On the Ree unital (Q1008978)

Let \({\mathcal R}_k\) denote the Ree unital of order \(3^k\) for any odd integer \(k\geq 1\), and let \(G_k\) denote the associated Ree group naturally acting on this unital. In [\textit{H. Lüneburg}, J. Algebra 3, 256--259 (1966; Zbl 0135.39401)], it was shown that one cannot embed \({\mathcal R}_k\) in a finite projective plane \(\pi\) of order \(3^{2k}\) in such a way that \(G_k\) is induced by a collineation group of \(\pi\) stabilizing the embedded unital. In the lovely paper under review, the author extends this classical result by showing that if \(k>1\), then \({\mathcal R}_k\) cannot be embedded in a finite projective plane \(\pi\) of order \(n\), for any \(n\leq 3^{4k}\), in such a way that \(G_k\) is induced by a collineation group \(G\) of \(\pi\) stabilizing the unital. The proof proceeds roughly as follows. First the author shows that in any such embedding the group \(G\) would act faithfully on \({\mathcal R}_k\) as \(G_k\). Then, using known properties of the Ree group \(G_k\), in a tour de force involving numerous counting arguments a contradiction is finally obtained. In the case \(k=1\), it is again shown that in any such embedding of \({\mathcal R}_1\) in a projective plane \(\pi\) of order \(n\leq 3^4\) the collineation group \(G\) will act faithfully on \({\mathcal R}_1\) as \(G_1\). Moreover, it is shown that in any such embedding either \(\pi\) is the desarguesian plane of order \(8\) or \(\pi\) is a plane of order \(64\). In fact, an example is given of \({\mathcal R}_1\) embedded in the desarguesian plane \(\pi=PG(2,8)\) in such a way that the collineation group \(G\) acting faithfully on \({\mathcal R}_1\) as \(G_1\) leaves invariant a line conic of \(\pi\), and \({\mathcal R}_1\) consists of the \(28\) external points to this line conic.