Thank you very much for the the answer Mr. Ulrich. I can't understand the Wilk's test. How does a proportion two determinats measure a relationship? I am clear with ANOVA case, but I am very confused with the multivariate case. These determinants are called generalized variance, but I can't understand what is generalized variance. I think understanding of generalized variance and understandind those eigevalues will make someone fully understand of multivariate procedures. Thanks a lot.