Abstract
Following a critique of the various arguments leading to Ginzburg's criterion for the width of the critical region the author shows that if the unperturbed propagator of fluctuations is expressed in terms of the full susceptibility (mass renormalization) then the width of the critical region follows systematically. The procedure gives, automatically, the width of the region about the real transition temperature for which both the Landau and the Ginzburg-Landau theories are valid. Exactly the same logic applies above and below the transition temperature.