David Joyner reports in his Adventures in Group Theory, 2nd ed. of $2008$ that this problem was the work of Christoph Bandelow, but he gives no further details.