In a linear optimization problem there are generally variables being Zero for each optimal solution. A method for the determination of these variables is given. By means of these variables it is possible to calculate a separate basic representation for those variables which can take positive values in the set of optimal solutions. Hence the dimension of this set follows directly.