When you have a determined % of martensite (because you did not get to the Mf temperature in quenching), the diagram tells that lowering temperature will transform the remaining austenite into martensite, but would that transformation not require fast cooling?