The paper introduces a measure of information provided by an experiment, derived from Shannon's information theory. This measure is based on prior probability distributions over the parameter space and is used to compare experiments without reference to prior distributions. It contrasts with methods discussed by Blackwell and is applied to experimental design where the goal is to gain knowledge, not make decisions. The measure is defined using the Kullback-Leibler divergence and is shown to be related to the average information gained from an experiment. The paper also discusses the properties of this measure, including its concavity and convexity, and its application to comparing experiments. It shows that experiments can be compared absolutely, and that some experiments are more informative than others. The paper also discusses the relationship between different methods of comparing experiments, including those based on loss functions and sufficiency. The results are illustrated with examples, including the binomial dichotomy and normal distribution cases, and the paper concludes that the purpose of some statistical experimentation is to gain and measure information about the state of nature.The paper introduces a measure of information provided by an experiment, derived from Shannon's information theory. This measure is based on prior probability distributions over the parameter space and is used to compare experiments without reference to prior distributions. It contrasts with methods discussed by Blackwell and is applied to experimental design where the goal is to gain knowledge, not make decisions. The measure is defined using the Kullback-Leibler divergence and is shown to be related to the average information gained from an experiment. The paper also discusses the properties of this measure, including its concavity and convexity, and its application to comparing experiments. It shows that experiments can be compared absolutely, and that some experiments are more informative than others. The paper also discusses the relationship between different methods of comparing experiments, including those based on loss functions and sufficiency. The results are illustrated with examples, including the binomial dichotomy and normal distribution cases, and the paper concludes that the purpose of some statistical experimentation is to gain and measure information about the state of nature.