A toxicity test measured average swimming speed only among fish alive after 72 hours. At the highest dose, 60% of fish died, and the survivors swam nearly as fast as controls. Why could this procedure understate toxicity?