Studio del problema di Behrens-Fisher mediante pseudodensità

A classical problem, the study of the difference between the unknown means of two normal distributions (with unknown and different variances), whose solution is still hotly debated among statisticians, has been treated using the pseudodensity concept. As we know, this tool allows a rigorous approach to the so-called improper distributions. By assuming uniform initial pseudodensities we have found a final distribution which, as the size of the two samples becomes larger, approaches the Behrens-Fisher distribution, that is the same result which was determined by Jeffreys and by Fisher himself through different, although not quite beyond criticism, procedures.