Yield-density relations and Jarvis's lucerne data

Abstract

1. Methods of fitting certain yield-density relations are given which are appropriate when the proportional rather than the absolute standard error of the yields is constant.2. These methods are applied to relations discussed by Jarvis in connexion with his data on lucerne, and it is shown that no clear advantage can be claimed for any of the relations.3. The role of significance tests in the interpretation of the data is briefly discussed.