Corollary 2.4. h:M→N is an L-embedding if and only if the conclusion of Theorem 2.2 holds for all quantifier free formulas φ(x1,…,xn).