This note reexamines the general class of Sargan densities analyzed by
Goldfeld and Quandt (1981), Kafei and Schmidt (1985), and others in t
he context of approximating normal densities in certain econometric mo
dels. It suggests a third-order Sargan density which complies with the
Goldfeld and Quandt criteria, is unimodal, and is the best approximat
ion to the normal for orders less than or equal to 3. Some errors in G
oldfeld and Quandt paper are corrected.