A reconsideration of stress experiments and population heterogeneity