On the Frattinidual of Doerk and Hauck (Q1078663)

Let G be a finite group and \(\Psi\) (G) the smallest normal subgroup of G such that \(N\cap \Psi (G)\) generates the same Fitting class as N for each normal subgroup N of G. G/\(\Psi\) (G) is always soluble, as was proved by \textit{K. Doerk} and \textit{P. Hauck} [Arch. Math. 35, 218-227 (1980; Zbl 0465.20017)] who left as an open question whether or not G/\(\Psi\) (G) must always be of nilpotent length not exceeding two. In the present paper the authors answer this question in the negative: they construct for each integer \(n\geq 1\) a group \(G_ n\) of nilpotent length \(n+1\) such that \(G_ n/\Psi (G_ n)\) has nilpotent length \(n+1\). They seem to have reason to believe that more than they have proved is almost certainly true when conjecturing: for any finite soluble group S, there is a finite soluble group G such that S is isomorphic to a subgroup of G/\(\Psi\) (G).