strength, except at one count. A system of making estimates by using two dilutions and correcting the counts to provide interchangeable estimates is suggested.The data from Experiment 3 were corrected on this basis and analysed. The corrected data yield an F value for dilutions of different strength about the same as that obtained by the use of parallel dilutions of one strength.