The method commonly used in counting bacteria in soil is a modification of the plating method of Koch. In counting bacteria two difficulties have to be overcome--their immense numbers and their small size. The numbers of bacteria in soil are so large that the bacterial population of a gram of soil could not, of course, be counted directly. The method adopted, therefore, is to make a suspension of soil in sterile salt solution, and to dilute this suspension to a convenient and known extent, which will depend on the numbers of bacteria expected. In ordinary field soils it is found convenient, for example, to dilute the soil suspension so that one cubic centimeter of the diluted suspension will contain 1/250,000th of a gram of soil. Such a volume will commonly contain a number of bacteria sufficiently small to count. The second difficulty is that the organisms are microscopic, and yet cannot be readily counted under the microscope owing to the presence of soil particles in the suspension. Hence recourse is had to plating. One cubic centimeter of diluted suspension is placed in a petri dish and mixed with a suitable nutrient agar medium, melted, and cooled to about 40° C. The medium sets, and after a few days’ incubation the organisms multiply and produce colonies visible to the naked eye. By counting these colonies we obtain an estimate of the number of bacteria in the one cubic centimeter of suspension, it being assumed that every organism has developed into one colony, and by multiplying this number by the degree of dilution we obtain the numbers per gram of soil. In practice a number of parallel platings are made from one cubic centimeter portions of the diluted suspension and the mean number of colonies per plate is taken. By this means the error due to the random distribution of bacteria in the suspension is reduced, because of the greater number of organisms counted. In drawing conclusions from bacterial count data, it is necessary to distinguish between the indication which the method gives of the absolute numbers of bacteria in the soil and the accuracy with which it enables the numbers of two soil samples to be compared. The method cannot be used for the former purpose at present. We do not know how far the figures obtained by this counting method fall short of the actual number of bacteria in the soil. One reason for this is the difficulty of effecting a complete separation of the clumps of bacteria into discrete individuals in the suspension. Then again, there is no known medium upon which all the physiological groups of bacteria will develop and produce colonies. And even on a suitable medium some of the individuals may fail to multiply.