The Analysis of Latin Squares when Some Observations are Missing

Abstract The discussion of the missing-value problem is given explicitly for a biological assay which employs a 4×4 latin square in several replications. However, the methods are easily adapted to any latin square and to various other designs as well. Methods of analysis, when some observations are missing, are discussed for the following cases. 1. One or more “single” observations are missing. 2. Several columns are missing. 3. One column is missing. 4. Two columns are missing. 5. One column and one or more single observations are missing.