Burnside obstructions to the Montesinos-Nakanishi 3-move conjecture (Q1865108)

It is well known that a crossing change for a knot diagram is an unknotting operation, that is, a finite sequence of crossing changes transforms the diagram into that of the unknot. It is an important operation since it is used to calculate and define polynomial invariants such as the Alexander, Jones, and HOMFLY polynomials and also to introduce the Vassiliev (or finite type) invariants. There are other unknotting operations and among them a Borromeo transformation introduced by Matveev [\textit{S. V. Matveev}, Mat. Zametki 42 , No. 2, 268-278 (1987; Zbl 0634.57006)] is a key to the clasper theory by Goussarov and Habiro [\textit{S. Garoufalidis, M. Goussarov, M. Polyak}, Geom. Topol. 5, 75-108 (2001)(electronic); \textit{K. Habiro}, Geom. Topol. 4, 1-83 (2000; Zbl 0941.57015)]. On the other hand it was asked by Y. Nakanishi and is known as the Montesinos-Nakanishi conjecture that whether a 3-move (parallelizing three successive half-twists) is an unknotting operation. Note that a 3-move can change the connectivity and so, strictly speaking, the question is: Can any link be transformed into a trivial link by a finite sequence of 3-moves? In this paper the authors prove that this does not hold by showing that two links, proposed by Nakanishi and Chen as counterexamples, in fact cannot be transformed to trivial links by 3-moves. The proof uses the third Burnside group of the fundamental group of the double branched covering space of a link. Here the third Burnside group of a group \(G\) is its quotient by all relations of the form \(g^{3}=1\).