On the completeness of the calculus of logic