Heteroscedasticity is potentially a serious problem for research workers in statistics. The Literature on Inferential Problems for Linear Statistical, models under heteroscedasticity has grown enormously in the past three decades. The presence of heteroscedastic errors disturbs the optimal properties of OLS estimators of parameters of linear statistical model. In this case, the OLS estimators are no longer efficient estimators. There is a need of detecting the existence of the heteroscedastic errors in the linear model.