THE ACCURACY OF THE PLATING METHOD FOR ESTIMATING THE NUMBERS OF BACTERIA AND FUNGI FROM ONE DILUTION AND FROM ONE ALIQUOT OF A LABORATORY SAMPLE OF SOIL

Six 25-gm. aliquots taken from a well-mixed laboratory sample were plated, each in six replicate dilutions with four replicate plates from each dilution, for counts of fungi. The analysis of variance shows that there are not significant differences in counts of fungi among replicate dilutions from one aliquot, but that there are among aliquots from one laboratory sample. Each replicate dilution was raised and the final dilutions plated in four replicates for counts of bacteria. The data for bacterial counts show significant differences among dilutions from one aliquot, but not among aliquot samples.In a second experiment, one 25-gm. aliquot taken from a sample was diluted 1:10 and another was diluted 1:50. Each original dilution was raised to 1:5,000 in 11 replicate dilutions, which were plated in four replicates for fungi. The experiment was repeated 10 times. In this case, the data show that the 1:10 method of making the original dilution yields significant differences among the final dilutions and that the 1:50 system, which reached 1:5,000 in one transfer, is preferable. Each dilution was raised to 1:500,000 and the final dilutions were plated for bacteria in six replicates. The analysis shows that the 1:10 method is not reliable because of significant differences among dilutions and that the 1:50 method is preferable, although failing to reduce the differences to insignificance.The 1:50 and 1:100 systems of making the original dilution were compared in Experiment 3, as well as differences among aliquot samples. A fresh sample was plated in five aliquots for each system, each aliquot in ten replicate dilutions and each dilution in four replicate plates for bacteria. The 1:50 system again shows significant differences among dilutions and the 1:100 system is not preferable. Likewise, there are significant differences among aliquot samples in each case.In Experiment 4 all dilutions were raised from 1:50 original dilutions. Each trial consisted of six aliquots, raised in six replicate series of dilutions and plated in six replicate plates from each final dilution. This was repeated four times for fungal counts and six times for counts of bacteria. The analysis again shows that for fungal counts differences among dilutions are not significant, while for bacterial counts they are. Again, there are significant differences for aliquot samples in the case of both fungal and bacterial counts.In Experiments 2, 3, and 4, the plating, pouring, piling of plates in the incubator and counting of plates were carried out in one order. The analysis shows that none of these practices adds anything significant to the error of plating.As the errors of the sample used and of the dilution plated are significant, reliable information on the counts of bacteria, actinomyces, or fungi in a laboratory sample is not obtained by the usual procedures with one 25-gm. sample and one final dilution from it, regardless of the number of replicate plates made from the dilution. By the use of six aliquot samples with three replicate dilutions from each, and one plate for each dilution, the estimate would be based upon these three factors in about their proportionate weight.Only by carefully designed experiments and the application of statistical methods to check the validity of the results obtained can progress be made in developing the plate method of counting bacteria or fungi in soil to a stage where it may be used for practical application to the problems of agriculture.