The quantitative limitations of exponential curve fitting

Abstract

A detailed statistical study of the principal factors which affect exponential fitting of biological data has been carried out for two- and three-exponential functions. The results indicate the importance of complete definition of the function by prolonged observation, and draws attention to the extremely large errors liable to be obtained from even `good' biological data. The limitations in the experimental design, due to inadequate definition of the data, lead to practical criteria for the collection of data and to the design of future studies of this type.