Information and economics in Fisher’s design of experiments

In 1951 R. A. Fisher described what had been achieved in the 20th century so far: "we have learnt (i) To conserve in its statistical reduction the scientific information latent in any body of observations. (ii) To conduct experimental and observational inquiries so as to maximise the information obtained for a given expenditure." This paper asks what Fisher meant and, in particular, how he saw his work on experimental design as contributing to the objective of maximizing information for a given expenditure. The material examined ranges from detailed work on issues like "the information lost in measurement of error" to polemics against decision theory.