Counting involutions (Q5954289)

This is a very interesting paper. It is based on an idea of \textit{H. Bender} [Ill. J. Math. 18, 223-228 (1974; Zbl 0279.20015)] who proved a combinatorial order formula which enables to calculate the order of a group with just one class of involutions. Before, this was just possible by using characters. Using the Bender formula the authors prove: Assume \(M\) is a nontrivial TI-subgroup of the finite group \(G\), let \(M^*=N_G(M)\) and assume \(M^*=MC_{M^*}(z)\) for some involution \(z\in M^*\) and \(C_G(x)\) is odd for each \(1\neq x\in M\). Then \(M\trianglelefteq G\) or the number of involutions of \(G\) is at most \(|G:M|\) or \(G\cong L_2(m)\) or \(\text{PG}L_2(m)\) for suitable \(m\). In a technical lemma they show that under certain assumptions on the commuting graph the assumptions of the theorem are satisfied. As an application they also prove: Let \(G\) be a finite group containing an involution \(z\) such that \(T=C_G(z)\) is dihedral of order 8. Then \(G=O(G)T\) or \(G\cong\text{PGL}_2(q)\), \(q=3,5\), or \(G\cong\text{PSL}_2(q)\), \(q=7,9\). A similar theorem can be found in \textit{M. Suzuki}'s book [Group theory II, Springer (1986; Zbl 0586.20001), pp.~132-135]. All the proofs in this paper are elementary, quotations are just to standard textbooks in group theory.