研討双分類不等重擅資料具有受感鼓應與否之變積分析理論，求出處理效應推算值之變方和變積以供處。理效應線型函數之顯著性測驗及複合比較 (Multiple Comparisons)之用，又在固定型 (Fixed Model) 之前提下求出各因子效應迴歸矯正均方之期望值，進而求出離心性特徵常數 (Non-Centrality Parameter) 以供探討擬說測驗功率 (Power of the Test) 之用。再者，討論不能正確判斷觀測值之線型模式中竟含有交感效應抑或不含交感之變積分析方法。繼之，將上述各點推展到具有P種共軛變值 (Concomitant Variates) 之複變積分析，並用於分析歷年蓬稻區域試驗結果資料。 The theory of analysis of covariance for two-way classification with unequal numbers in subclasses is studied under the linear models with and without interaction. The dispersion matrices of estimators for parameters in the model concerned are obtained for use in testing significance of a linear function of parameters and in making multiple comparisons among treatment effects. Under the assumption of fixed effects in the model, the expected values of the mean squares adjusted for regression are obtained so as to find the noncentrality parameters which are necessary for computing the powers of tests of linear hypotheses. A method of analysis of covariance is derived in case when the test for interaction should be used as a preliminary test as an aid in specifying the model for subsequent inferences. The above mentioned discussions are also extended to the multiple covariance .analysis for a two-way classification with unequal numbers in subclasses and used in analysing the perennial experiments with Japonica rice varieties at different locations.