Estimates of the intensity of selection based on measurements of the living and the dead require knowledge of the fraction of the original population dying. We apply recently developed methods (Blanckenhorn et al. 1999) to estimate the intensity of selection in a population of cliff swallows. In this population the fraction of individuals dying was unknown, but certainly high. The inferred selection is very strong and impossible to achieve if the original population is assumed to have followed a normal distribution. We consider several alternative explanations for this result including measurement biases, undetected immigration, and sampling biases. Of these, sampling biases are perhaps the most likely. We conclude that the intensity of selection on the swallows was probably strong, but its absolute magnitude is unknown.

Corresponding Editor: D. Roff