Fitting a Mejzler distribution to extreme value data (Q1360971)

In his fundamental work on limit behaviour of normalized maxima from independent (not necessary identically distributed) random variables, D. Mejzler (1956) showed that the limit df \(G\) possesses a \(\log\)-concave property provided that the uniform negligibility condition is met. Denote the class of all such df's by \(\mathcal M\). The author demonstrates a method to fit a given df \(G\in\mathcal M\) to extreme value data, based on his former results (1975) on extremal processes generated by Poisson point processes whose mean measure at \(B=(x,\infty)\) is \(\lambda(x)=-\log G(x)\).