1) Guy lewis steele, jr 
Generalized Least Squares. A generalization of the OLS procedure to make an efficient linear regression estimate of a parameter from a sample in which the disturbances are heteroskedastic. That is, in y = Xb + e (equation 1) that the e's vary in magnitude with the X's. The estimator of b is: (X'O-1X)-1X...
Gesellschaft f?r Logistischen Service