A Historical Note on the Problem of Small Samples