Abstract: Let be a bijective map on such that both and are Borel measurable. For any and any real positive definite matrix let denote the -variate normal (gaussian) probability measure on with mean vector and covariance matrix Here we prove the following two results: (1) Suppose is gaussian for where is the identity matrix and is a basis for Then is an affine linear transformation; (2) Let where for every and is a basis of unit vectors in with denoting the transpose of the column vector Suppose and are gaussian. Then a.e. where runs over the set of diagonal matrices of order with diagonal entries are orthogonal matrices and is a collection of Borel subsets of such that and are partitions of modulo Lebesgue-null sets and for every is independent of all for which the Lebesgue measure of is positive. The converse of this result also holds. vskip0.1in Our results constitute a sharpening of the results of S. Nabeya and T. Kariya