In 1955, 26 racers were caught in the May-June-July period, in the headquarters field area. In the August-September-October period of the same year five racers were caught of which only one was a member of the original 26. The five-to-one ratio indicates that the original 26 may have represented an actual population of 130, but of course the single recapture is much too small a sample to provide a reliable ratio. Some of the racers caught in May were recaptured in June, others in July, and still others not until late summer or early autumn. Somewhat different estimates can be obtained for the population depending on how the season's records are divided. For instance, in the headquarters field area in May 1955, eleven racers were caught; in the remainder of the season 20 were caught, of which two were members of the original group of eleven. The 20 to 2 ratio indicates that the 11 caught in May represented an actual population of 110. In the period May-June, 18 racers were caught, and in July-August-September-October, 15 were caught, including four of the original 18, hence indicating a population of 67. Although obviously the population underwent some change during the course of the season, the three sets of census figures apply essentially to the same population, and the divergence in them illustrates the wide range of error arising from insufficiently small samples. Common sources of error in the censusing of natural populations of animals by the capture-recapture method ("Lincoln Index" or "Petersen Index") arise from the fact that the composition of a local population often changes between two sampling periods, or even within them. Some of the animals marked may move elsewhere, to be replaced by unmarked immigrants, or they may die and be replaced by unmarked maturing young. First-year racers that could pass through the quarter-inch mesh of the traps in spring and early summer became too large to escape in this way in the latter half of the summer, but these young were excluded from the census computations. There was doubtless some shifting of marked individuals away from the study areas and shifting of new individuals onto these areas in the periods of weeks between successive samplings. Jackson (1939) has explained a method of correcting census computations based on capture-recapture ratios when there is a consistent trend of diminishing recaptures with increase in elapsed time. However, in my records no such trend is discernible; furthermore it has been demonstrated that individual racers tend to stay within the same home range area throughout most of their season of activity. Therefore, I conclude that shifts of individuals away from the study areas or into them, in the intervals between samplings constitute only a minor source of error.