Extra-Period Latin-Square Change-Over Designs

Abstract

Latin-square change-over trials balanced for carry-over effects have been modified by adding an extra period of observation. During this extra period the treatment pattern is identical with that in the last period of the Latin-square design. This modification reduces somewhat the information per observation about treatments if residual effects do not exist, but increases the information on residual effects and especially on “permanent” effects of treatment. Use of the extra-period designs is discussed and patterns for two to five treatments are given. The statistical analysis is outlined symbolically and is illustrated numerically with uniformity data. © 1957, American Dairy Science Association. All rights reserved.