For an arbitrary rank general Gauss Markoff model Y=Xβ+μ,μ∽(0,Σ) , where Σ is a nonnegative ddefinite matrix, the effect of transforming the observable vector Y to FY is analyzed with respect to the variance of the Best Linear Unbiased Estimator (BLUE) of C′β . It is shown that there exists an estimable subspace S in which the BLUE of C′β can be get as a function of FY . The result obtained in [1] is generalized in this paper. |