Bacterial populations from four adjacent sites with different soils were compared. The sites sampled were acid forest humus, grassland soil, and adjacent duplicates of open field soil. About a hundred random isolates from each of the four soil samples, 467 in all, were submitted to 44 two-state tests. The correlations between the tests were computed in terms of the test results obtained with all the isolates. Forty-two of the tests were accepted for further consideration. Their correlation matrix was factorized, and the first nine principal factors, covering 99% of the total estimated communality, were taken into account after varimax rotation.With the aid of within/between population variance, the ecologically discriminating efficiency of the factors was determined. For qualitative description of the populations six significantly discriminating factors (P < 0.1%) were used. These factors covered 56% of the total factor variance considered. The populations were characterized by the position in the 6-factor space of the point centroids (population means or "centers of gravity") of all the item points making up a population.The six factors of ecologically discriminating efficiency were dominated by attributes on (1) oxidase, and nitrate reductase, (2) Bacillus characteristics, (3) protocatechuic acid utilization, (4) pH tolerance, (5) nutritional demands, and (6) pigment-correlated hydrolytic abilities.The similarity or difference between the populations was expressed as the squared distance (d2) between the population means in the selected discriminating 6-factor space. This parameter varied from 10 for the pair of duplicate field soil populations, to 172 between one of the field soil populations and the forest humus population.