On the Method of Maximum Likelihood