This paper considers (i) the robustness of the @t and Durbin-Watson bounds tests for first-order autocorrelation when disturbances in the linear regression model are heteroskedastic and (ii) the ...