Experiments in environmental toxicology can sometimes be manipulative experiments in which the researcher actively chooses and manipulates the independent variable. In Hunt's study, for example, dosages of BPA were manipulated and the effects were measured. In manipulative studies, the researcher controls all the other variables in the experiment, so any health effects observed in the test subjects can be attributed to differences in the independent variable. In other cases, researchers use natural experiments in which the dependent variable (typically a measure of organism health) is measured under differing contexts that are not manipulated. Say, for example, that an accidental chemical spill contaminates five ponds. To determine the possible effects of the toxic chemical on frogs, a researcher could compare the hatching rate of frog eggs laid in those five ponds to the hatching rate of eggs laid in five uncontaminated ponds nearby. This would be an example of a natural experiment because concentrations of the toxic chemical in the ponds were not controlled by the experimenter, but rather resulted from the chemical spill. Drag type of experiment on the left to the example of experiment on the right.