Finite 2-groups with no normal elementary Abelian subgroups of order 8 (Q5957543)

P. Roquette (1958) classified the \(p\)-groups without normal elementary Abelian subgroups of order \(p^2\). Next, N. Blackburn (1961) classified the \(p\)-groups, \(p>2\), (i) without normal elementary Abelian subgroups of order \(p^3\) and (ii) without normal subgroups of order \(p^p\) and exponent \(p\). In this important paper the author obtains a reasonably complete description of the \(2\)-groups without normal elementary Abelian subgroups of order \(8\). Main Theorem 1.1. Let \(G\) be a finite \(2\)-group which has no normal elementary Abelian subgroup of order \(8\). Suppose that \(G\) is neither Abelian nor a group of maximal class. Then \(G\) possesses a normal metacyclic subgroup \(N\) such that \(C_G(O_2(N))\leq N\) and one of the following holds: (a) \(|G:N|\leq 4\) and either \(\Omega_2(N)\) is Abelian of type \((4,4)\) or \(N\) is Abelian of type \((2^j,2)\), \(j\geq 2\). (b) \(G/N\cong D_8\) and (b1) \(N\) is Abelian either of type \((2^k,2^{k+1})\), \(k\geq 1\), or of type \((2^t,2^t)\), \(t\geq 2\), or (b2) \(N\) is minimal non-Abelian, \(\Omega_2(N)\) is Abelian of type \((4,4)\), and more precisely, \(N=\langle a,b\mid a^{2^m}=b^{2^n}=1,\;a^b=a^{1+2^{m-1}}\rangle\), where \(m=n\geq 3\) or \(m=n+1\) with \(n\geq 2\). As an easy corollary, one obtains that all subgroups of the groups of the title are generated by four elements (MacWilliams, 1970). It is remarkable that the proof is entirely elementary and unexpectedly short. In the proof, some results presented in the reviewer's unpublished book `Groups of prime power order, Parts 1-3', are used. Moreover, in that book, in Sect. 50, a very detailed proof of Theorem 1.1 is presented. The main theorem of this paper has surprisingly many deep consequences (some of them are presented in three subsequent papers). Note that Blackburn's result (i) is not a correct analog of Janko's theorem in the case \(p>2\). The correct analog must be such: Description of the \(p\)-groups, \(p>2\), without normal subgroups of order \(p^{p+1}\) and exponent \(p\). I think that there is little hope to prove this analog in the nearest future. As the author remarks, there were three unsuccessful trials to prove Theorem 1.1.