Abstract: This paper discusses the estimate class of parameter β such as~ β = AX’y: the necessary and sufficient condition that ~β is the linear admissible estimate of β; when the error is normal, the linear admissible subclass is equivalent to the class of the Bayes estimate; the principal component estimate ~β_k^* of k has properties of T-min max and A, Φ₁-max min in the estimate class Γ_k.